includes/upload/UploadBase.php

   1 <?php
   2 /**
   3  * Base class for the backend of file upload.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Upload
  22  */
  23
  24 /**
  25  * @defgroup Upload Upload related
  26  */
  27
  28 /**
  29  * @ingroup Upload
  30  *
  31  * UploadBase and subclasses are the backend of MediaWiki's file uploads.
  32  * The frontends are formed by ApiUpload and SpecialUpload.
  33  *
  34  * @author Brion Vibber
  35  * @author Bryan Tong Minh
  36  * @author Michael Dale
  37  */
  38 abstract class UploadBase {
  39         protected $mTempPath;
  40         protected $mDesiredDestName, $mDestName, $mRemoveTempFile, $mSourceType;
  41         protected $mTitle = false, $mTitleError = 0;
  42         protected $mFilteredName, $mFinalExtension;
  43         protected $mLocalFile, $mFileSize, $mFileProps;
  44         protected $mBlackListedExtensions;
  45         protected $mJavaDetected, $mSVGNSError;
  46
  47         protected static $safeXmlEncodings = array( '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 int $error
  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 WebRequest $request
 133          * @param string|null $type
 134          * @return null|UploadBase
 135          */
 136         public static function createFromRequest( &$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 WebRequest $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          * @param WebRequest $request
 216          */
 217         abstract public function initializeFromRequest( &$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 int
 238          */
 239         public function getFileSize() {
 240                 return $this->mFileSize;
 241         }
 242
 243         /**
 244          * Get the base 36 SHA1 of the file
 245          * @return string
 246          */
 247         public function getTempFileSha1Base36() {
 248                 return FSFile::getSha1Base36FromPath( $this->mTempPath );
 249         }
 250
 251         /**
 252          * @param string $srcPath The source path
 253          * @return string|bool The real path if it was a virtual URL Returns false on failure
 254          */
 255         function getRealPath( $srcPath ) {
 256                 wfProfileIn( __METHOD__ );
 257                 $repo = RepoGroup::singleton()->getLocalRepo();
 258                 if ( $repo->isVirtualUrl( $srcPath ) ) {
 259                         // @todo just make uploads work with storage paths
 260                         // UploadFromStash loads files via virtual URLs
 261                         $tmpFile = $repo->getLocalCopy( $srcPath );
 262                         if ( $tmpFile ) {
 263                                 $tmpFile->bind( $this ); // keep alive with $this
 264                         }
 265                         $path = $tmpFile ? $tmpFile->getPath() : false;
 266                 } else {
 267                         $path = $srcPath;
 268                 }
 269                 wfProfileOut( __METHOD__ );
 270                 return $path;
 271         }
 272
 273         /**
 274          * Verify whether the upload is sane.
 275          * @return mixed self::OK or else an array with error information
 276          */
 277         public function verifyUpload() {
 278                 wfProfileIn( __METHOD__ );
 279
 280                 /**
 281                  * If there was no filename or a zero size given, give up quick.
 282                  */
 283                 if ( $this->isEmptyFile() ) {
 284                         wfProfileOut( __METHOD__ );
 285                         return array( 'status' => self::EMPTY_FILE );
 286                 }
 287
 288                 /**
 289                  * Honor $wgMaxUploadSize
 290                  */
 291                 $maxSize = self::getMaxUploadSize( $this->getSourceType() );
 292                 if ( $this->mFileSize > $maxSize ) {
 293                         wfProfileOut( __METHOD__ );
 294                         return array(
 295                                 'status' => self::FILE_TOO_LARGE,
 296                                 'max' => $maxSize,
 297                         );
 298                 }
 299
 300                 /**
 301                  * Look at the contents of the file; if we can recognize the
 302                  * type but it's corrupt or data of the wrong type, we should
 303                  * probably not accept it.
 304                  */
 305                 $verification = $this->verifyFile();
 306                 if ( $verification !== true ) {
 307                         wfProfileOut( __METHOD__ );
 308                         return array(
 309                                 'status' => self::VERIFICATION_ERROR,
 310                                 'details' => $verification
 311                         );
 312                 }
 313
 314                 /**
 315                  * Make sure this file can be created
 316                  */
 317                 $result = $this->validateName();
 318                 if ( $result !== true ) {
 319                         wfProfileOut( __METHOD__ );
 320                         return $result;
 321                 }
 322
 323                 $error = '';
 324                 if ( !wfRunHooks( 'UploadVerification',
 325                         array( $this->mDestName, $this->mTempPath, &$error ) )
 326                 ) {
 327                         wfProfileOut( __METHOD__ );
 328                         return array( 'status' => self::HOOK_ABORTED, 'error' => $error );
 329                 }
 330
 331                 wfProfileOut( __METHOD__ );
 332                 return array( 'status' => self::OK );
 333         }
 334
 335         /**
 336          * Verify that the name is valid and, if necessary, that we can overwrite
 337          *
 338          * @return mixed True if valid, otherwise and array with 'status'
 339          * and other keys
 340          */
 341         public function validateName() {
 342                 $nt = $this->getTitle();
 343                 if ( is_null( $nt ) ) {
 344                         $result = array( 'status' => $this->mTitleError );
 345                         if ( $this->mTitleError == self::ILLEGAL_FILENAME ) {
 346                                 $result['filtered'] = $this->mFilteredName;
 347                         }
 348                         if ( $this->mTitleError == self::FILETYPE_BADTYPE ) {
 349                                 $result['finalExt'] = $this->mFinalExtension;
 350                                 if ( count( $this->mBlackListedExtensions ) ) {
 351                                         $result['blacklistedExt'] = $this->mBlackListedExtensions;
 352                                 }
 353                         }
 354                         return $result;
 355                 }
 356                 $this->mDestName = $this->getLocalFile()->getName();
 357
 358                 return true;
 359         }
 360
 361         /**
 362          * Verify the mime type.
 363          *
 364          * @note Only checks that it is not an evil mime. The does it have
 365          *  correct extension given its mime type check is in verifyFile.
 366          * @param string $mime Representing the mime
 367          * @return mixed True if the file is verified, an array otherwise
 368          */
 369         protected function verifyMimeType( $mime ) {
 370                 global $wgVerifyMimeType;
 371                 wfProfileIn( __METHOD__ );
 372                 if ( $wgVerifyMimeType ) {
 373                         wfDebug( "\n\nmime: <$mime> extension: <{$this->mFinalExtension}>\n\n" );
 374                         global $wgMimeTypeBlacklist;
 375                         if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
 376                                 wfProfileOut( __METHOD__ );
 377                                 return array( 'filetype-badmime', $mime );
 378                         }
 379
 380                         # Check IE type
 381                         $fp = fopen( $this->mTempPath, 'rb' );
 382                         $chunk = fread( $fp, 256 );
 383                         fclose( $fp );
 384
 385                         $magic = MimeMagic::singleton();
 386                         $extMime = $magic->guessTypesForExtension( $this->mFinalExtension );
 387                         $ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime );
 388                         foreach ( $ieTypes as $ieType ) {
 389                                 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
 390                                         wfProfileOut( __METHOD__ );
 391                                         return array( 'filetype-bad-ie-mime', $ieType );
 392                                 }
 393                         }
 394                 }
 395
 396                 wfProfileOut( __METHOD__ );
 397                 return true;
 398         }
 399
 400         /**
 401          * Verifies that it's ok to include the uploaded file
 402          *
 403          * @return mixed True of the file is verified, array otherwise.
 404          */
 405         protected function verifyFile() {
 406                 global $wgVerifyMimeType;
 407                 wfProfileIn( __METHOD__ );
 408
 409                 $status = $this->verifyPartialFile();
 410                 if ( $status !== true ) {
 411                         wfProfileOut( __METHOD__ );
 412                         return $status;
 413                 }
 414
 415                 $this->mFileProps = FSFile::getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 416                 $mime = $this->mFileProps['file-mime'];
 417
 418                 if ( $wgVerifyMimeType ) {
 419                         # XXX: Missing extension will be caught by validateName() via getTitle()
 420                         if ( $this->mFinalExtension != '' && !$this->verifyExtension( $mime, $this->mFinalExtension ) ) {
 421                                 wfProfileOut( __METHOD__ );
 422                                 return array( 'filetype-mime-mismatch', $this->mFinalExtension, $mime );
 423                         }
 424                 }
 425
 426                 $handler = MediaHandler::getHandler( $mime );
 427                 if ( $handler ) {
 428                         $handlerStatus = $handler->verifyUpload( $this->mTempPath );
 429                         if ( !$handlerStatus->isOK() ) {
 430                                 $errors = $handlerStatus->getErrorsArray();
 431                                 wfProfileOut( __METHOD__ );
 432                                 return reset( $errors );
 433                         }
 434                 }
 435
 436                 wfRunHooks( 'UploadVerifyFile', array( $this, $mime, &$status ) );
 437                 if ( $status !== true ) {
 438                         wfProfileOut( __METHOD__ );
 439                         return $status;
 440                 }
 441
 442                 wfDebug( __METHOD__ . ": all clear; passing.\n" );
 443                 wfProfileOut( __METHOD__ );
 444                 return true;
 445         }
 446
 447         /**
 448          * A verification routine suitable for partial files
 449          *
 450          * Runs the blacklist checks, but not any checks that may
 451          * assume the entire file is present.
 452          *
 453          * @return mixed True for valid or array with error message key.
 454          */
 455         protected function verifyPartialFile() {
 456                 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
 457                 wfProfileIn( __METHOD__ );
 458
 459                 # getTitle() sets some internal parameters like $this->mFinalExtension
 460                 $this->getTitle();
 461
 462                 $this->mFileProps = FSFile::getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 463
 464                 # check mime type, if desired
 465                 $mime = $this->mFileProps['file-mime'];
 466                 $status = $this->verifyMimeType( $mime );
 467                 if ( $status !== true ) {
 468                         wfProfileOut( __METHOD__ );
 469                         return $status;
 470                 }
 471
 472                 # check for htmlish code and javascript
 473                 if ( !$wgDisableUploadScriptChecks ) {
 474                         if ( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
 475                                 wfProfileOut( __METHOD__ );
 476                                 return array( 'uploadscripted' );
 477                         }
 478                         if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 479                                 $svgStatus = $this->detectScriptInSvg( $this->mTempPath );
 480                                 if ( $svgStatus !== false ) {
 481                                         wfProfileOut( __METHOD__ );
 482                                         return $svgStatus;
 483                                 }
 484                         }
 485                 }
 486
 487                 # Check for Java applets, which if uploaded can bypass cross-site
 488                 # restrictions.
 489                 if ( !$wgAllowJavaUploads ) {
 490                         $this->mJavaDetected = false;
 491                         $zipStatus = ZipDirectoryReader::read( $this->mTempPath,
 492                                 array( $this, 'zipEntryCallback' ) );
 493                         if ( !$zipStatus->isOK() ) {
 494                                 $errors = $zipStatus->getErrorsArray();
 495                                 $error = reset( $errors );
 496                                 if ( $error[0] !== 'zip-wrong-format' ) {
 497                                         wfProfileOut( __METHOD__ );
 498                                         return $error;
 499                                 }
 500                         }
 501                         if ( $this->mJavaDetected ) {
 502                                 wfProfileOut( __METHOD__ );
 503                                 return array( 'uploadjava' );
 504                         }
 505                 }
 506
 507                 # Scan the uploaded file for viruses
 508                 $virus = $this->detectVirus( $this->mTempPath );
 509                 if ( $virus ) {
 510                         wfProfileOut( __METHOD__ );
 511                         return array( 'uploadvirus', $virus );
 512                 }
 513
 514                 wfProfileOut( __METHOD__ );
 515                 return true;
 516         }
 517
 518         /**
 519          * Callback for ZipDirectoryReader to detect Java class files.
 520          *
 521          * @param array $entry
 522          */
 523         function zipEntryCallback( $entry ) {
 524                 $names = array( $entry['name'] );
 525
 526                 // If there is a null character, cut off the name at it, because JDK's
 527                 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
 528                 // were constructed which had ".class\0" followed by a string chosen to
 529                 // make the hash collide with the truncated name, that file could be
 530                 // returned in response to a request for the .class file.
 531                 $nullPos = strpos( $entry['name'], "\000" );
 532                 if ( $nullPos !== false ) {
 533                         $names[] = substr( $entry['name'], 0, $nullPos );
 534                 }
 535
 536                 // If there is a trailing slash in the file name, we have to strip it,
 537                 // because that's what ZIP_GetEntry() does.
 538                 if ( preg_grep( '!\.class/?$!', $names ) ) {
 539                         $this->mJavaDetected = true;
 540                 }
 541         }
 542
 543         /**
 544          * Alias for verifyTitlePermissions. The function was originally 'verifyPermissions'
 545          * but that suggests it's checking the user, when it's really checking the title + user combination.
 546          * @param User $user User object to verify the permissions against
 547          * @return mixed An array as returned by getUserPermissionsErrors or true
 548          *   in case the user has proper permissions.
 549          */
 550         public function verifyPermissions( $user ) {
 551                 return $this->verifyTitlePermissions( $user );
 552         }
 553
 554         /**
 555          * Check whether the user can edit, upload and create the image. This
 556          * checks only against the current title; if it returns errors, it may
 557          * very well be that another title will not give errors. Therefore
 558          * isAllowed() should be called as well for generic is-user-blocked or
 559          * can-user-upload checking.
 560          *
 561          * @param User $user object to verify the permissions against
 562          * @return mixed An array as returned by getUserPermissionsErrors or true
 563          *   in case the user has proper permissions.
 564          */
 565         public function verifyTitlePermissions( $user ) {
 566                 /**
 567                  * If the image is protected, non-sysop users won't be able
 568                  * to modify it by uploading a new revision.
 569                  */
 570                 $nt = $this->getTitle();
 571                 if ( is_null( $nt ) ) {
 572                         return true;
 573                 }
 574                 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
 575                 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
 576                 if ( !$nt->exists() ) {
 577                         $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
 578                 } else {
 579                         $permErrorsCreate = array();
 580                 }
 581                 if ( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
 582                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
 583                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
 584                         return $permErrors;
 585                 }
 586
 587                 $overwriteError = $this->checkOverwrite( $user );
 588                 if ( $overwriteError !== true ) {
 589                         return array( $overwriteError );
 590                 }
 591
 592                 return true;
 593         }
 594
 595         /**
 596          * Check for non fatal problems with the file.
 597          *
 598          * This should not assume that mTempPath is set.
 599          *
 600          * @return array Array of warnings
 601          */
 602         public function checkWarnings() {
 603                 global $wgLang;
 604                 wfProfileIn( __METHOD__ );
 605
 606                 $warnings = array();
 607
 608                 $localFile = $this->getLocalFile();
 609                 $filename = $localFile->getName();
 610
 611                 /**
 612                  * Check whether the resulting filename is different from the desired one,
 613                  * but ignore things like ucfirst() and spaces/underscore things
 614                  */
 615                 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName );
 616                 $comparableName = Title::capitalize( $comparableName, NS_FILE );
 617
 618                 if ( $this->mDesiredDestName != $filename && $comparableName != $filename ) {
 619                         $warnings['badfilename'] = $filename;
 620                         // Debugging for bug 62241
 621                         wfDebugLog( 'upload', "Filename: '$filename', mDesiredDestName: '$this->mDesiredDestName', comparableName: '$comparableName'" );
 622                 }
 623
 624                 // Check whether the file extension is on the unwanted list
 625                 global $wgCheckFileExtensions, $wgFileExtensions;
 626                 if ( $wgCheckFileExtensions ) {
 627                         $extensions = array_unique( $wgFileExtensions );
 628                         if ( !$this->checkFileExtension( $this->mFinalExtension, $extensions ) ) {
 629                                 $warnings['filetype-unwanted-type'] = array( $this->mFinalExtension,
 630                                         $wgLang->commaList( $extensions ), count( $extensions ) );
 631                         }
 632                 }
 633
 634                 global $wgUploadSizeWarning;
 635                 if ( $wgUploadSizeWarning && ( $this->mFileSize > $wgUploadSizeWarning ) ) {
 636                         $warnings['large-file'] = array( $wgUploadSizeWarning, $this->mFileSize );
 637                 }
 638
 639                 if ( $this->mFileSize == 0 ) {
 640                         $warnings['emptyfile'] = true;
 641                 }
 642
 643                 $exists = self::getExistsWarning( $localFile );
 644                 if ( $exists !== false ) {
 645                         $warnings['exists'] = $exists;
 646                 }
 647
 648                 // Check dupes against existing files
 649                 $hash = $this->getTempFileSha1Base36();
 650                 $dupes = RepoGroup::singleton()->findBySha1( $hash );
 651                 $title = $this->getTitle();
 652                 // Remove all matches against self
 653                 foreach ( $dupes as $key => $dupe ) {
 654                         if ( $title->equals( $dupe->getTitle() ) ) {
 655                                 unset( $dupes[$key] );
 656                         }
 657                 }
 658                 if ( $dupes ) {
 659                         $warnings['duplicate'] = $dupes;
 660                 }
 661
 662                 // Check dupes against archives
 663                 $archivedImage = new ArchivedFile( null, 0, "{$hash}.{$this->mFinalExtension}" );
 664                 if ( $archivedImage->getID() > 0 ) {
 665                         if ( $archivedImage->userCan( File::DELETED_FILE ) ) {
 666                                 $warnings['duplicate-archive'] = $archivedImage->getName();
 667                         } else {
 668                                 $warnings['duplicate-archive'] = '';
 669                         }
 670                 }
 671
 672                 wfProfileOut( __METHOD__ );
 673                 return $warnings;
 674         }
 675
 676         /**
 677          * Really perform the upload. Stores the file in the local repo, watches
 678          * if necessary and runs the UploadComplete hook.
 679          *
 680          * @param string $comment
 681          * @param string $pageText
 682          * @param bool $watch
 683          * @param User $user
 684          *
 685          * @return Status Indicating the whether the upload succeeded.
 686          */
 687         public function performUpload( $comment, $pageText, $watch, $user ) {
 688                 wfProfileIn( __METHOD__ );
 689
 690                 $status = $this->getLocalFile()->upload(
 691                         $this->mTempPath,
 692                         $comment,
 693                         $pageText,
 694                         File::DELETE_SOURCE,
 695                         $this->mFileProps,
 696                         false,
 697                         $user
 698                 );
 699
 700                 if ( $status->isGood() ) {
 701                         if ( $watch ) {
 702                                 WatchAction::doWatch( $this->getLocalFile()->getTitle(), $user, WatchedItem::IGNORE_USER_RIGHTS );
 703                         }
 704                         wfRunHooks( 'UploadComplete', array( &$this ) );
 705                 }
 706
 707                 wfProfileOut( __METHOD__ );
 708                 return $status;
 709         }
 710
 711         /**
 712          * Returns the title of the file to be uploaded. Sets mTitleError in case
 713          * the name was illegal.
 714          *
 715          * @return Title The title of the file or null in case the name was illegal
 716          */
 717         public function getTitle() {
 718                 if ( $this->mTitle !== false ) {
 719                         return $this->mTitle;
 720                 }
 721                 /* Assume that if a user specified File:Something.jpg, this is an error
 722                  * and that the namespace prefix needs to be stripped of.
 723                  */
 724                 $title = Title::newFromText( $this->mDesiredDestName );
 725                 if ( $title && $title->getNamespace() == NS_FILE ) {
 726                         $this->mFilteredName = $title->getDBkey();
 727                 } else {
 728                         $this->mFilteredName = $this->mDesiredDestName;
 729                 }
 730
 731                 # oi_archive_name is max 255 bytes, which include a timestamp and an
 732                 # exclamation mark, so restrict file name to 240 bytes.
 733                 if ( strlen( $this->mFilteredName ) > 240 ) {
 734                         $this->mTitleError = self::FILENAME_TOO_LONG;
 735                         $this->mTitle = null;
 736                         return $this->mTitle;
 737                 }
 738
 739                 /**
 740                  * Chop off any directories in the given filename. Then
 741                  * filter out illegal characters, and try to make a legible name
 742                  * out of it. We'll strip some silently that Title would die on.
 743                  */
 744                 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mFilteredName );
 745                 /* Normalize to title form before we do any further processing */
 746                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 747                 if ( is_null( $nt ) ) {
 748                         $this->mTitleError = self::ILLEGAL_FILENAME;
 749                         $this->mTitle = null;
 750                         return $this->mTitle;
 751                 }
 752                 $this->mFilteredName = $nt->getDBkey();
 753
 754                 /**
 755                  * We'll want to blacklist against *any* 'extension', and use
 756                  * only the final one for the whitelist.
 757                  */
 758                 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
 759
 760                 if ( count( $ext ) ) {
 761                         $this->mFinalExtension = trim( $ext[count( $ext ) - 1] );
 762                 } else {
 763                         $this->mFinalExtension = '';
 764
 765                         # No extension, try guessing one
 766                         $magic = MimeMagic::singleton();
 767                         $mime = $magic->guessMimeType( $this->mTempPath );
 768                         if ( $mime !== 'unknown/unknown' ) {
 769                                 # Get a space separated list of extensions
 770                                 $extList = $magic->getExtensionsForType( $mime );
 771                                 if ( $extList ) {
 772                                         # Set the extension to the canonical extension
 773                                         $this->mFinalExtension = strtok( $extList, ' ' );
 774
 775                                         # Fix up the other variables
 776                                         $this->mFilteredName .= ".{$this->mFinalExtension}";
 777                                         $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 778                                         $ext = array( $this->mFinalExtension );
 779                                 }
 780                         }
 781                 }
 782
 783                 /* Don't allow users to override the blacklist (check file extension) */
 784                 global $wgCheckFileExtensions, $wgStrictFileExtensions;
 785                 global $wgFileExtensions, $wgFileBlacklist;
 786
 787                 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
 788
 789                 if ( $this->mFinalExtension == '' ) {
 790                         $this->mTitleError = self::FILETYPE_MISSING;
 791                         $this->mTitle = null;
 792                         return $this->mTitle;
 793                 } elseif ( $blackListedExtensions ||
 794                                 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
 795                                         !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) ) ) {
 796                         $this->mBlackListedExtensions = $blackListedExtensions;
 797                         $this->mTitleError = self::FILETYPE_BADTYPE;
 798                         $this->mTitle = null;
 799                         return $this->mTitle;
 800                 }
 801
 802                 // Windows may be broken with special characters, see bug XXX
 803                 if ( wfIsWindows() && !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() ) ) {
 804                         $this->mTitleError = self::WINDOWS_NONASCII_FILENAME;
 805                         $this->mTitle = null;
 806                         return $this->mTitle;
 807                 }
 808
 809                 # If there was more than one "extension", reassemble the base
 810                 # filename to prevent bogus complaints about length
 811                 if ( count( $ext ) > 1 ) {
 812                         for ( $i = 0; $i < count( $ext ) - 1; $i++ ) {
 813                                 $partname .= '.' . $ext[$i];
 814                         }
 815                 }
 816
 817                 if ( strlen( $partname ) < 1 ) {
 818                         $this->mTitleError = self::MIN_LENGTH_PARTNAME;
 819                         $this->mTitle = null;
 820                         return $this->mTitle;
 821                 }
 822
 823                 $this->mTitle = $nt;
 824                 return $this->mTitle;
 825         }
 826
 827         /**
 828          * Return the local file and initializes if necessary.
 829          *
 830          * @return LocalFile|null
 831          */
 832         public function getLocalFile() {
 833                 if ( is_null( $this->mLocalFile ) ) {
 834                         $nt = $this->getTitle();
 835                         $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
 836                 }
 837                 return $this->mLocalFile;
 838         }
 839
 840         /**
 841          * If the user does not supply all necessary information in the first upload form submission (either by accident or
 842          * by design) then we may want to stash the file temporarily, get more information, and publish the file later.
 843          *
 844          * This method will stash a file in a temporary directory for later processing, and save the necessary descriptive info
 845          * into the database.
 846          * This method returns the file object, which also has a 'fileKey' property which can be passed through a form or
 847          * API request to find this stashed file again.
 848          *
 849          * @param User $user
 850          * @return UploadStashFile Stashed file
 851          */
 852         public function stashFile( User $user = null ) {
 853                 // was stashSessionFile
 854                 wfProfileIn( __METHOD__ );
 855
 856                 $stash = RepoGroup::singleton()->getLocalRepo()->getUploadStash( $user );
 857                 $file = $stash->stashFile( $this->mTempPath, $this->getSourceType() );
 858                 $this->mLocalFile = $file;
 859
 860                 wfProfileOut( __METHOD__ );
 861                 return $file;
 862         }
 863
 864         /**
 865          * Stash a file in a temporary directory, returning a key which can be used to find the file again. See stashFile().
 866          *
 867          * @return string File key
 868          */
 869         public function stashFileGetKey() {
 870                 return $this->stashFile()->getFileKey();
 871         }
 872
 873         /**
 874          * alias for stashFileGetKey, for backwards compatibility
 875          *
 876          * @return string File key
 877          */
 878         public function stashSession() {
 879                 return $this->stashFileGetKey();
 880         }
 881
 882         /**
 883          * If we've modified the upload file we need to manually remove it
 884          * on exit to clean up.
 885          */
 886         public function cleanupTempFile() {
 887                 if ( $this->mRemoveTempFile && $this->mTempPath && file_exists( $this->mTempPath ) ) {
 888                         wfDebug( __METHOD__ . ": Removing temporary file {$this->mTempPath}\n" );
 889                         unlink( $this->mTempPath );
 890                 }
 891         }
 892
 893         public function getTempPath() {
 894                 return $this->mTempPath;
 895         }
 896
 897         /**
 898          * Split a file into a base name and all dot-delimited 'extensions'
 899          * on the end. Some web server configurations will fall back to
 900          * earlier pseudo-'extensions' to determine type and execute
 901          * scripts, so the blacklist needs to check them all.
 902          *
 903          * @param string $filename
 904          * @return array
 905          */
 906         public static function splitExtensions( $filename ) {
 907                 $bits = explode( '.', $filename );
 908                 $basename = array_shift( $bits );
 909                 return array( $basename, $bits );
 910         }
 911
 912         /**
 913          * Perform case-insensitive match against a list of file extensions.
 914          * Returns true if the extension is in the list.
 915          *
 916          * @param string $ext
 917          * @param array $list
 918          * @return bool
 919          */
 920         public static function checkFileExtension( $ext, $list ) {
 921                 return in_array( strtolower( $ext ), $list );
 922         }
 923
 924         /**
 925          * Perform case-insensitive match against a list of file extensions.
 926          * Returns an array of matching extensions.
 927          *
 928          * @param array $ext
 929          * @param array $list
 930          * @return bool
 931          */
 932         public static function checkFileExtensionList( $ext, $list ) {
 933                 return array_intersect( array_map( 'strtolower', $ext ), $list );
 934         }
 935
 936         /**
 937          * Checks if the mime type of the uploaded file matches the file extension.
 938          *
 939          * @param string $mime The mime type of the uploaded file
 940          * @param string $extension The filename extension that the file is to be served with
 941          * @return bool
 942          */
 943         public static function verifyExtension( $mime, $extension ) {
 944                 $magic = MimeMagic::singleton();
 945
 946                 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' ) {
 947                         if ( !$magic->isRecognizableExtension( $extension ) ) {
 948                                 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
 949                                         "unrecognized extension '$extension', can't verify\n" );
 950                                 return true;
 951                         } else {
 952                                 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; " .
 953                                         "recognized extension '$extension', so probably invalid file\n" );
 954                                 return false;
 955                         }
 956                 }
 957
 958                 $match = $magic->isMatchingExtension( $extension, $mime );
 959
 960                 if ( $match === null ) {
 961                         if ( $magic->getTypesForExtension( $extension ) !== null ) {
 962                                 wfDebug( __METHOD__ . ": No extension known for $mime, but we know a mime for $extension\n" );
 963                                 return false;
 964                         } else {
 965                                 wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
 966                                 return true;
 967                         }
 968                 } elseif ( $match === true ) {
 969                         wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
 970
 971                         #TODO: if it's a bitmap, make sure PHP or ImageMagic resp. can handle it!
 972                         return true;
 973
 974                 } else {
 975                         wfDebug( __METHOD__ . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
 976                         return false;
 977                 }
 978         }
 979
 980         /**
 981          * Heuristic for detecting files that *could* contain JavaScript instructions or
 982          * things that may look like HTML to a browser and are thus
 983          * potentially harmful. The present implementation will produce false
 984          * positives in some situations.
 985          *
 986          * @param string $file Pathname to the temporary upload file
 987          * @param string $mime The mime type of the file
 988          * @param string $extension The extension of the file
 989          * @return bool True if the file contains something looking like embedded scripts
 990          */
 991         public static function detectScript( $file, $mime, $extension ) {
 992                 global $wgAllowTitlesInSVG;
 993                 wfProfileIn( __METHOD__ );
 994
 995                 # ugly hack: for text files, always look at the entire file.
 996                 # For binary field, just check the first K.
 997
 998                 if ( strpos( $mime, 'text/' ) === 0 ) {
 999                         $chunk = file_get_contents( $file );
1000                 } else {
1001                         $fp = fopen( $file, 'rb' );
1002                         $chunk = fread( $fp, 1024 );
1003                         fclose( $fp );
1004                 }
1005
1006                 $chunk = strtolower( $chunk );
1007
1008                 if ( !$chunk ) {
1009                         wfProfileOut( __METHOD__ );
1010                         return false;
1011                 }
1012
1013                 # decode from UTF-16 if needed (could be used for obfuscation).
1014                 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1015                         $enc = 'UTF-16BE';
1016                 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1017                         $enc = 'UTF-16LE';
1018                 } else {
1019                         $enc = null;
1020                 }
1021
1022                 if ( $enc ) {
1023                         $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1024                 }
1025
1026                 $chunk = trim( $chunk );
1027
1028                 # @todo FIXME: Convert from UTF-16 if necessary!
1029                 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
1030
1031                 # check for HTML doctype
1032                 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1033                         wfProfileOut( __METHOD__ );
1034                         return true;
1035                 }
1036
1037                 // Some browsers will interpret obscure xml encodings as UTF-8, while
1038                 // PHP/expat will interpret the given encoding in the xml declaration (bug 47304)
1039                 if ( $extension == 'svg' || strpos( $mime, 'image/svg' ) === 0 ) {
1040                         if ( self::checkXMLEncodingMissmatch( $file ) ) {
1041                                 wfProfileOut( __METHOD__ );
1042                                 return true;
1043                         }
1044                 }
1045
1046                 /**
1047                  * Internet Explorer for Windows performs some really stupid file type
1048                  * autodetection which can cause it to interpret valid image files as HTML
1049                  * and potentially execute JavaScript, creating a cross-site scripting
1050                  * attack vectors.
1051                  *
1052                  * Apple's Safari browser also performs some unsafe file type autodetection
1053                  * which can cause legitimate files to be interpreted as HTML if the
1054                  * web server is not correctly configured to send the right content-type
1055                  * (or if you're really uploading plain text and octet streams!)
1056                  *
1057                  * Returns true if IE is likely to mistake the given file for HTML.
1058                  * Also returns true if Safari would mistake the given file for HTML
1059                  * when served with a generic content-type.
1060                  */
1061                 $tags = array(
1062                         '<a href',
1063                         '<body',
1064                         '<head',
1065                         '<html',   #also in safari
1066                         '<img',
1067                         '<pre',
1068                         '<script', #also in safari
1069                         '<table'
1070                 );
1071
1072                 if ( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
1073                         $tags[] = '<title';
1074                 }
1075
1076                 foreach ( $tags as $tag ) {
1077                         if ( false !== strpos( $chunk, $tag ) ) {
1078                                 wfDebug( __METHOD__ . ": found something that may make it be mistaken for html: $tag\n" );
1079                                 wfProfileOut( __METHOD__ );
1080                                 return true;
1081                         }
1082                 }
1083
1084                 /*
1085                  * look for JavaScript
1086                  */
1087
1088                 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1089                 $chunk = Sanitizer::decodeCharReferences( $chunk );
1090
1091                 # look for script-types
1092                 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1093                         wfDebug( __METHOD__ . ": found script types\n" );
1094                         wfProfileOut( __METHOD__ );
1095                         return true;
1096                 }
1097
1098                 # look for html-style script-urls
1099                 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1100                         wfDebug( __METHOD__ . ": found html-style script urls\n" );
1101                         wfProfileOut( __METHOD__ );
1102                         return true;
1103                 }
1104
1105                 # look for css-style script-urls
1106                 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1107                         wfDebug( __METHOD__ . ": found css-style script urls\n" );
1108                         wfProfileOut( __METHOD__ );
1109                         return true;
1110                 }
1111
1112                 wfDebug( __METHOD__ . ": no scripts found\n" );
1113                 wfProfileOut( __METHOD__ );
1114                 return false;
1115         }
1116
1117         /**
1118          * Check a whitelist of xml encodings that are known not to be interpreted differently
1119          * by the server's xml parser (expat) and some common browsers.
1120          *
1121          * @param string $file Pathname to the temporary upload file
1122          * @return bool True if the file contains an encoding that could be misinterpreted
1123          */
1124         public static function checkXMLEncodingMissmatch( $file ) {
1125                 global $wgSVGMetadataCutoff;
1126                 $contents = file_get_contents( $file, false, null, -1, $wgSVGMetadataCutoff );
1127                 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1128
1129                 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1130                         if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1131                                 && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1132                         ) {
1133                                 wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1134                                 return true;
1135                         }
1136                 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1137                         // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1138                         // bytes. There shouldn't be a legitimate reason for this to happen.
1139                         wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1140                         return true;
1141                 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1142                         // EBCDIC encoded XML
1143                         wfDebug( __METHOD__ . ": EBCDIC Encoded XML\n" );
1144                         return true;
1145                 }
1146
1147                 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1148                 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1149                 $attemptEncodings = array( 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' );
1150                 foreach ( $attemptEncodings as $encoding ) {
1151                         wfSuppressWarnings();
1152                         $str = iconv( $encoding, 'UTF-8', $contents );
1153                         wfRestoreWarnings();
1154                         if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1155                                 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1156                                         && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1157                                 ) {
1158                                         wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1159                                         return true;
1160                                 }
1161                         } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1162                                 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1163                                 // bytes. There shouldn't be a legitimate reason for this to happen.
1164                                 wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1165                                 return true;
1166                         }
1167                 }
1168
1169                 return false;
1170         }
1171
1172         /**
1173          * @param string $filename
1174          * @return mixed False of the file is verified (does not contain scripts), array otherwise.
1175          */
1176         protected function detectScriptInSvg( $filename ) {
1177                 $this->mSVGNSError = false;
1178                 $check = new XmlTypeCheck(
1179                         $filename,
1180                         array( $this, 'checkSvgScriptCallback' ),
1181                         true,
1182                         array( 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback' )
1183                 );
1184                 if ( $check->wellFormed !== true ) {
1185                         // Invalid xml (bug 58553)
1186                         return array( 'uploadinvalidxml' );
1187                 } elseif ( $check->filterMatch ) {
1188                         if ( $this->mSVGNSError ) {
1189                                 return array( 'uploadscriptednamespace', $this->mSVGNSError );
1190                         }
1191                         return array( 'uploadscripted' );
1192                 }
1193                 return false;
1194         }
1195
1196         /**
1197          * Callback to filter SVG Processing Instructions.
1198          * @param string $target processing instruction name
1199          * @param string $data processing instruction attribute and value
1200          * @return bool (true if the filter identified something bad)
1201          */
1202         public static function checkSvgPICallback( $target, $data ) {
1203                 // Don't allow external stylesheets (bug 57550)
1204                 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1205                         return true;
1206                 }
1207                 return false;
1208         }
1209
1210         /**
1211          * @todo Replace this with a whitelist filter!
1212          * @param string $element
1213          * @param array $attribs
1214          * @return bool
1215          */
1216         public function checkSvgScriptCallback( $element, $attribs ) {
1217                 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1218
1219                 static $validNamespaces = array(
1220                         '',
1221                         'adobe:ns:meta/',
1222                         'http://creativecommons.org/ns#',
1223                         'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1224                         'http://ns.adobe.com/adobeillustrator/10.0/',
1225                         'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1226                         'http://ns.adobe.com/extensibility/1.0/',
1227                         'http://ns.adobe.com/flows/1.0/',
1228                         'http://ns.adobe.com/illustrator/1.0/',
1229                         'http://ns.adobe.com/imagereplacement/1.0/',
1230                         'http://ns.adobe.com/pdf/1.3/',
1231                         'http://ns.adobe.com/photoshop/1.0/',
1232                         'http://ns.adobe.com/saveforweb/1.0/',
1233                         'http://ns.adobe.com/variables/1.0/',
1234                         'http://ns.adobe.com/xap/1.0/',
1235                         'http://ns.adobe.com/xap/1.0/g/',
1236                         'http://ns.adobe.com/xap/1.0/g/img/',
1237                         'http://ns.adobe.com/xap/1.0/mm/',
1238                         'http://ns.adobe.com/xap/1.0/rights/',
1239                         'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1240                         'http://ns.adobe.com/xap/1.0/stype/font#',
1241                         'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1242                         'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1243                         'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1244                         'http://ns.adobe.com/xap/1.0/t/pg/',
1245                         'http://purl.org/dc/elements/1.1/',
1246                         'http://purl.org/dc/elements/1.1',
1247                         'http://schemas.microsoft.com/visio/2003/svgextensions/',
1248                         'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1249                         'http://web.resource.org/cc/',
1250                         'http://www.freesoftware.fsf.org/bkchem/cdml',
1251                         'http://www.inkscape.org/namespaces/inkscape',
1252                         'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1253                         'http://www.w3.org/2000/svg',
1254                 );
1255
1256                 if ( !in_array( $namespace, $validNamespaces ) ) {
1257                         wfDebug( __METHOD__ . ": Non-svg namespace '$namespace' in uploaded file.\n" );
1258                         // @TODO return a status object to a closure in XmlTypeCheck, for MW1.21+
1259                         $this->mSVGNSError = $namespace;
1260                         return true;
1261                 }
1262
1263                 /*
1264                  * check for elements that can contain javascript
1265                  */
1266                 if ( $strippedElement == 'script' ) {
1267                         wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
1268                         return true;
1269                 }
1270
1271                 # 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>
1272                 if ( $strippedElement == 'handler' ) {
1273                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1274                         return true;
1275                 }
1276
1277                 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1278                 if ( $strippedElement == 'stylesheet' ) {
1279                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1280                         return true;
1281                 }
1282
1283                 # Block iframes, in case they pass the namespace check
1284                 if ( $strippedElement == 'iframe' ) {
1285                         wfDebug( __METHOD__ . ": iframe in uploaded file.\n" );
1286                         return true;
1287                 }
1288
1289                 foreach ( $attribs as $attrib => $value ) {
1290                         $stripped = $this->stripXmlNamespace( $attrib );
1291                         $value = strtolower( $value );
1292
1293                         if ( substr( $stripped, 0, 2 ) == 'on' ) {
1294                                 wfDebug( __METHOD__ . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1295                                 return true;
1296                         }
1297
1298                         # href with javascript target
1299                         if ( $stripped == 'href' && strpos( strtolower( $value ), 'javascript:' ) !== false ) {
1300                                 wfDebug( __METHOD__ . ": Found script in href attribute '$attrib'='$value' in uploaded file.\n" );
1301                                 return true;
1302                         }
1303
1304                         # href with embedded svg as target
1305                         if ( $stripped == 'href' && preg_match( '!data:[^,]*image/svg[^,]*,!sim', $value ) ) {
1306                                 wfDebug( __METHOD__ . ": Found href to embedded svg \"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1307                                 return true;
1308                         }
1309
1310                         # href with embedded (text/xml) svg as target
1311                         if ( $stripped == 'href' && preg_match( '!data:[^,]*text/xml[^,]*,!sim', $value ) ) {
1312                                 wfDebug( __METHOD__ . ": Found href to embedded svg \"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1313                                 return true;
1314                         }
1315
1316                         # use set/animate to add event-handler attribute to parent
1317                         if ( ( $strippedElement == 'set' || $strippedElement == 'animate' ) && $stripped == 'attributename' && substr( $value, 0, 2 ) == 'on' ) {
1318                                 wfDebug( __METHOD__ . ": Found svg setting event-handler attribute with \"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1319                                 return true;
1320                         }
1321
1322                         # use set to add href attribute to parent element
1323                         if ( $strippedElement == 'set' && $stripped == 'attributename' && strpos( $value, 'href' ) !== false ) {
1324                                 wfDebug( __METHOD__ . ": Found svg setting href attribute '$value' in uploaded file.\n" );
1325                                 return true;
1326                         }
1327
1328                         # use set to add a remote / data / script target to an element
1329                         if ( $strippedElement == 'set' && $stripped == 'to' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1330                                 wfDebug( __METHOD__ . ": Found svg setting attribute to '$value' in uploaded file.\n" );
1331                                 return true;
1332                         }
1333
1334                         # use handler attribute with remote / data / script
1335                         if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1336                                 wfDebug( __METHOD__ . ": Found svg setting handler with remote/data/script '$attrib'='$value' in uploaded file.\n" );
1337                                 return true;
1338                         }
1339
1340                         # use CSS styles to bring in remote code
1341                         # catch url("http:..., url('http:..., url(http:..., but not url("#..., url('#..., url(#....
1342                         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 ) ) {
1343                                 foreach ( $matches[1] as $match ) {
1344                                         if ( !preg_match( '!(?:font|clip-path|fill|filter|marker|marker-end|marker-mid|marker-start|mask|stroke)\s*:\s*url\s*\(\s*(#|\'#|"#)!sim', $match ) ) {
1345                                                 wfDebug( __METHOD__ . ": Found svg setting a style with remote url '$attrib'='$value' in uploaded file.\n" );
1346                                                 return true;
1347                                         }
1348                                 }
1349                         }
1350
1351                         # image filters can pull in url, which could be svg that executes scripts
1352                         if ( $strippedElement == 'image' && $stripped == 'filter' && preg_match( '!url\s*\(!sim', $value ) ) {
1353                                 wfDebug( __METHOD__ . ": Found image filter with url: \"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1354                                 return true;
1355                         }
1356
1357                 }
1358
1359                 return false; //No scripts detected
1360         }
1361
1362         /**
1363          * Divide the element name passed by the xml parser to the callback into URI and prifix.
1364          * @param string $name
1365          * @return array Containing the namespace URI and prefix
1366          */
1367         private static function splitXmlNamespace( $element ) {
1368                 // 'http://www.w3.org/2000/svg:script' -> array( 'http://www.w3.org/2000/svg', 'script' )
1369                 $parts = explode( ':', strtolower( $element ) );
1370                 $name = array_pop( $parts );
1371                 $ns = implode( ':', $parts );
1372                 return array( $ns, $name );
1373         }
1374
1375         /**
1376          * @param string $name
1377          * @return string
1378          */
1379         private function stripXmlNamespace( $name ) {
1380                 // 'http://www.w3.org/2000/svg:script' -> 'script'
1381                 $parts = explode( ':', strtolower( $name ) );
1382                 return array_pop( $parts );
1383         }
1384
1385         /**
1386          * Generic wrapper function for a virus scanner program.
1387          * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1388          * $wgAntivirusRequired may be used to deny upload if the scan fails.
1389          *
1390          * @param string $file Pathname to the temporary upload file
1391          * @return mixed False if not virus is found, null if the scan fails or is disabled,
1392          *   or a string containing feedback from the virus scanner if a virus was found.
1393          *   If textual feedback is missing but a virus was found, this function returns true.
1394          */
1395         public static function detectVirus( $file ) {
1396                 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1397                 wfProfileIn( __METHOD__ );
1398
1399                 if ( !$wgAntivirus ) {
1400                         wfDebug( __METHOD__ . ": virus scanner disabled\n" );
1401                         wfProfileOut( __METHOD__ );
1402                         return null;
1403                 }
1404
1405                 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1406                         wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
1407                         $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1408                                 array( 'virus-badscanner', $wgAntivirus ) );
1409                         wfProfileOut( __METHOD__ );
1410                         return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1411                 }
1412
1413                 # look up scanner configuration
1414                 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1415                 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1416                 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
1417                         $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
1418
1419                 if ( strpos( $command, "%f" ) === false ) {
1420                         # simple pattern: append file to scan
1421                         $command .= " " . wfEscapeShellArg( $file );
1422                 } else {
1423                         # complex pattern: replace "%f" with file to scan
1424                         $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
1425                 }
1426
1427                 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
1428
1429                 # execute virus scanner
1430                 $exitCode = false;
1431
1432                 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1433                 #      that does not seem to be worth the pain.
1434                 #      Ask me (Duesentrieb) about it if it's ever needed.
1435                 $output = wfShellExecWithStderr( $command, $exitCode );
1436
1437                 # map exit code to AV_xxx constants.
1438                 $mappedCode = $exitCode;
1439                 if ( $exitCodeMap ) {
1440                         if ( isset( $exitCodeMap[$exitCode] ) ) {
1441                                 $mappedCode = $exitCodeMap[$exitCode];
1442                         } elseif ( isset( $exitCodeMap["*"] ) ) {
1443                                 $mappedCode = $exitCodeMap["*"];
1444                         }
1445                 }
1446
1447                 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1448                  * so we need the strict equalities === and thus can't use a switch here
1449                  */
1450                 if ( $mappedCode === AV_SCAN_FAILED ) {
1451                         # scan failed (code was mapped to false by $exitCodeMap)
1452                         wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
1453
1454                         $output = $wgAntivirusRequired ? wfMessage( 'virus-scanfailed', array( $exitCode ) )->text() : null;
1455                 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
1456                         # scan failed because filetype is unknown (probably imune)
1457                         wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
1458                         $output = null;
1459                 } elseif ( $mappedCode === AV_NO_VIRUS ) {
1460                         # no virus found
1461                         wfDebug( __METHOD__ . ": file passed virus scan.\n" );
1462                         $output = false;
1463                 } else {
1464                         $output = trim( $output );
1465
1466                         if ( !$output ) {
1467                                 $output = true; #if there's no output, return true
1468                         } elseif ( $msgPattern ) {
1469                                 $groups = array();
1470                                 if ( preg_match( $msgPattern, $output, $groups ) ) {
1471                                         if ( $groups[1] ) {
1472                                                 $output = $groups[1];
1473                                         }
1474                                 }
1475                         }
1476
1477                         wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
1478                 }
1479
1480                 wfProfileOut( __METHOD__ );
1481                 return $output;
1482         }
1483
1484         /**
1485          * Check if there's an overwrite conflict and, if so, if restrictions
1486          * forbid this user from performing the upload.
1487          *
1488          * @param User $user
1489          *
1490          * @return mixed True on success, array on failure
1491          */
1492         private function checkOverwrite( $user ) {
1493                 // First check whether the local file can be overwritten
1494                 $file = $this->getLocalFile();
1495                 if ( $file->exists() ) {
1496                         if ( !self::userCanReUpload( $user, $file ) ) {
1497                                 return array( 'fileexists-forbidden', $file->getName() );
1498                         } else {
1499                                 return true;
1500                         }
1501                 }
1502
1503                 /* Check shared conflicts: if the local file does not exist, but
1504                  * wfFindFile finds a file, it exists in a shared repository.
1505                  */
1506                 $file = wfFindFile( $this->getTitle() );
1507                 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1508                         return array( 'fileexists-shared-forbidden', $file->getName() );
1509                 }
1510
1511                 return true;
1512         }
1513
1514         /**
1515          * Check if a user is the last uploader
1516          *
1517          * @param User $user
1518          * @param string $img Image name
1519          * @return bool
1520          */
1521         public static function userCanReUpload( User $user, $img ) {
1522                 if ( $user->isAllowed( 'reupload' ) ) {
1523                         return true; // non-conditional
1524                 }
1525                 if ( !$user->isAllowed( 'reupload-own' ) ) {
1526                         return false;
1527                 }
1528                 if ( is_string( $img ) ) {
1529                         $img = wfLocalFile( $img );
1530                 }
1531                 if ( !( $img instanceof LocalFile ) ) {
1532                         return false;
1533                 }
1534
1535                 return $user->getId() == $img->getUser( 'id' );
1536         }
1537
1538         /**
1539          * Helper function that does various existence checks for a file.
1540          * The following checks are performed:
1541          * - The file exists
1542          * - Article with the same name as the file exists
1543          * - File exists with normalized extension
1544          * - The file looks like a thumbnail and the original exists
1545          *
1546          * @param File $file The File object to check
1547          * @return mixed False if the file does not exists, else an array
1548          */
1549         public static function getExistsWarning( $file ) {
1550                 if ( $file->exists() ) {
1551                         return array( 'warning' => 'exists', 'file' => $file );
1552                 }
1553
1554                 if ( $file->getTitle()->getArticleID() ) {
1555                         return array( 'warning' => 'page-exists', 'file' => $file );
1556                 }
1557
1558                 if ( $file->wasDeleted() && !$file->exists() ) {
1559                         return array( 'warning' => 'was-deleted', 'file' => $file );
1560                 }
1561
1562                 if ( strpos( $file->getName(), '.' ) == false ) {
1563                         $partname = $file->getName();
1564                         $extension = '';
1565                 } else {
1566                         $n = strrpos( $file->getName(), '.' );
1567                         $extension = substr( $file->getName(), $n + 1 );
1568                         $partname = substr( $file->getName(), 0, $n );
1569                 }
1570                 $normalizedExtension = File::normalizeExtension( $extension );
1571
1572                 if ( $normalizedExtension != $extension ) {
1573                         // We're not using the normalized form of the extension.
1574                         // Normal form is lowercase, using most common of alternate
1575                         // extensions (eg 'jpg' rather than 'JPEG').
1576                         //
1577                         // Check for another file using the normalized form...
1578                         $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
1579                         $file_lc = wfLocalFile( $nt_lc );
1580
1581                         if ( $file_lc->exists() ) {
1582                                 return array(
1583                                         'warning' => 'exists-normalized',
1584                                         'file' => $file,
1585                                         'normalizedFile' => $file_lc
1586                                 );
1587                         }
1588                 }
1589
1590                 // Check for files with the same name but a different extension
1591                 $similarFiles = RepoGroup::singleton()->getLocalRepo()->findFilesByPrefix(
1592                                 "{$partname}.", 1 );
1593                 if ( count( $similarFiles ) ) {
1594                         return array(
1595                                 'warning' => 'exists-normalized',
1596                                 'file' => $file,
1597                                 'normalizedFile' => $similarFiles[0],
1598                         );
1599                 }
1600
1601                 if ( self::isThumbName( $file->getName() ) ) {
1602                         # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1603                         $nt_thb = Title::newFromText( substr( $partname, strpos( $partname, '-' ) + 1 ) . '.' . $extension, NS_FILE );
1604                         $file_thb = wfLocalFile( $nt_thb );
1605                         if ( $file_thb->exists() ) {
1606                                 return array(
1607                                         'warning' => 'thumb',
1608                                         'file' => $file,
1609                                         'thumbFile' => $file_thb
1610                                 );
1611                         } else {
1612                                 // File does not exist, but we just don't like the name
1613                                 return array(
1614                                         'warning' => 'thumb-name',
1615                                         'file' => $file,
1616                                         'thumbFile' => $file_thb
1617                                 );
1618                         }
1619                 }
1620
1621                 foreach ( self::getFilenamePrefixBlacklist() as $prefix ) {
1622                         if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1623                                 return array(
1624                                         'warning' => 'bad-prefix',
1625                                         'file' => $file,
1626                                         'prefix' => $prefix
1627                                 );
1628                         }
1629                 }
1630
1631                 return false;
1632         }
1633
1634         /**
1635          * Helper function that checks whether the filename looks like a thumbnail
1636          * @param string $filename
1637          * @return bool
1638          */
1639         public static function isThumbName( $filename ) {
1640                 $n = strrpos( $filename, '.' );
1641                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
1642                 return (
1643                                         substr( $partname, 3, 3 ) == 'px-' ||
1644                                         substr( $partname, 2, 3 ) == 'px-'
1645                                 ) &&
1646                                 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
1647         }
1648
1649         /**
1650          * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
1651          *
1652          * @return array List of prefixes
1653          */
1654         public static function getFilenamePrefixBlacklist() {
1655                 $blacklist = array();
1656                 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
1657                 if ( !$message->isDisabled() ) {
1658                         $lines = explode( "\n", $message->plain() );
1659                         foreach ( $lines as $line ) {
1660                                 // Remove comment lines
1661                                 $comment = substr( trim( $line ), 0, 1 );
1662                                 if ( $comment == '#' || $comment == '' ) {
1663                                         continue;
1664                                 }
1665                                 // Remove additional comments after a prefix
1666                                 $comment = strpos( $line, '#' );
1667                                 if ( $comment > 0 ) {
1668                                         $line = substr( $line, 0, $comment - 1 );
1669                                 }
1670                                 $blacklist[] = trim( $line );
1671                         }
1672                 }
1673                 return $blacklist;
1674         }
1675
1676         /**
1677          * Gets image info about the file just uploaded.
1678          *
1679          * Also has the effect of setting metadata to be an 'indexed tag name' in returned API result if
1680          * 'metadata' was requested. Oddly, we have to pass the "result" object down just so it can do that
1681          * with the appropriate format, presumably.
1682          *
1683          * @param ApiResult $result
1684          * @return array Image info
1685          */
1686         public function getImageInfo( $result ) {
1687                 $file = $this->getLocalFile();
1688                 // TODO This cries out for refactoring. We really want to say $file->getAllInfo(); here.
1689                 // Perhaps "info" methods should be moved into files, and the API should just wrap them in queries.
1690                 if ( $file instanceof UploadStashFile ) {
1691                         $imParam = ApiQueryStashImageInfo::getPropertyNames();
1692                         $info = ApiQueryStashImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1693                 } else {
1694                         $imParam = ApiQueryImageInfo::getPropertyNames();
1695                         $info = ApiQueryImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1696                 }
1697                 return $info;
1698         }
1699
1700         /**
1701          * @param array $error
1702          * @return Status
1703          */
1704         public function convertVerifyErrorToStatus( $error ) {
1705                 $code = $error['status'];
1706                 unset( $code['status'] );
1707                 return Status::newFatal( $this->getVerificationErrorCode( $code ), $error );
1708         }
1709
1710         /**
1711          * @param null|string $forType
1712          * @return int
1713          */
1714         public static function getMaxUploadSize( $forType = null ) {
1715                 global $wgMaxUploadSize;
1716
1717                 if ( is_array( $wgMaxUploadSize ) ) {
1718                         if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
1719                                 return $wgMaxUploadSize[$forType];
1720                         } else {
1721                                 return $wgMaxUploadSize['*'];
1722                         }
1723                 } else {
1724                         return intval( $wgMaxUploadSize );
1725                 }
1726         }
1727
1728         /**
1729          * Get the current status of a chunked upload (used for polling).
1730          * The status will be read from the *current* user session.
1731          * @param string $statusKey
1732          * @return array|bool
1733          */
1734         public static function getSessionStatus( $statusKey ) {
1735                 return isset( $_SESSION[self::SESSION_STATUS_KEY][$statusKey] )
1736                         ? $_SESSION[self::SESSION_STATUS_KEY][$statusKey]
1737                         : false;
1738         }
1739
1740         /**
1741          * Set the current status of a chunked upload (used for polling).
1742          * The status will be stored in the *current* user session.
1743          * @param string $statusKey
1744          * @param array|bool $value
1745          * @return void
1746          */
1747         public static function setSessionStatus( $statusKey, $value ) {
1748                 if ( $value === false ) {
1749                         unset( $_SESSION[self::SESSION_STATUS_KEY][$statusKey] );
1750                 } else {
1751                         $_SESSION[self::SESSION_STATUS_KEY][$statusKey] = $value;
1752                 }
1753         }
1754 }