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