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 use MediaWiki\MediaWikiServices;
  24
  25 /**
  26  * @defgroup Upload Upload related
  27  */
  28
  29 /**
  30  * @ingroup Upload
  31  *
  32  * UploadBase and subclasses are the backend of MediaWiki's file uploads.
  33  * The frontends are formed by ApiUpload and SpecialUpload.
  34  *
  35  * @author Brion Vibber
  36  * @author Bryan Tong Minh
  37  * @author Michael Dale
  38  */
  39 abstract class UploadBase {
  40         /** @var string Local file system path to the file to upload (or a local copy) */
  41         protected $mTempPath;
  42         /** @var TempFSFile|null Wrapper to handle deleting the temp file */
  43         protected $tempFileObj;
  44
  45         protected $mDesiredDestName, $mDestName, $mRemoveTempFile, $mSourceType;
  46         protected $mTitle = false, $mTitleError = 0;
  47         protected $mFilteredName, $mFinalExtension;
  48         protected $mLocalFile, $mStashFile, $mFileSize, $mFileProps;
  49         protected $mBlackListedExtensions;
  50         protected $mJavaDetected, $mSVGNSError;
  51
  52         protected static $safeXmlEncodings = [
  53                 'UTF-8',
  54                 'ISO-8859-1',
  55                 'ISO-8859-2',
  56                 'UTF-16',
  57                 'UTF-32',
  58                 'WINDOWS-1250',
  59                 'WINDOWS-1251',
  60                 'WINDOWS-1252',
  61                 'WINDOWS-1253',
  62                 'WINDOWS-1254',
  63                 'WINDOWS-1255',
  64                 'WINDOWS-1256',
  65                 'WINDOWS-1257',
  66                 'WINDOWS-1258',
  67         ];
  68
  69         const SUCCESS = 0;
  70         const OK = 0;
  71         const EMPTY_FILE = 3;
  72         const MIN_LENGTH_PARTNAME = 4;
  73         const ILLEGAL_FILENAME = 5;
  74         const OVERWRITE_EXISTING_FILE = 7; # Not used anymore; handled by verifyTitlePermissions()
  75         const FILETYPE_MISSING = 8;
  76         const FILETYPE_BADTYPE = 9;
  77         const VERIFICATION_ERROR = 10;
  78         const HOOK_ABORTED = 11;
  79         const FILE_TOO_LARGE = 12;
  80         const WINDOWS_NONASCII_FILENAME = 13;
  81         const FILENAME_TOO_LONG = 14;
  82
  83         /**
  84          * @param int $error
  85          * @return string
  86          */
  87         public function getVerificationErrorCode( $error ) {
  88                 $code_to_status = [
  89                         self::EMPTY_FILE => 'empty-file',
  90                         self::FILE_TOO_LARGE => 'file-too-large',
  91                         self::FILETYPE_MISSING => 'filetype-missing',
  92                         self::FILETYPE_BADTYPE => 'filetype-banned',
  93                         self::MIN_LENGTH_PARTNAME => 'filename-tooshort',
  94                         self::ILLEGAL_FILENAME => 'illegal-filename',
  95                         self::OVERWRITE_EXISTING_FILE => 'overwrite',
  96                         self::VERIFICATION_ERROR => 'verification-error',
  97                         self::HOOK_ABORTED => 'hookaborted',
  98                         self::WINDOWS_NONASCII_FILENAME => 'windows-nonascii-filename',
  99                         self::FILENAME_TOO_LONG => 'filename-toolong',
 100                 ];
 101                 if ( isset( $code_to_status[$error] ) ) {
 102                         return $code_to_status[$error];
 103                 }
 104
 105                 return 'unknown-error';
 106         }
 107
 108         /**
 109          * Returns true if uploads are enabled.
 110          * Can be override by subclasses.
 111          * @return bool
 112          */
 113         public static function isEnabled() {
 114                 global $wgEnableUploads;
 115
 116                 if ( !$wgEnableUploads ) {
 117                         return false;
 118                 }
 119
 120                 # Check php's file_uploads setting
 121                 return wfIsHHVM() || wfIniGetBool( 'file_uploads' );
 122         }
 123
 124         /**
 125          * Returns true if the user can use this upload module or else a string
 126          * identifying the missing permission.
 127          * Can be overridden by subclasses.
 128          *
 129          * @param User $user
 130          * @return bool|string
 131          */
 132         public static function isAllowed( $user ) {
 133                 foreach ( [ 'upload', 'edit' ] as $permission ) {
 134                         if ( !$user->isAllowed( $permission ) ) {
 135                                 return $permission;
 136                         }
 137                 }
 138
 139                 return true;
 140         }
 141
 142         /**
 143          * Returns true if the user has surpassed the upload rate limit, false otherwise.
 144          *
 145          * @param User $user
 146          * @return bool
 147          */
 148         public static function isThrottled( $user ) {
 149                 return $user->pingLimiter( 'upload' );
 150         }
 151
 152         // Upload handlers. Should probably just be a global.
 153         private static $uploadHandlers = [ 'Stash', 'File', 'Url' ];
 154
 155         /**
 156          * Create a form of UploadBase depending on wpSourceType and initializes it
 157          *
 158          * @param WebRequest $request
 159          * @param string|null $type
 160          * @return null|UploadBase
 161          */
 162         public static function createFromRequest( &$request, $type = null ) {
 163                 $type = $type ? $type : $request->getVal( 'wpSourceType', 'File' );
 164
 165                 if ( !$type ) {
 166                         return null;
 167                 }
 168
 169                 // Get the upload class
 170                 $type = ucfirst( $type );
 171
 172                 // Give hooks the chance to handle this request
 173                 $className = null;
 174                 Hooks::run( 'UploadCreateFromRequest', [ $type, &$className ] );
 175                 if ( is_null( $className ) ) {
 176                         $className = 'UploadFrom' . $type;
 177                         wfDebug( __METHOD__ . ": class name: $className\n" );
 178                         if ( !in_array( $type, self::$uploadHandlers ) ) {
 179                                 return null;
 180                         }
 181                 }
 182
 183                 // Check whether this upload class is enabled
 184                 if ( !call_user_func( [ $className, 'isEnabled' ] ) ) {
 185                         return null;
 186                 }
 187
 188                 // Check whether the request is valid
 189                 if ( !call_user_func( [ $className, 'isValidRequest' ], $request ) ) {
 190                         return null;
 191                 }
 192
 193                 /** @var UploadBase $handler */
 194                 $handler = new $className;
 195
 196                 $handler->initializeFromRequest( $request );
 197
 198                 return $handler;
 199         }
 200
 201         /**
 202          * Check whether a request if valid for this handler
 203          * @param WebRequest $request
 204          * @return bool
 205          */
 206         public static function isValidRequest( $request ) {
 207                 return false;
 208         }
 209
 210         public function __construct() {
 211         }
 212
 213         /**
 214          * Returns the upload type. Should be overridden by child classes
 215          *
 216          * @since 1.18
 217          * @return string
 218          */
 219         public function getSourceType() {
 220                 return null;
 221         }
 222
 223         /**
 224          * Initialize the path information
 225          * @param string $name The desired destination name
 226          * @param string $tempPath The temporary path
 227          * @param int $fileSize The file size
 228          * @param bool $removeTempFile (false) remove the temporary file?
 229          * @throws MWException
 230          */
 231         public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
 232                 $this->mDesiredDestName = $name;
 233                 if ( FileBackend::isStoragePath( $tempPath ) ) {
 234                         throw new MWException( __METHOD__ . " given storage path `$tempPath`." );
 235                 }
 236
 237                 $this->setTempFile( $tempPath, $fileSize );
 238                 $this->mRemoveTempFile = $removeTempFile;
 239         }
 240
 241         /**
 242          * Initialize from a WebRequest. Override this in a subclass.
 243          *
 244          * @param WebRequest $request
 245          */
 246         abstract public function initializeFromRequest( &$request );
 247
 248         /**
 249          * @param string $tempPath File system path to temporary file containing the upload
 250          * @param integer $fileSize
 251          */
 252         protected function setTempFile( $tempPath, $fileSize = null ) {
 253                 $this->mTempPath = $tempPath;
 254                 $this->mFileSize = $fileSize ?: null;
 255                 if ( strlen( $this->mTempPath ) && file_exists( $this->mTempPath ) ) {
 256                         $this->tempFileObj = new TempFSFile( $this->mTempPath );
 257                         if ( !$fileSize ) {
 258                                 $this->mFileSize = filesize( $this->mTempPath );
 259                         }
 260                 } else {
 261                         $this->tempFileObj = null;
 262                 }
 263         }
 264
 265         /**
 266          * Fetch the file. Usually a no-op
 267          * @return Status
 268          */
 269         public function fetchFile() {
 270                 return Status::newGood();
 271         }
 272
 273         /**
 274          * Return true if the file is empty
 275          * @return bool
 276          */
 277         public function isEmptyFile() {
 278                 return empty( $this->mFileSize );
 279         }
 280
 281         /**
 282          * Return the file size
 283          * @return int
 284          */
 285         public function getFileSize() {
 286                 return $this->mFileSize;
 287         }
 288
 289         /**
 290          * Get the base 36 SHA1 of the file
 291          * @return string
 292          */
 293         public function getTempFileSha1Base36() {
 294                 return FSFile::getSha1Base36FromPath( $this->mTempPath );
 295         }
 296
 297         /**
 298          * @param string $srcPath The source path
 299          * @return string|bool The real path if it was a virtual URL Returns false on failure
 300          */
 301         public function getRealPath( $srcPath ) {
 302                 $repo = RepoGroup::singleton()->getLocalRepo();
 303                 if ( $repo->isVirtualUrl( $srcPath ) ) {
 304                         /** @todo Just make uploads work with storage paths UploadFromStash
 305                          *  loads files via virtual URLs.
 306                          */
 307                         $tmpFile = $repo->getLocalCopy( $srcPath );
 308                         if ( $tmpFile ) {
 309                                 $tmpFile->bind( $this ); // keep alive with $this
 310                         }
 311                         $path = $tmpFile ? $tmpFile->getPath() : false;
 312                 } else {
 313                         $path = $srcPath;
 314                 }
 315
 316                 return $path;
 317         }
 318
 319         /**
 320          * Verify whether the upload is sane.
 321          * @return mixed Const self::OK or else an array with error information
 322          */
 323         public function verifyUpload() {
 324                 /**
 325                  * If there was no filename or a zero size given, give up quick.
 326                  */
 327                 if ( $this->isEmptyFile() ) {
 328                         return [ 'status' => self::EMPTY_FILE ];
 329                 }
 330
 331                 /**
 332                  * Honor $wgMaxUploadSize
 333                  */
 334                 $maxSize = self::getMaxUploadSize( $this->getSourceType() );
 335                 if ( $this->mFileSize > $maxSize ) {
 336                         return [
 337                                 'status' => self::FILE_TOO_LARGE,
 338                                 'max' => $maxSize,
 339                         ];
 340                 }
 341
 342                 /**
 343                  * Look at the contents of the file; if we can recognize the
 344                  * type but it's corrupt or data of the wrong type, we should
 345                  * probably not accept it.
 346                  */
 347                 $verification = $this->verifyFile();
 348                 if ( $verification !== true ) {
 349                         return [
 350                                 'status' => self::VERIFICATION_ERROR,
 351                                 'details' => $verification
 352                         ];
 353                 }
 354
 355                 /**
 356                  * Make sure this file can be created
 357                  */
 358                 $result = $this->validateName();
 359                 if ( $result !== true ) {
 360                         return $result;
 361                 }
 362
 363                 $error = '';
 364                 if ( !Hooks::run( 'UploadVerification',
 365                         [ $this->mDestName, $this->mTempPath, &$error ], '1.28' )
 366                 ) {
 367                         return [ 'status' => self::HOOK_ABORTED, 'error' => $error ];
 368                 }
 369
 370                 return [ 'status' => self::OK ];
 371         }
 372
 373         /**
 374          * Verify that the name is valid and, if necessary, that we can overwrite
 375          *
 376          * @return mixed True if valid, otherwise and array with 'status'
 377          * and other keys
 378          */
 379         public function validateName() {
 380                 $nt = $this->getTitle();
 381                 if ( is_null( $nt ) ) {
 382                         $result = [ 'status' => $this->mTitleError ];
 383                         if ( $this->mTitleError == self::ILLEGAL_FILENAME ) {
 384                                 $result['filtered'] = $this->mFilteredName;
 385                         }
 386                         if ( $this->mTitleError == self::FILETYPE_BADTYPE ) {
 387                                 $result['finalExt'] = $this->mFinalExtension;
 388                                 if ( count( $this->mBlackListedExtensions ) ) {
 389                                         $result['blacklistedExt'] = $this->mBlackListedExtensions;
 390                                 }
 391                         }
 392
 393                         return $result;
 394                 }
 395                 $this->mDestName = $this->getLocalFile()->getName();
 396
 397                 return true;
 398         }
 399
 400         /**
 401          * Verify the MIME type.
 402          *
 403          * @note Only checks that it is not an evil MIME. The "does it have
 404          *  correct extension given its MIME type?" check is in verifyFile.
 405          *  in `verifyFile()` that MIME type and file extension correlate.
 406          * @param string $mime Representing the MIME
 407          * @return mixed True if the file is verified, an array otherwise
 408          */
 409         protected function verifyMimeType( $mime ) {
 410                 global $wgVerifyMimeType;
 411                 if ( $wgVerifyMimeType ) {
 412                         wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>\n" );
 413                         global $wgMimeTypeBlacklist;
 414                         if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
 415                                 return [ 'filetype-badmime', $mime ];
 416                         }
 417
 418                         # Check what Internet Explorer would detect
 419                         $fp = fopen( $this->mTempPath, 'rb' );
 420                         $chunk = fread( $fp, 256 );
 421                         fclose( $fp );
 422
 423                         $magic = MimeMagic::singleton();
 424                         $extMime = $magic->guessTypesForExtension( $this->mFinalExtension );
 425                         $ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime );
 426                         foreach ( $ieTypes as $ieType ) {
 427                                 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
 428                                         return [ 'filetype-bad-ie-mime', $ieType ];
 429                                 }
 430                         }
 431                 }
 432
 433                 return true;
 434         }
 435
 436         /**
 437          * Verifies that it's ok to include the uploaded file
 438          *
 439          * @return mixed True of the file is verified, array otherwise.
 440          */
 441         protected function verifyFile() {
 442                 global $wgVerifyMimeType, $wgDisableUploadScriptChecks;
 443
 444                 $status = $this->verifyPartialFile();
 445                 if ( $status !== true ) {
 446                         return $status;
 447                 }
 448
 449                 $mwProps = new MWFileProps( MimeMagic::singleton() );
 450                 $this->mFileProps = $mwProps->getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 451                 $mime = $this->mFileProps['mime'];
 452
 453                 if ( $wgVerifyMimeType ) {
 454                         # XXX: Missing extension will be caught by validateName() via getTitle()
 455                         if ( $this->mFinalExtension != '' && !$this->verifyExtension( $mime, $this->mFinalExtension ) ) {
 456                                 return [ 'filetype-mime-mismatch', $this->mFinalExtension, $mime ];
 457                         }
 458                 }
 459
 460                 # check for htmlish code and javascript
 461                 if ( !$wgDisableUploadScriptChecks ) {
 462                         if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 463                                 $svgStatus = $this->detectScriptInSvg( $this->mTempPath, false );
 464                                 if ( $svgStatus !== false ) {
 465                                         return $svgStatus;
 466                                 }
 467                         }
 468                 }
 469
 470                 $handler = MediaHandler::getHandler( $mime );
 471                 if ( $handler ) {
 472                         $handlerStatus = $handler->verifyUpload( $this->mTempPath );
 473                         if ( !$handlerStatus->isOK() ) {
 474                                 $errors = $handlerStatus->getErrorsArray();
 475
 476                                 return reset( $errors );
 477                         }
 478                 }
 479
 480                 $error = true;
 481                 Hooks::run( 'UploadVerifyFile', [ $this, $mime, &$error ] );
 482                 if ( $error !== true ) {
 483                         if ( !is_array( $error ) ) {
 484                                 $error = [ $error ];
 485                         }
 486                         return $error;
 487                 }
 488
 489                 wfDebug( __METHOD__ . ": all clear; passing.\n" );
 490
 491                 return true;
 492         }
 493
 494         /**
 495          * A verification routine suitable for partial files
 496          *
 497          * Runs the blacklist checks, but not any checks that may
 498          * assume the entire file is present.
 499          *
 500          * @return mixed True for valid or array with error message key.
 501          */
 502         protected function verifyPartialFile() {
 503                 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
 504
 505                 # getTitle() sets some internal parameters like $this->mFinalExtension
 506                 $this->getTitle();
 507
 508                 $mwProps = new MWFileProps( MimeMagic::singleton() );
 509                 $this->mFileProps = $mwProps->getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 510
 511                 # check MIME type, if desired
 512                 $mime = $this->mFileProps['file-mime'];
 513                 $status = $this->verifyMimeType( $mime );
 514                 if ( $status !== true ) {
 515                         return $status;
 516                 }
 517
 518                 # check for htmlish code and javascript
 519                 if ( !$wgDisableUploadScriptChecks ) {
 520                         if ( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
 521                                 return [ 'uploadscripted' ];
 522                         }
 523                         if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 524                                 $svgStatus = $this->detectScriptInSvg( $this->mTempPath, true );
 525                                 if ( $svgStatus !== false ) {
 526                                         return $svgStatus;
 527                                 }
 528                         }
 529                 }
 530
 531                 # Check for Java applets, which if uploaded can bypass cross-site
 532                 # restrictions.
 533                 if ( !$wgAllowJavaUploads ) {
 534                         $this->mJavaDetected = false;
 535                         $zipStatus = ZipDirectoryReader::read( $this->mTempPath,
 536                                 [ $this, 'zipEntryCallback' ] );
 537                         if ( !$zipStatus->isOK() ) {
 538                                 $errors = $zipStatus->getErrorsArray();
 539                                 $error = reset( $errors );
 540                                 if ( $error[0] !== 'zip-wrong-format' ) {
 541                                         return $error;
 542                                 }
 543                         }
 544                         if ( $this->mJavaDetected ) {
 545                                 return [ 'uploadjava' ];
 546                         }
 547                 }
 548
 549                 # Scan the uploaded file for viruses
 550                 $virus = $this->detectVirus( $this->mTempPath );
 551                 if ( $virus ) {
 552                         return [ 'uploadvirus', $virus ];
 553                 }
 554
 555                 return true;
 556         }
 557
 558         /**
 559          * Callback for ZipDirectoryReader to detect Java class files.
 560          *
 561          * @param array $entry
 562          */
 563         public function zipEntryCallback( $entry ) {
 564                 $names = [ $entry['name'] ];
 565
 566                 // If there is a null character, cut off the name at it, because JDK's
 567                 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
 568                 // were constructed which had ".class\0" followed by a string chosen to
 569                 // make the hash collide with the truncated name, that file could be
 570                 // returned in response to a request for the .class file.
 571                 $nullPos = strpos( $entry['name'], "\000" );
 572                 if ( $nullPos !== false ) {
 573                         $names[] = substr( $entry['name'], 0, $nullPos );
 574                 }
 575
 576                 // If there is a trailing slash in the file name, we have to strip it,
 577                 // because that's what ZIP_GetEntry() does.
 578                 if ( preg_grep( '!\.class/?$!', $names ) ) {
 579                         $this->mJavaDetected = true;
 580                 }
 581         }
 582
 583         /**
 584          * Alias for verifyTitlePermissions. The function was originally
 585          * 'verifyPermissions', but that suggests it's checking the user, when it's
 586          * really checking the title + user combination.
 587          *
 588          * @param User $user User object to verify the permissions against
 589          * @return mixed An array as returned by getUserPermissionsErrors or true
 590          *   in case the user has proper permissions.
 591          */
 592         public function verifyPermissions( $user ) {
 593                 return $this->verifyTitlePermissions( $user );
 594         }
 595
 596         /**
 597          * Check whether the user can edit, upload and create the image. This
 598          * checks only against the current title; if it returns errors, it may
 599          * very well be that another title will not give errors. Therefore
 600          * isAllowed() should be called as well for generic is-user-blocked or
 601          * can-user-upload checking.
 602          *
 603          * @param User $user User object to verify the permissions against
 604          * @return mixed An array as returned by getUserPermissionsErrors or true
 605          *   in case the user has proper permissions.
 606          */
 607         public function verifyTitlePermissions( $user ) {
 608                 /**
 609                  * If the image is protected, non-sysop users won't be able
 610                  * to modify it by uploading a new revision.
 611                  */
 612                 $nt = $this->getTitle();
 613                 if ( is_null( $nt ) ) {
 614                         return true;
 615                 }
 616                 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
 617                 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
 618                 if ( !$nt->exists() ) {
 619                         $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
 620                 } else {
 621                         $permErrorsCreate = [];
 622                 }
 623                 if ( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
 624                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
 625                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
 626
 627                         return $permErrors;
 628                 }
 629
 630                 $overwriteError = $this->checkOverwrite( $user );
 631                 if ( $overwriteError !== true ) {
 632                         return [ $overwriteError ];
 633                 }
 634
 635                 return true;
 636         }
 637
 638         /**
 639          * Check for non fatal problems with the file.
 640          *
 641          * This should not assume that mTempPath is set.
 642          *
 643          * @return array Array of warnings
 644          */
 645         public function checkWarnings() {
 646                 global $wgLang;
 647
 648                 $warnings = [];
 649
 650                 $localFile = $this->getLocalFile();
 651                 $localFile->load( File::READ_LATEST );
 652                 $filename = $localFile->getName();
 653
 654                 /**
 655                  * Check whether the resulting filename is different from the desired one,
 656                  * but ignore things like ucfirst() and spaces/underscore things
 657                  */
 658                 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName );
 659                 $comparableName = Title::capitalize( $comparableName, NS_FILE );
 660
 661                 if ( $this->mDesiredDestName != $filename && $comparableName != $filename ) {
 662                         $warnings['badfilename'] = $filename;
 663                 }
 664
 665                 // Check whether the file extension is on the unwanted list
 666                 global $wgCheckFileExtensions, $wgFileExtensions;
 667                 if ( $wgCheckFileExtensions ) {
 668                         $extensions = array_unique( $wgFileExtensions );
 669                         if ( !$this->checkFileExtension( $this->mFinalExtension, $extensions ) ) {
 670                                 $warnings['filetype-unwanted-type'] = [ $this->mFinalExtension,
 671                                         $wgLang->commaList( $extensions ), count( $extensions ) ];
 672                         }
 673                 }
 674
 675                 global $wgUploadSizeWarning;
 676                 if ( $wgUploadSizeWarning && ( $this->mFileSize > $wgUploadSizeWarning ) ) {
 677                         $warnings['large-file'] = [ $wgUploadSizeWarning, $this->mFileSize ];
 678                 }
 679
 680                 if ( $this->mFileSize == 0 ) {
 681                         $warnings['empty-file'] = true;
 682                 }
 683
 684                 $hash = $this->getTempFileSha1Base36();
 685                 $exists = self::getExistsWarning( $localFile );
 686                 if ( $exists !== false ) {
 687                         $warnings['exists'] = $exists;
 688
 689                         // check if file is an exact duplicate of current file version
 690                         if ( $hash === $localFile->getSha1() ) {
 691                                 $warnings['no-change'] = $localFile;
 692                         }
 693
 694                         // check if file is an exact duplicate of older versions of this file
 695                         $history = $localFile->getHistory();
 696                         foreach ( $history as $oldFile ) {
 697                                 if ( $hash === $oldFile->getSha1() ) {
 698                                         $warnings['duplicate-version'][] = $oldFile;
 699                                 }
 700                         }
 701                 }
 702
 703                 if ( $localFile->wasDeleted() && !$localFile->exists() ) {
 704                         $warnings['was-deleted'] = $filename;
 705                 }
 706
 707                 // Check dupes against existing files
 708                 $dupes = RepoGroup::singleton()->findBySha1( $hash );
 709                 $title = $this->getTitle();
 710                 // Remove all matches against self
 711                 foreach ( $dupes as $key => $dupe ) {
 712                         if ( $title->equals( $dupe->getTitle() ) ) {
 713                                 unset( $dupes[$key] );
 714                         }
 715                 }
 716                 if ( $dupes ) {
 717                         $warnings['duplicate'] = $dupes;
 718                 }
 719
 720                 // Check dupes against archives
 721                 $archivedFile = new ArchivedFile( null, 0, '', $hash );
 722                 if ( $archivedFile->getID() > 0 ) {
 723                         if ( $archivedFile->userCan( File::DELETED_FILE ) ) {
 724                                 $warnings['duplicate-archive'] = $archivedFile->getName();
 725                         } else {
 726                                 $warnings['duplicate-archive'] = '';
 727                         }
 728                 }
 729
 730                 return $warnings;
 731         }
 732
 733         /**
 734          * Really perform the upload. Stores the file in the local repo, watches
 735          * if necessary and runs the UploadComplete hook.
 736          *
 737          * @param string $comment
 738          * @param string $pageText
 739          * @param bool $watch Whether the file page should be added to user's watchlist.
 740          *   (This doesn't check $user's permissions.)
 741          * @param User $user
 742          * @param string[] $tags Change tags to add to the log entry and page revision.
 743          *   (This doesn't check $user's permissions.)
 744          * @return Status Indicating the whether the upload succeeded.
 745          */
 746         public function performUpload( $comment, $pageText, $watch, $user, $tags = [] ) {
 747                 $this->getLocalFile()->load( File::READ_LATEST );
 748                 $props = $this->mFileProps;
 749
 750                 $error = null;
 751                 Hooks::run( 'UploadVerifyUpload', [ $this, $user, $props, $comment, $pageText, &$error ] );
 752                 if ( $error ) {
 753                         if ( !is_array( $error ) ) {
 754                                 $error = [ $error ];
 755                         }
 756                         return call_user_func_array( 'Status::newFatal', $error );
 757                 }
 758
 759                 $status = $this->getLocalFile()->upload(
 760                         $this->mTempPath,
 761                         $comment,
 762                         $pageText,
 763                         File::DELETE_SOURCE,
 764                         $props,
 765                         false,
 766                         $user,
 767                         $tags
 768                 );
 769
 770                 if ( $status->isGood() ) {
 771                         if ( $watch ) {
 772                                 WatchAction::doWatch(
 773                                         $this->getLocalFile()->getTitle(),
 774                                         $user,
 775                                         User::IGNORE_USER_RIGHTS
 776                                 );
 777                         }
 778                         // Avoid PHP 7.1 warning of passing $this by reference
 779                         $uploadBase = $this;
 780                         Hooks::run( 'UploadComplete', [ &$uploadBase ] );
 781
 782                         $this->postProcessUpload();
 783                 }
 784
 785                 return $status;
 786         }
 787
 788         /**
 789          * Perform extra steps after a successful upload.
 790          *
 791          * @since  1.25
 792          */
 793         public function postProcessUpload() {
 794         }
 795
 796         /**
 797          * Returns the title of the file to be uploaded. Sets mTitleError in case
 798          * the name was illegal.
 799          *
 800          * @return Title|null The title of the file or null in case the name was illegal
 801          */
 802         public function getTitle() {
 803                 if ( $this->mTitle !== false ) {
 804                         return $this->mTitle;
 805                 }
 806                 if ( !is_string( $this->mDesiredDestName ) ) {
 807                         $this->mTitleError = self::ILLEGAL_FILENAME;
 808                         $this->mTitle = null;
 809
 810                         return $this->mTitle;
 811                 }
 812                 /* Assume that if a user specified File:Something.jpg, this is an error
 813                  * and that the namespace prefix needs to be stripped of.
 814                  */
 815                 $title = Title::newFromText( $this->mDesiredDestName );
 816                 if ( $title && $title->getNamespace() == NS_FILE ) {
 817                         $this->mFilteredName = $title->getDBkey();
 818                 } else {
 819                         $this->mFilteredName = $this->mDesiredDestName;
 820                 }
 821
 822                 # oi_archive_name is max 255 bytes, which include a timestamp and an
 823                 # exclamation mark, so restrict file name to 240 bytes.
 824                 if ( strlen( $this->mFilteredName ) > 240 ) {
 825                         $this->mTitleError = self::FILENAME_TOO_LONG;
 826                         $this->mTitle = null;
 827
 828                         return $this->mTitle;
 829                 }
 830
 831                 /**
 832                  * Chop off any directories in the given filename. Then
 833                  * filter out illegal characters, and try to make a legible name
 834                  * out of it. We'll strip some silently that Title would die on.
 835                  */
 836                 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mFilteredName );
 837                 /* Normalize to title form before we do any further processing */
 838                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 839                 if ( is_null( $nt ) ) {
 840                         $this->mTitleError = self::ILLEGAL_FILENAME;
 841                         $this->mTitle = null;
 842
 843                         return $this->mTitle;
 844                 }
 845                 $this->mFilteredName = $nt->getDBkey();
 846
 847                 /**
 848                  * We'll want to blacklist against *any* 'extension', and use
 849                  * only the final one for the whitelist.
 850                  */
 851                 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
 852
 853                 if ( count( $ext ) ) {
 854                         $this->mFinalExtension = trim( $ext[count( $ext ) - 1] );
 855                 } else {
 856                         $this->mFinalExtension = '';
 857
 858                         # No extension, try guessing one
 859                         $magic = MimeMagic::singleton();
 860                         $mime = $magic->guessMimeType( $this->mTempPath );
 861                         if ( $mime !== 'unknown/unknown' ) {
 862                                 # Get a space separated list of extensions
 863                                 $extList = $magic->getExtensionsForType( $mime );
 864                                 if ( $extList ) {
 865                                         # Set the extension to the canonical extension
 866                                         $this->mFinalExtension = strtok( $extList, ' ' );
 867
 868                                         # Fix up the other variables
 869                                         $this->mFilteredName .= ".{$this->mFinalExtension}";
 870                                         $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 871                                         $ext = [ $this->mFinalExtension ];
 872                                 }
 873                         }
 874                 }
 875
 876                 /* Don't allow users to override the blacklist (check file extension) */
 877                 global $wgCheckFileExtensions, $wgStrictFileExtensions;
 878                 global $wgFileExtensions, $wgFileBlacklist;
 879
 880                 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
 881
 882                 if ( $this->mFinalExtension == '' ) {
 883                         $this->mTitleError = self::FILETYPE_MISSING;
 884                         $this->mTitle = null;
 885
 886                         return $this->mTitle;
 887                 } elseif ( $blackListedExtensions ||
 888                         ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
 889                                 !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) )
 890                 ) {
 891                         $this->mBlackListedExtensions = $blackListedExtensions;
 892                         $this->mTitleError = self::FILETYPE_BADTYPE;
 893                         $this->mTitle = null;
 894
 895                         return $this->mTitle;
 896                 }
 897
 898                 // Windows may be broken with special characters, see T3780
 899                 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
 900                         && !RepoGroup::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
 901                 ) {
 902                         $this->mTitleError = self::WINDOWS_NONASCII_FILENAME;
 903                         $this->mTitle = null;
 904
 905                         return $this->mTitle;
 906                 }
 907
 908                 # If there was more than one "extension", reassemble the base
 909                 # filename to prevent bogus complaints about length
 910                 if ( count( $ext ) > 1 ) {
 911                         $iterations = count( $ext ) - 1;
 912                         for ( $i = 0; $i < $iterations; $i++ ) {
 913                                 $partname .= '.' . $ext[$i];
 914                         }
 915                 }
 916
 917                 if ( strlen( $partname ) < 1 ) {
 918                         $this->mTitleError = self::MIN_LENGTH_PARTNAME;
 919                         $this->mTitle = null;
 920
 921                         return $this->mTitle;
 922                 }
 923
 924                 $this->mTitle = $nt;
 925
 926                 return $this->mTitle;
 927         }
 928
 929         /**
 930          * Return the local file and initializes if necessary.
 931          *
 932          * @return LocalFile|null
 933          */
 934         public function getLocalFile() {
 935                 if ( is_null( $this->mLocalFile ) ) {
 936                         $nt = $this->getTitle();
 937                         $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
 938                 }
 939
 940                 return $this->mLocalFile;
 941         }
 942
 943         /**
 944          * @return UploadStashFile|null
 945          */
 946         public function getStashFile() {
 947                 return $this->mStashFile;
 948         }
 949
 950         /**
 951          * Like stashFile(), but respects extensions' wishes to prevent the stashing. verifyUpload() must
 952          * be called before calling this method (unless $isPartial is true).
 953          *
 954          * Upload stash exceptions are also caught and converted to an error status.
 955          *
 956          * @since 1.28
 957          * @param User $user
 958          * @param bool $isPartial Pass `true` if this is a part of a chunked upload (not a complete file).
 959          * @return Status If successful, value is an UploadStashFile instance
 960          */
 961         public function tryStashFile( User $user, $isPartial = false ) {
 962                 if ( !$isPartial ) {
 963                         $error = $this->runUploadStashFileHook( $user );
 964                         if ( $error ) {
 965                                 return call_user_func_array( 'Status::newFatal', $error );
 966                         }
 967                 }
 968                 try {
 969                         $file = $this->doStashFile( $user );
 970                         return Status::newGood( $file );
 971                 } catch ( UploadStashException $e ) {
 972                         return Status::newFatal( 'uploadstash-exception', get_class( $e ), $e->getMessage() );
 973                 }
 974         }
 975
 976         /**
 977          * @param User $user
 978          * @return array|null Error message and parameters, null if there's no error
 979          */
 980         protected function runUploadStashFileHook( User $user ) {
 981                 $props = $this->mFileProps;
 982                 $error = null;
 983                 Hooks::run( 'UploadStashFile', [ $this, $user, $props, &$error ] );
 984                 if ( $error ) {
 985                         if ( !is_array( $error ) ) {
 986                                 $error = [ $error ];
 987                         }
 988                 }
 989                 return $error;
 990         }
 991
 992         /**
 993          * If the user does not supply all necessary information in the first upload
 994          * form submission (either by accident or by design) then we may want to
 995          * stash the file temporarily, get more information, and publish the file
 996          * later.
 997          *
 998          * This method will stash a file in a temporary directory for later
 999          * processing, and save the necessary descriptive info into the database.
1000          * This method returns the file object, which also has a 'fileKey' property
1001          * which can be passed through a form or API request to find this stashed
1002          * file again.
1003          *
1004          * @deprecated since 1.28 Use tryStashFile() instead
1005          * @param User $user
1006          * @return UploadStashFile Stashed file
1007          * @throws UploadStashBadPathException
1008          * @throws UploadStashFileException
1009          * @throws UploadStashNotLoggedInException
1010          */
1011         public function stashFile( User $user = null ) {
1012                 return $this->doStashFile( $user );
1013         }
1014
1015         /**
1016          * Implementation for stashFile() and tryStashFile().
1017          *
1018          * @param User $user
1019          * @return UploadStashFile Stashed file
1020          */
1021         protected function doStashFile( User $user = null ) {
1022                 $stash = RepoGroup::singleton()->getLocalRepo()->getUploadStash( $user );
1023                 $file = $stash->stashFile( $this->mTempPath, $this->getSourceType() );
1024                 $this->mStashFile = $file;
1025
1026                 return $file;
1027         }
1028
1029         /**
1030          * Stash a file in a temporary directory, returning a key which can be used
1031          * to find the file again. See stashFile().
1032          *
1033          * @deprecated since 1.28
1034          * @return string File key
1035          */
1036         public function stashFileGetKey() {
1037                 wfDeprecated( __METHOD__, '1.28' );
1038                 return $this->doStashFile()->getFileKey();
1039         }
1040
1041         /**
1042          * alias for stashFileGetKey, for backwards compatibility
1043          *
1044          * @deprecated since 1.28
1045          * @return string File key
1046          */
1047         public function stashSession() {
1048                 wfDeprecated( __METHOD__, '1.28' );
1049                 return $this->doStashFile()->getFileKey();
1050         }
1051
1052         /**
1053          * If we've modified the upload file we need to manually remove it
1054          * on exit to clean up.
1055          */
1056         public function cleanupTempFile() {
1057                 if ( $this->mRemoveTempFile && $this->tempFileObj ) {
1058                         // Delete when all relevant TempFSFile handles go out of scope
1059                         wfDebug( __METHOD__ . ": Marked temporary file '{$this->mTempPath}' for removal\n" );
1060                         $this->tempFileObj->autocollect();
1061                 }
1062         }
1063
1064         public function getTempPath() {
1065                 return $this->mTempPath;
1066         }
1067
1068         /**
1069          * Split a file into a base name and all dot-delimited 'extensions'
1070          * on the end. Some web server configurations will fall back to
1071          * earlier pseudo-'extensions' to determine type and execute
1072          * scripts, so the blacklist needs to check them all.
1073          *
1074          * @param string $filename
1075          * @return array
1076          */
1077         public static function splitExtensions( $filename ) {
1078                 $bits = explode( '.', $filename );
1079                 $basename = array_shift( $bits );
1080
1081                 return [ $basename, $bits ];
1082         }
1083
1084         /**
1085          * Perform case-insensitive match against a list of file extensions.
1086          * Returns true if the extension is in the list.
1087          *
1088          * @param string $ext
1089          * @param array $list
1090          * @return bool
1091          */
1092         public static function checkFileExtension( $ext, $list ) {
1093                 return in_array( strtolower( $ext ), $list );
1094         }
1095
1096         /**
1097          * Perform case-insensitive match against a list of file extensions.
1098          * Returns an array of matching extensions.
1099          *
1100          * @param array $ext
1101          * @param array $list
1102          * @return bool
1103          */
1104         public static function checkFileExtensionList( $ext, $list ) {
1105                 return array_intersect( array_map( 'strtolower', $ext ), $list );
1106         }
1107
1108         /**
1109          * Checks if the MIME type of the uploaded file matches the file extension.
1110          *
1111          * @param string $mime The MIME type of the uploaded file
1112          * @param string $extension The filename extension that the file is to be served with
1113          * @return bool
1114          */
1115         public static function verifyExtension( $mime, $extension ) {
1116                 $magic = MimeMagic::singleton();
1117
1118                 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' ) {
1119                         if ( !$magic->isRecognizableExtension( $extension ) ) {
1120                                 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
1121                                         "unrecognized extension '$extension', can't verify\n" );
1122
1123                                 return true;
1124                         } else {
1125                                 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; " .
1126                                         "recognized extension '$extension', so probably invalid file\n" );
1127
1128                                 return false;
1129                         }
1130                 }
1131
1132                 $match = $magic->isMatchingExtension( $extension, $mime );
1133
1134                 if ( $match === null ) {
1135                         if ( $magic->getTypesForExtension( $extension ) !== null ) {
1136                                 wfDebug( __METHOD__ . ": No extension known for $mime, but we know a mime for $extension\n" );
1137
1138                                 return false;
1139                         } else {
1140                                 wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
1141
1142                                 return true;
1143                         }
1144                 } elseif ( $match === true ) {
1145                         wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
1146
1147                         /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1148                         return true;
1149                 } else {
1150                         wfDebug( __METHOD__
1151                                 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1152
1153                         return false;
1154                 }
1155         }
1156
1157         /**
1158          * Heuristic for detecting files that *could* contain JavaScript instructions or
1159          * things that may look like HTML to a browser and are thus
1160          * potentially harmful. The present implementation will produce false
1161          * positives in some situations.
1162          *
1163          * @param string $file Pathname to the temporary upload file
1164          * @param string $mime The MIME type of the file
1165          * @param string $extension The extension of the file
1166          * @return bool True if the file contains something looking like embedded scripts
1167          */
1168         public static function detectScript( $file, $mime, $extension ) {
1169                 global $wgAllowTitlesInSVG;
1170
1171                 # ugly hack: for text files, always look at the entire file.
1172                 # For binary field, just check the first K.
1173
1174                 if ( strpos( $mime, 'text/' ) === 0 ) {
1175                         $chunk = file_get_contents( $file );
1176                 } else {
1177                         $fp = fopen( $file, 'rb' );
1178                         $chunk = fread( $fp, 1024 );
1179                         fclose( $fp );
1180                 }
1181
1182                 $chunk = strtolower( $chunk );
1183
1184                 if ( !$chunk ) {
1185                         return false;
1186                 }
1187
1188                 # decode from UTF-16 if needed (could be used for obfuscation).
1189                 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1190                         $enc = 'UTF-16BE';
1191                 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1192                         $enc = 'UTF-16LE';
1193                 } else {
1194                         $enc = null;
1195                 }
1196
1197                 if ( $enc ) {
1198                         $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1199                 }
1200
1201                 $chunk = trim( $chunk );
1202
1203                 /** @todo FIXME: Convert from UTF-16 if necessary! */
1204                 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
1205
1206                 # check for HTML doctype
1207                 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1208                         return true;
1209                 }
1210
1211                 // Some browsers will interpret obscure xml encodings as UTF-8, while
1212                 // PHP/expat will interpret the given encoding in the xml declaration (T49304)
1213                 if ( $extension == 'svg' || strpos( $mime, 'image/svg' ) === 0 ) {
1214                         if ( self::checkXMLEncodingMissmatch( $file ) ) {
1215                                 return true;
1216                         }
1217                 }
1218
1219                 /**
1220                  * Internet Explorer for Windows performs some really stupid file type
1221                  * autodetection which can cause it to interpret valid image files as HTML
1222                  * and potentially execute JavaScript, creating a cross-site scripting
1223                  * attack vectors.
1224                  *
1225                  * Apple's Safari browser also performs some unsafe file type autodetection
1226                  * which can cause legitimate files to be interpreted as HTML if the
1227                  * web server is not correctly configured to send the right content-type
1228                  * (or if you're really uploading plain text and octet streams!)
1229                  *
1230                  * Returns true if IE is likely to mistake the given file for HTML.
1231                  * Also returns true if Safari would mistake the given file for HTML
1232                  * when served with a generic content-type.
1233                  */
1234                 $tags = [
1235                         '<a href',
1236                         '<body',
1237                         '<head',
1238                         '<html', # also in safari
1239                         '<img',
1240                         '<pre',
1241                         '<script', # also in safari
1242                         '<table'
1243                 ];
1244
1245                 if ( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
1246                         $tags[] = '<title';
1247                 }
1248
1249                 foreach ( $tags as $tag ) {
1250                         if ( false !== strpos( $chunk, $tag ) ) {
1251                                 wfDebug( __METHOD__ . ": found something that may make it be mistaken for html: $tag\n" );
1252
1253                                 return true;
1254                         }
1255                 }
1256
1257                 /*
1258                  * look for JavaScript
1259                  */
1260
1261                 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1262                 $chunk = Sanitizer::decodeCharReferences( $chunk );
1263
1264                 # look for script-types
1265                 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1266                         wfDebug( __METHOD__ . ": found script types\n" );
1267
1268                         return true;
1269                 }
1270
1271                 # look for html-style script-urls
1272                 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1273                         wfDebug( __METHOD__ . ": found html-style script urls\n" );
1274
1275                         return true;
1276                 }
1277
1278                 # look for css-style script-urls
1279                 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1280                         wfDebug( __METHOD__ . ": found css-style script urls\n" );
1281
1282                         return true;
1283                 }
1284
1285                 wfDebug( __METHOD__ . ": no scripts found\n" );
1286
1287                 return false;
1288         }
1289
1290         /**
1291          * Check a whitelist of xml encodings that are known not to be interpreted differently
1292          * by the server's xml parser (expat) and some common browsers.
1293          *
1294          * @param string $file Pathname to the temporary upload file
1295          * @return bool True if the file contains an encoding that could be misinterpreted
1296          */
1297         public static function checkXMLEncodingMissmatch( $file ) {
1298                 global $wgSVGMetadataCutoff;
1299                 $contents = file_get_contents( $file, false, null, -1, $wgSVGMetadataCutoff );
1300                 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1301
1302                 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1303                         if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1304                                 && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1305                         ) {
1306                                 wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1307
1308                                 return true;
1309                         }
1310                 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1311                         // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1312                         // bytes. There shouldn't be a legitimate reason for this to happen.
1313                         wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1314
1315                         return true;
1316                 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1317                         // EBCDIC encoded XML
1318                         wfDebug( __METHOD__ . ": EBCDIC Encoded XML\n" );
1319
1320                         return true;
1321                 }
1322
1323                 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1324                 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1325                 $attemptEncodings = [ 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' ];
1326                 foreach ( $attemptEncodings as $encoding ) {
1327                         MediaWiki\suppressWarnings();
1328                         $str = iconv( $encoding, 'UTF-8', $contents );
1329                         MediaWiki\restoreWarnings();
1330                         if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1331                                 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1332                                         && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1333                                 ) {
1334                                         wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1335
1336                                         return true;
1337                                 }
1338                         } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1339                                 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1340                                 // bytes. There shouldn't be a legitimate reason for this to happen.
1341                                 wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1342
1343                                 return true;
1344                         }
1345                 }
1346
1347                 return false;
1348         }
1349
1350         /**
1351          * @param string $filename
1352          * @param bool $partial
1353          * @return mixed False of the file is verified (does not contain scripts), array otherwise.
1354          */
1355         protected function detectScriptInSvg( $filename, $partial ) {
1356                 $this->mSVGNSError = false;
1357                 $check = new XmlTypeCheck(
1358                         $filename,
1359                         [ $this, 'checkSvgScriptCallback' ],
1360                         true,
1361                         [
1362                                 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback',
1363                                 'external_dtd_handler' => 'UploadBase::checkSvgExternalDTD',
1364                         ]
1365                 );
1366                 if ( $check->wellFormed !== true ) {
1367                         // Invalid xml (T60553)
1368                         // But only when non-partial (T67724)
1369                         return $partial ? false : [ 'uploadinvalidxml' ];
1370                 } elseif ( $check->filterMatch ) {
1371                         if ( $this->mSVGNSError ) {
1372                                 return [ 'uploadscriptednamespace', $this->mSVGNSError ];
1373                         }
1374
1375                         return $check->filterMatchType;
1376                 }
1377
1378                 return false;
1379         }
1380
1381         /**
1382          * Callback to filter SVG Processing Instructions.
1383          * @param string $target Processing instruction name
1384          * @param string $data Processing instruction attribute and value
1385          * @return bool (true if the filter identified something bad)
1386          */
1387         public static function checkSvgPICallback( $target, $data ) {
1388                 // Don't allow external stylesheets (T59550)
1389                 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1390                         return [ 'upload-scripted-pi-callback' ];
1391                 }
1392
1393                 return false;
1394         }
1395
1396         /**
1397          * Verify that DTD urls referenced are only the standard dtds
1398          *
1399          * Browsers seem to ignore external dtds. However just to be on the
1400          * safe side, only allow dtds from the svg standard.
1401          *
1402          * @param string $type PUBLIC or SYSTEM
1403          * @param string $publicId The well-known public identifier for the dtd
1404          * @param string $systemId The url for the external dtd
1405          */
1406         public static function checkSvgExternalDTD( $type, $publicId, $systemId ) {
1407                 // This doesn't include the XHTML+MathML+SVG doctype since we don't
1408                 // allow XHTML anyways.
1409                 $allowedDTDs = [
1410                         'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd',
1411                         'http://www.w3.org/TR/2001/REC-SVG-20010904/DTD/svg10.dtd',
1412                         'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-basic.dtd',
1413                         'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-tiny.dtd',
1414                         // https://phabricator.wikimedia.org/T168856
1415                         'http://www.w3.org/TR/2001/PR-SVG-20010719/DTD/svg10.dtd',
1416                 ];
1417                 if ( $type !== 'PUBLIC'
1418                         || !in_array( $systemId, $allowedDTDs )
1419                         || strpos( $publicId, "-//W3C//" ) !== 0
1420                 ) {
1421                         return [ 'upload-scripted-dtd' ];
1422                 }
1423                 return false;
1424         }
1425
1426         /**
1427          * @todo Replace this with a whitelist filter!
1428          * @param string $element
1429          * @param array $attribs
1430          * @return bool
1431          */
1432         public function checkSvgScriptCallback( $element, $attribs, $data = null ) {
1433                 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1434
1435                 // We specifically don't include:
1436                 // http://www.w3.org/1999/xhtml (T62771)
1437                 static $validNamespaces = [
1438                         '',
1439                         'adobe:ns:meta/',
1440                         'http://creativecommons.org/ns#',
1441                         'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1442                         'http://ns.adobe.com/adobeillustrator/10.0/',
1443                         'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1444                         'http://ns.adobe.com/extensibility/1.0/',
1445                         'http://ns.adobe.com/flows/1.0/',
1446                         'http://ns.adobe.com/illustrator/1.0/',
1447                         'http://ns.adobe.com/imagereplacement/1.0/',
1448                         'http://ns.adobe.com/pdf/1.3/',
1449                         'http://ns.adobe.com/photoshop/1.0/',
1450                         'http://ns.adobe.com/saveforweb/1.0/',
1451                         'http://ns.adobe.com/variables/1.0/',
1452                         'http://ns.adobe.com/xap/1.0/',
1453                         'http://ns.adobe.com/xap/1.0/g/',
1454                         'http://ns.adobe.com/xap/1.0/g/img/',
1455                         'http://ns.adobe.com/xap/1.0/mm/',
1456                         'http://ns.adobe.com/xap/1.0/rights/',
1457                         'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1458                         'http://ns.adobe.com/xap/1.0/stype/font#',
1459                         'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1460                         'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1461                         'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1462                         'http://ns.adobe.com/xap/1.0/t/pg/',
1463                         'http://purl.org/dc/elements/1.1/',
1464                         'http://purl.org/dc/elements/1.1',
1465                         'http://schemas.microsoft.com/visio/2003/svgextensions/',
1466                         'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1467                         'http://taptrix.com/inkpad/svg_extensions',
1468                         'http://web.resource.org/cc/',
1469                         'http://www.freesoftware.fsf.org/bkchem/cdml',
1470                         'http://www.inkscape.org/namespaces/inkscape',
1471                         'http://www.opengis.net/gml',
1472                         'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1473                         'http://www.w3.org/2000/svg',
1474                         'http://www.w3.org/tr/rec-rdf-syntax/',
1475                         'http://www.w3.org/2000/01/rdf-schema#',
1476                 ];
1477
1478                 // Inkscape mangles namespace definitions created by Adobe Illustrator.
1479                 // This is nasty but harmless. (T144827)
1480                 $isBuggyInkscape = preg_match( '/^&(#38;)*ns_[a-z_]+;$/', $namespace );
1481
1482                 if ( !( $isBuggyInkscape || in_array( $namespace, $validNamespaces ) ) ) {
1483                         wfDebug( __METHOD__ . ": Non-svg namespace '$namespace' in uploaded file.\n" );
1484                         /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1485                         $this->mSVGNSError = $namespace;
1486
1487                         return true;
1488                 }
1489
1490                 /*
1491                  * check for elements that can contain javascript
1492                  */
1493                 if ( $strippedElement == 'script' ) {
1494                         wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
1495
1496                         return [ 'uploaded-script-svg', $strippedElement ];
1497                 }
1498
1499                 # e.g., <svg xmlns="http://www.w3.org/2000/svg">
1500                 #  <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1501                 if ( $strippedElement == 'handler' ) {
1502                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1503
1504                         return [ 'uploaded-script-svg', $strippedElement ];
1505                 }
1506
1507                 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1508                 if ( $strippedElement == 'stylesheet' ) {
1509                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1510
1511                         return [ 'uploaded-script-svg', $strippedElement ];
1512                 }
1513
1514                 # Block iframes, in case they pass the namespace check
1515                 if ( $strippedElement == 'iframe' ) {
1516                         wfDebug( __METHOD__ . ": iframe in uploaded file.\n" );
1517
1518                         return [ 'uploaded-script-svg', $strippedElement ];
1519                 }
1520
1521                 # Check <style> css
1522                 if ( $strippedElement == 'style'
1523                         && self::checkCssFragment( Sanitizer::normalizeCss( $data ) )
1524                 ) {
1525                         wfDebug( __METHOD__ . ": hostile css in style element.\n" );
1526                         return [ 'uploaded-hostile-svg' ];
1527                 }
1528
1529                 foreach ( $attribs as $attrib => $value ) {
1530                         $stripped = $this->stripXmlNamespace( $attrib );
1531                         $value = strtolower( $value );
1532
1533                         if ( substr( $stripped, 0, 2 ) == 'on' ) {
1534                                 wfDebug( __METHOD__
1535                                         . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1536
1537                                 return [ 'uploaded-event-handler-on-svg', $attrib, $value ];
1538                         }
1539
1540                         # Do not allow relative links, or unsafe url schemas.
1541                         # For <a> tags, only data:, http: and https: and same-document
1542                         # fragment links are allowed. For all other tags, only data:
1543                         # and fragment are allowed.
1544                         if ( $stripped == 'href'
1545                                 && $value !== ''
1546                                 && strpos( $value, 'data:' ) !== 0
1547                                 && strpos( $value, '#' ) !== 0
1548                         ) {
1549                                 if ( !( $strippedElement === 'a'
1550                                         && preg_match( '!^https?://!i', $value ) )
1551                                 ) {
1552                                         wfDebug( __METHOD__ . ": Found href attribute <$strippedElement "
1553                                                 . "'$attrib'='$value' in uploaded file.\n" );
1554
1555                                         return [ 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value ];
1556                                 }
1557                         }
1558
1559                         # only allow data: targets that should be safe. This prevents vectors like,
1560                         # image/svg, text/xml, application/xml, and text/html, which can contain scripts
1561                         if ( $stripped == 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) {
1562                                 // rfc2397 parameters. This is only slightly slower than (;[\w;]+)*.
1563                                 // @codingStandardsIgnoreStart Generic.Files.LineLength
1564                                 $parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?';
1565                                 // @codingStandardsIgnoreEnd
1566
1567                                 if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|png)$parameters,!i", $value ) ) {
1568                                         wfDebug( __METHOD__ . ": Found href to unwhitelisted data: uri "
1569                                                 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1570                                         return [ 'uploaded-href-unsafe-target-svg', $strippedElement, $attrib, $value ];
1571                                 }
1572                         }
1573
1574                         # Change href with animate from (http://html5sec.org/#137).
1575                         if ( $stripped === 'attributename'
1576                                 && $strippedElement === 'animate'
1577                                 && $this->stripXmlNamespace( $value ) == 'href'
1578                         ) {
1579                                 wfDebug( __METHOD__ . ": Found animate that might be changing href using from "
1580                                         . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1581
1582                                 return [ 'uploaded-animate-svg', $strippedElement, $attrib, $value ];
1583                         }
1584
1585                         # use set/animate to add event-handler attribute to parent
1586                         if ( ( $strippedElement == 'set' || $strippedElement == 'animate' )
1587                                 && $stripped == 'attributename'
1588                                 && substr( $value, 0, 2 ) == 'on'
1589                         ) {
1590                                 wfDebug( __METHOD__ . ": Found svg setting event-handler attribute with "
1591                                         . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1592
1593                                 return [ 'uploaded-setting-event-handler-svg', $strippedElement, $stripped, $value ];
1594                         }
1595
1596                         # use set to add href attribute to parent element
1597                         if ( $strippedElement == 'set'
1598                                 && $stripped == 'attributename'
1599                                 && strpos( $value, 'href' ) !== false
1600                         ) {
1601                                 wfDebug( __METHOD__ . ": Found svg setting href attribute '$value' in uploaded file.\n" );
1602
1603                                 return [ 'uploaded-setting-href-svg' ];
1604                         }
1605
1606                         # use set to add a remote / data / script target to an element
1607                         if ( $strippedElement == 'set'
1608                                 && $stripped == 'to'
1609                                 && preg_match( '!(http|https|data|script):!sim', $value )
1610                         ) {
1611                                 wfDebug( __METHOD__ . ": Found svg setting attribute to '$value' in uploaded file.\n" );
1612
1613                                 return [ 'uploaded-wrong-setting-svg', $value ];
1614                         }
1615
1616                         # use handler attribute with remote / data / script
1617                         if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1618                                 wfDebug( __METHOD__ . ": Found svg setting handler with remote/data/script "
1619                                         . "'$attrib'='$value' in uploaded file.\n" );
1620
1621                                 return [ 'uploaded-setting-handler-svg', $attrib, $value ];
1622                         }
1623
1624                         # use CSS styles to bring in remote code
1625                         if ( $stripped == 'style'
1626                                 && self::checkCssFragment( Sanitizer::normalizeCss( $value ) )
1627                         ) {
1628                                 wfDebug( __METHOD__ . ": Found svg setting a style with "
1629                                         . "remote url '$attrib'='$value' in uploaded file.\n" );
1630                                 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1631                         }
1632
1633                         # Several attributes can include css, css character escaping isn't allowed
1634                         $cssAttrs = [ 'font', 'clip-path', 'fill', 'filter', 'marker',
1635                                 'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke' ];
1636                         if ( in_array( $stripped, $cssAttrs )
1637                                 && self::checkCssFragment( $value )
1638                         ) {
1639                                 wfDebug( __METHOD__ . ": Found svg setting a style with "
1640                                         . "remote url '$attrib'='$value' in uploaded file.\n" );
1641                                 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1642                         }
1643
1644                         # image filters can pull in url, which could be svg that executes scripts
1645                         if ( $strippedElement == 'image'
1646                                 && $stripped == 'filter'
1647                                 && preg_match( '!url\s*\(!sim', $value )
1648                         ) {
1649                                 wfDebug( __METHOD__ . ": Found image filter with url: "
1650                                         . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1651
1652                                 return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ];
1653                         }
1654                 }
1655
1656                 return false; // No scripts detected
1657         }
1658
1659         /**
1660          * Check a block of CSS or CSS fragment for anything that looks like
1661          * it is bringing in remote code.
1662          * @param string $value a string of CSS
1663          * @param bool $propOnly only check css properties (start regex with :)
1664          * @return bool true if the CSS contains an illegal string, false if otherwise
1665          */
1666         private static function checkCssFragment( $value ) {
1667                 # Forbid external stylesheets, for both reliability and to protect viewer's privacy
1668                 if ( stripos( $value, '@import' ) !== false ) {
1669                         return true;
1670                 }
1671
1672                 # We allow @font-face to embed fonts with data: urls, so we snip the string
1673                 # 'url' out so this case won't match when we check for urls below
1674                 $pattern = '!(@font-face\s*{[^}]*src:)url(\("data:;base64,)!im';
1675                 $value = preg_replace( $pattern, '$1$2', $value );
1676
1677                 # Check for remote and executable CSS. Unlike in Sanitizer::checkCss, the CSS
1678                 # properties filter and accelerator don't seem to be useful for xss in SVG files.
1679                 # Expression and -o-link don't seem to work either, but filtering them here in case.
1680                 # Additionally, we catch remote urls like url("http:..., url('http:..., url(http:...,
1681                 # but not local ones such as url("#..., url('#..., url(#....
1682                 if ( preg_match( '!expression
1683                                 | -o-link\s*:
1684                                 | -o-link-source\s*:
1685                                 | -o-replace\s*:!imx', $value ) ) {
1686                         return true;
1687                 }
1688
1689                 if ( preg_match_all(
1690                                 "!(\s*(url|image|image-set)\s*\(\s*[\"']?\s*[^#]+.*?\))!sim",
1691                                 $value,
1692                                 $matches
1693                         ) !== 0
1694                 ) {
1695                         # TODO: redo this in one regex. Until then, url("#whatever") matches the first
1696                         foreach ( $matches[1] as $match ) {
1697                                 if ( !preg_match( "!\s*(url|image|image-set)\s*\(\s*(#|'#|\"#)!im", $match ) ) {
1698                                         return true;
1699                                 }
1700                         }
1701                 }
1702
1703                 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
1704                         return true;
1705                 }
1706
1707                 return false;
1708         }
1709
1710         /**
1711          * Divide the element name passed by the xml parser to the callback into URI and prifix.
1712          * @param string $element
1713          * @return array Containing the namespace URI and prefix
1714          */
1715         private static function splitXmlNamespace( $element ) {
1716                 // 'http://www.w3.org/2000/svg:script' -> [ 'http://www.w3.org/2000/svg', 'script' ]
1717                 $parts = explode( ':', strtolower( $element ) );
1718                 $name = array_pop( $parts );
1719                 $ns = implode( ':', $parts );
1720
1721                 return [ $ns, $name ];
1722         }
1723
1724         /**
1725          * @param string $name
1726          * @return string
1727          */
1728         private function stripXmlNamespace( $name ) {
1729                 // 'http://www.w3.org/2000/svg:script' -> 'script'
1730                 $parts = explode( ':', strtolower( $name ) );
1731
1732                 return array_pop( $parts );
1733         }
1734
1735         /**
1736          * Generic wrapper function for a virus scanner program.
1737          * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1738          * $wgAntivirusRequired may be used to deny upload if the scan fails.
1739          *
1740          * @param string $file Pathname to the temporary upload file
1741          * @return mixed False if not virus is found, null if the scan fails or is disabled,
1742          *   or a string containing feedback from the virus scanner if a virus was found.
1743          *   If textual feedback is missing but a virus was found, this function returns true.
1744          */
1745         public static function detectVirus( $file ) {
1746                 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1747
1748                 if ( !$wgAntivirus ) {
1749                         wfDebug( __METHOD__ . ": virus scanner disabled\n" );
1750
1751                         return null;
1752                 }
1753
1754                 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1755                         wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
1756                         $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1757                                 [ 'virus-badscanner', $wgAntivirus ] );
1758
1759                         return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1760                 }
1761
1762                 # look up scanner configuration
1763                 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1764                 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1765                 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
1766                         $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
1767
1768                 if ( strpos( $command, "%f" ) === false ) {
1769                         # simple pattern: append file to scan
1770                         $command .= " " . wfEscapeShellArg( $file );
1771                 } else {
1772                         # complex pattern: replace "%f" with file to scan
1773                         $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
1774                 }
1775
1776                 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
1777
1778                 # execute virus scanner
1779                 $exitCode = false;
1780
1781                 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1782                 #      that does not seem to be worth the pain.
1783                 #      Ask me (Duesentrieb) about it if it's ever needed.
1784                 $output = wfShellExecWithStderr( $command, $exitCode );
1785
1786                 # map exit code to AV_xxx constants.
1787                 $mappedCode = $exitCode;
1788                 if ( $exitCodeMap ) {
1789                         if ( isset( $exitCodeMap[$exitCode] ) ) {
1790                                 $mappedCode = $exitCodeMap[$exitCode];
1791                         } elseif ( isset( $exitCodeMap["*"] ) ) {
1792                                 $mappedCode = $exitCodeMap["*"];
1793                         }
1794                 }
1795
1796                 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1797                  * so we need the strict equalities === and thus can't use a switch here
1798                  */
1799                 if ( $mappedCode === AV_SCAN_FAILED ) {
1800                         # scan failed (code was mapped to false by $exitCodeMap)
1801                         wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
1802
1803                         $output = $wgAntivirusRequired
1804                                 ? wfMessage( 'virus-scanfailed', [ $exitCode ] )->text()
1805                                 : null;
1806                 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
1807                         # scan failed because filetype is unknown (probably imune)
1808                         wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
1809                         $output = null;
1810                 } elseif ( $mappedCode === AV_NO_VIRUS ) {
1811                         # no virus found
1812                         wfDebug( __METHOD__ . ": file passed virus scan.\n" );
1813                         $output = false;
1814                 } else {
1815                         $output = trim( $output );
1816
1817                         if ( !$output ) {
1818                                 $output = true; # if there's no output, return true
1819                         } elseif ( $msgPattern ) {
1820                                 $groups = [];
1821                                 if ( preg_match( $msgPattern, $output, $groups ) ) {
1822                                         if ( $groups[1] ) {
1823                                                 $output = $groups[1];
1824                                         }
1825                                 }
1826                         }
1827
1828                         wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
1829                 }
1830
1831                 return $output;
1832         }
1833
1834         /**
1835          * Check if there's an overwrite conflict and, if so, if restrictions
1836          * forbid this user from performing the upload.
1837          *
1838          * @param User $user
1839          *
1840          * @return mixed True on success, array on failure
1841          */
1842         private function checkOverwrite( $user ) {
1843                 // First check whether the local file can be overwritten
1844                 $file = $this->getLocalFile();
1845                 $file->load( File::READ_LATEST );
1846                 if ( $file->exists() ) {
1847                         if ( !self::userCanReUpload( $user, $file ) ) {
1848                                 return [ 'fileexists-forbidden', $file->getName() ];
1849                         } else {
1850                                 return true;
1851                         }
1852                 }
1853
1854                 /* Check shared conflicts: if the local file does not exist, but
1855                  * wfFindFile finds a file, it exists in a shared repository.
1856                  */
1857                 $file = wfFindFile( $this->getTitle(), [ 'latest' => true ] );
1858                 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1859                         return [ 'fileexists-shared-forbidden', $file->getName() ];
1860                 }
1861
1862                 return true;
1863         }
1864
1865         /**
1866          * Check if a user is the last uploader
1867          *
1868          * @param User $user
1869          * @param File $img
1870          * @return bool
1871          */
1872         public static function userCanReUpload( User $user, File $img ) {
1873                 if ( $user->isAllowed( 'reupload' ) ) {
1874                         return true; // non-conditional
1875                 } elseif ( !$user->isAllowed( 'reupload-own' ) ) {
1876                         return false;
1877                 }
1878
1879                 if ( !( $img instanceof LocalFile ) ) {
1880                         return false;
1881                 }
1882
1883                 $img->load();
1884
1885                 return $user->getId() == $img->getUser( 'id' );
1886         }
1887
1888         /**
1889          * Helper function that does various existence checks for a file.
1890          * The following checks are performed:
1891          * - The file exists
1892          * - Article with the same name as the file exists
1893          * - File exists with normalized extension
1894          * - The file looks like a thumbnail and the original exists
1895          *
1896          * @param File $file The File object to check
1897          * @return mixed False if the file does not exists, else an array
1898          */
1899         public static function getExistsWarning( $file ) {
1900                 if ( $file->exists() ) {
1901                         return [ 'warning' => 'exists', 'file' => $file ];
1902                 }
1903
1904                 if ( $file->getTitle()->getArticleID() ) {
1905                         return [ 'warning' => 'page-exists', 'file' => $file ];
1906                 }
1907
1908                 if ( strpos( $file->getName(), '.' ) == false ) {
1909                         $partname = $file->getName();
1910                         $extension = '';
1911                 } else {
1912                         $n = strrpos( $file->getName(), '.' );
1913                         $extension = substr( $file->getName(), $n + 1 );
1914                         $partname = substr( $file->getName(), 0, $n );
1915                 }
1916                 $normalizedExtension = File::normalizeExtension( $extension );
1917
1918                 if ( $normalizedExtension != $extension ) {
1919                         // We're not using the normalized form of the extension.
1920                         // Normal form is lowercase, using most common of alternate
1921                         // extensions (eg 'jpg' rather than 'JPEG').
1922
1923                         // Check for another file using the normalized form...
1924                         $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
1925                         $file_lc = wfLocalFile( $nt_lc );
1926
1927                         if ( $file_lc->exists() ) {
1928                                 return [
1929                                         'warning' => 'exists-normalized',
1930                                         'file' => $file,
1931                                         'normalizedFile' => $file_lc
1932                                 ];
1933                         }
1934                 }
1935
1936                 // Check for files with the same name but a different extension
1937                 $similarFiles = RepoGroup::singleton()->getLocalRepo()->findFilesByPrefix(
1938                         "{$partname}.", 1 );
1939                 if ( count( $similarFiles ) ) {
1940                         return [
1941                                 'warning' => 'exists-normalized',
1942                                 'file' => $file,
1943                                 'normalizedFile' => $similarFiles[0],
1944                         ];
1945                 }
1946
1947                 if ( self::isThumbName( $file->getName() ) ) {
1948                         # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1949                         $nt_thb = Title::newFromText(
1950                                 substr( $partname, strpos( $partname, '-' ) + 1 ) . '.' . $extension,
1951                                 NS_FILE
1952                         );
1953                         $file_thb = wfLocalFile( $nt_thb );
1954                         if ( $file_thb->exists() ) {
1955                                 return [
1956                                         'warning' => 'thumb',
1957                                         'file' => $file,
1958                                         'thumbFile' => $file_thb
1959                                 ];
1960                         } else {
1961                                 // File does not exist, but we just don't like the name
1962                                 return [
1963                                         'warning' => 'thumb-name',
1964                                         'file' => $file,
1965                                         'thumbFile' => $file_thb
1966                                 ];
1967                         }
1968                 }
1969
1970                 foreach ( self::getFilenamePrefixBlacklist() as $prefix ) {
1971                         if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1972                                 return [
1973                                         'warning' => 'bad-prefix',
1974                                         'file' => $file,
1975                                         'prefix' => $prefix
1976                                 ];
1977                         }
1978                 }
1979
1980                 return false;
1981         }
1982
1983         /**
1984          * Helper function that checks whether the filename looks like a thumbnail
1985          * @param string $filename
1986          * @return bool
1987          */
1988         public static function isThumbName( $filename ) {
1989                 $n = strrpos( $filename, '.' );
1990                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
1991
1992                 return (
1993                         substr( $partname, 3, 3 ) == 'px-' ||
1994                         substr( $partname, 2, 3 ) == 'px-'
1995                 ) &&
1996                 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
1997         }
1998
1999         /**
2000          * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
2001          *
2002          * @return array List of prefixes
2003          */
2004         public static function getFilenamePrefixBlacklist() {
2005                 $blacklist = [];
2006                 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
2007                 if ( !$message->isDisabled() ) {
2008                         $lines = explode( "\n", $message->plain() );
2009                         foreach ( $lines as $line ) {
2010                                 // Remove comment lines
2011                                 $comment = substr( trim( $line ), 0, 1 );
2012                                 if ( $comment == '#' || $comment == '' ) {
2013                                         continue;
2014                                 }
2015                                 // Remove additional comments after a prefix
2016                                 $comment = strpos( $line, '#' );
2017                                 if ( $comment > 0 ) {
2018                                         $line = substr( $line, 0, $comment - 1 );
2019                                 }
2020                                 $blacklist[] = trim( $line );
2021                         }
2022                 }
2023
2024                 return $blacklist;
2025         }
2026
2027         /**
2028          * Gets image info about the file just uploaded.
2029          *
2030          * Also has the effect of setting metadata to be an 'indexed tag name' in
2031          * returned API result if 'metadata' was requested. Oddly, we have to pass
2032          * the "result" object down just so it can do that with the appropriate
2033          * format, presumably.
2034          *
2035          * @param ApiResult $result
2036          * @return array Image info
2037          */
2038         public function getImageInfo( $result ) {
2039                 $localFile = $this->getLocalFile();
2040                 $stashFile = $this->getStashFile();
2041                 // Calling a different API module depending on whether the file was stashed is less than optimal.
2042                 // In fact, calling API modules here at all is less than optimal. Maybe it should be refactored.
2043                 if ( $stashFile ) {
2044                         $imParam = ApiQueryStashImageInfo::getPropertyNames();
2045                         $info = ApiQueryStashImageInfo::getInfo( $stashFile, array_flip( $imParam ), $result );
2046                 } else {
2047                         $imParam = ApiQueryImageInfo::getPropertyNames();
2048                         $info = ApiQueryImageInfo::getInfo( $localFile, array_flip( $imParam ), $result );
2049                 }
2050
2051                 return $info;
2052         }
2053
2054         /**
2055          * @param array $error
2056          * @return Status
2057          */
2058         public function convertVerifyErrorToStatus( $error ) {
2059                 $code = $error['status'];
2060                 unset( $code['status'] );
2061
2062                 return Status::newFatal( $this->getVerificationErrorCode( $code ), $error );
2063         }
2064
2065         /**
2066          * Get the MediaWiki maximum uploaded file size for given type of upload, based on
2067          * $wgMaxUploadSize.
2068          *
2069          * @param null|string $forType
2070          * @return int
2071          */
2072         public static function getMaxUploadSize( $forType = null ) {
2073                 global $wgMaxUploadSize;
2074
2075                 if ( is_array( $wgMaxUploadSize ) ) {
2076                         if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
2077                                 return $wgMaxUploadSize[$forType];
2078                         } else {
2079                                 return $wgMaxUploadSize['*'];
2080                         }
2081                 } else {
2082                         return intval( $wgMaxUploadSize );
2083                 }
2084         }
2085
2086         /**
2087          * Get the PHP maximum uploaded file size, based on ini settings. If there is no limit or the
2088          * limit can't be guessed, returns a very large number (PHP_INT_MAX).
2089          *
2090          * @since 1.27
2091          * @return int
2092          */
2093         public static function getMaxPhpUploadSize() {
2094                 $phpMaxFileSize = wfShorthandToInteger(
2095                         ini_get( 'upload_max_filesize' ) ?: ini_get( 'hhvm.server.upload.upload_max_file_size' ),
2096                         PHP_INT_MAX
2097                 );
2098                 $phpMaxPostSize = wfShorthandToInteger(
2099                         ini_get( 'post_max_size' ) ?: ini_get( 'hhvm.server.max_post_size' ),
2100                         PHP_INT_MAX
2101                 ) ?: PHP_INT_MAX;
2102                 return min( $phpMaxFileSize, $phpMaxPostSize );
2103         }
2104
2105         /**
2106          * Get the current status of a chunked upload (used for polling)
2107          *
2108          * The value will be read from cache.
2109          *
2110          * @param User $user
2111          * @param string $statusKey
2112          * @return Status[]|bool
2113          */
2114         public static function getSessionStatus( User $user, $statusKey ) {
2115                 $cache = MediaWikiServices::getInstance()->getMainObjectStash();
2116                 $key = $cache->makeKey( 'uploadstatus', $user->getId() ?: md5( $user->getName() ), $statusKey );
2117
2118                 return $cache->get( $key );
2119         }
2120
2121         /**
2122          * Set the current status of a chunked upload (used for polling)
2123          *
2124          * The value will be set in cache for 1 day
2125          *
2126          * @param User $user
2127          * @param string $statusKey
2128          * @param array|bool $value
2129          * @return void
2130          */
2131         public static function setSessionStatus( User $user, $statusKey, $value ) {
2132                 $cache = MediaWikiServices::getInstance()->getMainObjectStash();
2133                 $key = $cache->makeKey( 'uploadstatus', $user->getId() ?: md5( $user->getName() ), $statusKey );
2134
2135                 if ( $value === false ) {
2136                         $cache->delete( $key );
2137                 } else {
2138                         $cache->set( $key, $value, $cache::TTL_DAY );
2139                 }
2140         }
2141 }