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\Shell\Shell;
  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         /** @var string|null */
  45         protected $mDesiredDestName;
  46         /** @var string|null */
  47         protected $mDestName;
  48         /** @var string|null */
  49         protected $mRemoveTempFile;
  50         /** @var string|null */
  51         protected $mSourceType;
  52         /** @var Title|bool */
  53         protected $mTitle = false;
  54         /** @var int */
  55         protected $mTitleError = 0;
  56         /** @var string|null */
  57         protected $mFilteredName;
  58         /** @var string|null */
  59         protected $mFinalExtension;
  60         /** @var LocalFile */
  61         protected $mLocalFile;
  62         /** @var UploadStashFile */
  63         protected $mStashFile;
  64         /** @var int|null */
  65         protected $mFileSize;
  66         /** @var array|null */
  67         protected $mFileProps;
  68         /** @var string[] */
  69         protected $mBlackListedExtensions;
  70         /** @var bool|null */
  71         protected $mJavaDetected;
  72         /** @var string|null */
  73         protected $mSVGNSError;
  74
  75         protected static $safeXmlEncodings = [
  76                 'UTF-8',
  77                 'ISO-8859-1',
  78                 'ISO-8859-2',
  79                 'UTF-16',
  80                 'UTF-32',
  81                 'WINDOWS-1250',
  82                 'WINDOWS-1251',
  83                 'WINDOWS-1252',
  84                 'WINDOWS-1253',
  85                 'WINDOWS-1254',
  86                 'WINDOWS-1255',
  87                 'WINDOWS-1256',
  88                 'WINDOWS-1257',
  89                 'WINDOWS-1258',
  90         ];
  91
  92         const SUCCESS = 0;
  93         const OK = 0;
  94         const EMPTY_FILE = 3;
  95         const MIN_LENGTH_PARTNAME = 4;
  96         const ILLEGAL_FILENAME = 5;
  97         const OVERWRITE_EXISTING_FILE = 7; # Not used anymore; handled by verifyTitlePermissions()
  98         const FILETYPE_MISSING = 8;
  99         const FILETYPE_BADTYPE = 9;
 100         const VERIFICATION_ERROR = 10;
 101         const HOOK_ABORTED = 11;
 102         const FILE_TOO_LARGE = 12;
 103         const WINDOWS_NONASCII_FILENAME = 13;
 104         const FILENAME_TOO_LONG = 14;
 105
 106         /**
 107          * @param int $error
 108          * @return string
 109          */
 110         public function getVerificationErrorCode( $error ) {
 111                 $code_to_status = [
 112                         self::EMPTY_FILE => 'empty-file',
 113                         self::FILE_TOO_LARGE => 'file-too-large',
 114                         self::FILETYPE_MISSING => 'filetype-missing',
 115                         self::FILETYPE_BADTYPE => 'filetype-banned',
 116                         self::MIN_LENGTH_PARTNAME => 'filename-tooshort',
 117                         self::ILLEGAL_FILENAME => 'illegal-filename',
 118                         self::OVERWRITE_EXISTING_FILE => 'overwrite',
 119                         self::VERIFICATION_ERROR => 'verification-error',
 120                         self::HOOK_ABORTED => 'hookaborted',
 121                         self::WINDOWS_NONASCII_FILENAME => 'windows-nonascii-filename',
 122                         self::FILENAME_TOO_LONG => 'filename-toolong',
 123                 ];
 124                 return $code_to_status[$error] ?? 'unknown-error';
 125         }
 126
 127         /**
 128          * Returns true if uploads are enabled.
 129          * Can be override by subclasses.
 130          * @return bool
 131          */
 132         public static function isEnabled() {
 133                 global $wgEnableUploads;
 134
 135                 if ( !$wgEnableUploads ) {
 136                         return false;
 137                 }
 138
 139                 # Check php's file_uploads setting
 140                 return wfIsHHVM() || wfIniGetBool( 'file_uploads' );
 141         }
 142
 143         /**
 144          * Returns true if the user can use this upload module or else a string
 145          * identifying the missing permission.
 146          * Can be overridden by subclasses.
 147          *
 148          * @param User $user
 149          * @return bool|string
 150          */
 151         public static function isAllowed( $user ) {
 152                 foreach ( [ 'upload', 'edit' ] as $permission ) {
 153                         if ( !$user->isAllowed( $permission ) ) {
 154                                 return $permission;
 155                         }
 156                 }
 157
 158                 return true;
 159         }
 160
 161         /**
 162          * Returns true if the user has surpassed the upload rate limit, false otherwise.
 163          *
 164          * @param User $user
 165          * @return bool
 166          */
 167         public static function isThrottled( $user ) {
 168                 return $user->pingLimiter( 'upload' );
 169         }
 170
 171         // Upload handlers. Should probably just be a global.
 172         private static $uploadHandlers = [ 'Stash', 'File', 'Url' ];
 173
 174         /**
 175          * Create a form of UploadBase depending on wpSourceType and initializes it
 176          *
 177          * @param WebRequest &$request
 178          * @param string|null $type
 179          * @return null|UploadBase
 180          */
 181         public static function createFromRequest( &$request, $type = null ) {
 182                 $type = $type ?: $request->getVal( 'wpSourceType', 'File' );
 183
 184                 if ( !$type ) {
 185                         return null;
 186                 }
 187
 188                 // Get the upload class
 189                 $type = ucfirst( $type );
 190
 191                 // Give hooks the chance to handle this request
 192                 $className = null;
 193                 Hooks::run( 'UploadCreateFromRequest', [ $type, &$className ] );
 194                 if ( is_null( $className ) ) {
 195                         $className = 'UploadFrom' . $type;
 196                         wfDebug( __METHOD__ . ": class name: $className\n" );
 197                         if ( !in_array( $type, self::$uploadHandlers ) ) {
 198                                 return null;
 199                         }
 200                 }
 201
 202                 // Check whether this upload class is enabled
 203                 if ( !call_user_func( [ $className, 'isEnabled' ] ) ) {
 204                         return null;
 205                 }
 206
 207                 // Check whether the request is valid
 208                 if ( !call_user_func( [ $className, 'isValidRequest' ], $request ) ) {
 209                         return null;
 210                 }
 211
 212                 /** @var UploadBase $handler */
 213                 $handler = new $className;
 214
 215                 $handler->initializeFromRequest( $request );
 216
 217                 return $handler;
 218         }
 219
 220         /**
 221          * Check whether a request if valid for this handler
 222          * @param WebRequest $request
 223          * @return bool
 224          */
 225         public static function isValidRequest( $request ) {
 226                 return false;
 227         }
 228
 229         public function __construct() {
 230         }
 231
 232         /**
 233          * Returns the upload type. Should be overridden by child classes
 234          *
 235          * @since 1.18
 236          * @return string
 237          */
 238         public function getSourceType() {
 239                 return null;
 240         }
 241
 242         /**
 243          * Initialize the path information
 244          * @param string $name The desired destination name
 245          * @param string $tempPath The temporary path
 246          * @param int $fileSize The file size
 247          * @param bool $removeTempFile (false) remove the temporary file?
 248          * @throws MWException
 249          */
 250         public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
 251                 $this->mDesiredDestName = $name;
 252                 if ( FileBackend::isStoragePath( $tempPath ) ) {
 253                         throw new MWException( __METHOD__ . " given storage path `$tempPath`." );
 254                 }
 255
 256                 $this->setTempFile( $tempPath, $fileSize );
 257                 $this->mRemoveTempFile = $removeTempFile;
 258         }
 259
 260         /**
 261          * Initialize from a WebRequest. Override this in a subclass.
 262          *
 263          * @param WebRequest &$request
 264          */
 265         abstract public function initializeFromRequest( &$request );
 266
 267         /**
 268          * @param string $tempPath File system path to temporary file containing the upload
 269          * @param int|null $fileSize
 270          */
 271         protected function setTempFile( $tempPath, $fileSize = null ) {
 272                 $this->mTempPath = $tempPath;
 273                 $this->mFileSize = $fileSize ?: null;
 274                 if ( strlen( $this->mTempPath ) && file_exists( $this->mTempPath ) ) {
 275                         $this->tempFileObj = new TempFSFile( $this->mTempPath );
 276                         if ( !$fileSize ) {
 277                                 $this->mFileSize = filesize( $this->mTempPath );
 278                         }
 279                 } else {
 280                         $this->tempFileObj = null;
 281                 }
 282         }
 283
 284         /**
 285          * Fetch the file. Usually a no-op
 286          * @return Status
 287          */
 288         public function fetchFile() {
 289                 return Status::newGood();
 290         }
 291
 292         /**
 293          * Return true if the file is empty
 294          * @return bool
 295          */
 296         public function isEmptyFile() {
 297                 return empty( $this->mFileSize );
 298         }
 299
 300         /**
 301          * Return the file size
 302          * @return int
 303          */
 304         public function getFileSize() {
 305                 return $this->mFileSize;
 306         }
 307
 308         /**
 309          * Get the base 36 SHA1 of the file
 310          * @return string
 311          */
 312         public function getTempFileSha1Base36() {
 313                 return FSFile::getSha1Base36FromPath( $this->mTempPath );
 314         }
 315
 316         /**
 317          * @param string $srcPath The source path
 318          * @return string|bool The real path if it was a virtual URL Returns false on failure
 319          */
 320         public function getRealPath( $srcPath ) {
 321                 $repo = RepoGroup::singleton()->getLocalRepo();
 322                 if ( FileRepo::isVirtualUrl( $srcPath ) ) {
 323                         /** @todo Just make uploads work with storage paths UploadFromStash
 324                          *  loads files via virtual URLs.
 325                          */
 326                         $tmpFile = $repo->getLocalCopy( $srcPath );
 327                         if ( $tmpFile ) {
 328                                 $tmpFile->bind( $this ); // keep alive with $this
 329                         }
 330                         $path = $tmpFile ? $tmpFile->getPath() : false;
 331                 } else {
 332                         $path = $srcPath;
 333                 }
 334
 335                 return $path;
 336         }
 337
 338         /**
 339          * Verify whether the upload is sane.
 340          * @return mixed Const self::OK or else an array with error information
 341          */
 342         public function verifyUpload() {
 343                 /**
 344                  * If there was no filename or a zero size given, give up quick.
 345                  */
 346                 if ( $this->isEmptyFile() ) {
 347                         return [ 'status' => self::EMPTY_FILE ];
 348                 }
 349
 350                 /**
 351                  * Honor $wgMaxUploadSize
 352                  */
 353                 $maxSize = self::getMaxUploadSize( $this->getSourceType() );
 354                 if ( $this->mFileSize > $maxSize ) {
 355                         return [
 356                                 'status' => self::FILE_TOO_LARGE,
 357                                 'max' => $maxSize,
 358                         ];
 359                 }
 360
 361                 /**
 362                  * Look at the contents of the file; if we can recognize the
 363                  * type but it's corrupt or data of the wrong type, we should
 364                  * probably not accept it.
 365                  */
 366                 $verification = $this->verifyFile();
 367                 if ( $verification !== true ) {
 368                         return [
 369                                 'status' => self::VERIFICATION_ERROR,
 370                                 'details' => $verification
 371                         ];
 372                 }
 373
 374                 /**
 375                  * Make sure this file can be created
 376                  */
 377                 $result = $this->validateName();
 378                 if ( $result !== true ) {
 379                         return $result;
 380                 }
 381
 382                 $error = '';
 383                 if ( !Hooks::run( 'UploadVerification',
 384                         [ $this->mDestName, $this->mTempPath, &$error ], '1.28' )
 385                 ) {
 386                         return [ 'status' => self::HOOK_ABORTED, 'error' => $error ];
 387                 }
 388
 389                 return [ 'status' => self::OK ];
 390         }
 391
 392         /**
 393          * Verify that the name is valid and, if necessary, that we can overwrite
 394          *
 395          * @return mixed True if valid, otherwise and array with 'status'
 396          * and other keys
 397          */
 398         public function validateName() {
 399                 $nt = $this->getTitle();
 400                 if ( is_null( $nt ) ) {
 401                         $result = [ 'status' => $this->mTitleError ];
 402                         if ( $this->mTitleError == self::ILLEGAL_FILENAME ) {
 403                                 $result['filtered'] = $this->mFilteredName;
 404                         }
 405                         if ( $this->mTitleError == self::FILETYPE_BADTYPE ) {
 406                                 $result['finalExt'] = $this->mFinalExtension;
 407                                 if ( count( $this->mBlackListedExtensions ) ) {
 408                                         $result['blacklistedExt'] = $this->mBlackListedExtensions;
 409                                 }
 410                         }
 411
 412                         return $result;
 413                 }
 414                 $this->mDestName = $this->getLocalFile()->getName();
 415
 416                 return true;
 417         }
 418
 419         /**
 420          * Verify the MIME type.
 421          *
 422          * @note Only checks that it is not an evil MIME. The "does it have
 423          *  correct extension given its MIME type?" check is in verifyFile.
 424          *  in `verifyFile()` that MIME type and file extension correlate.
 425          * @param string $mime Representing the MIME
 426          * @return mixed True if the file is verified, an array otherwise
 427          */
 428         protected function verifyMimeType( $mime ) {
 429                 global $wgVerifyMimeType, $wgVerifyMimeTypeIE;
 430                 if ( $wgVerifyMimeType ) {
 431                         wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>\n" );
 432                         global $wgMimeTypeBlacklist;
 433                         if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
 434                                 return [ 'filetype-badmime', $mime ];
 435                         }
 436
 437                         if ( $wgVerifyMimeTypeIE ) {
 438                                 # Check what Internet Explorer would detect
 439                                 $fp = fopen( $this->mTempPath, 'rb' );
 440                                 $chunk = fread( $fp, 256 );
 441                                 fclose( $fp );
 442
 443                                 $magic = MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer();
 444                                 $extMime = $magic->guessTypesForExtension( $this->mFinalExtension );
 445                                 $ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime );
 446                                 foreach ( $ieTypes as $ieType ) {
 447                                         if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
 448                                                 return [ 'filetype-bad-ie-mime', $ieType ];
 449                                         }
 450                                 }
 451                         }
 452                 }
 453
 454                 return true;
 455         }
 456
 457         /**
 458          * Verifies that it's ok to include the uploaded file
 459          *
 460          * @return mixed True of the file is verified, array otherwise.
 461          */
 462         protected function verifyFile() {
 463                 global $wgVerifyMimeType, $wgDisableUploadScriptChecks;
 464
 465                 $status = $this->verifyPartialFile();
 466                 if ( $status !== true ) {
 467                         return $status;
 468                 }
 469
 470                 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer() );
 471                 $this->mFileProps = $mwProps->getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 472                 $mime = $this->mFileProps['mime'];
 473
 474                 if ( $wgVerifyMimeType ) {
 475                         # XXX: Missing extension will be caught by validateName() via getTitle()
 476                         if ( $this->mFinalExtension != '' && !$this->verifyExtension( $mime, $this->mFinalExtension ) ) {
 477                                 return [ 'filetype-mime-mismatch', $this->mFinalExtension, $mime ];
 478                         }
 479                 }
 480
 481                 # check for htmlish code and javascript
 482                 if ( !$wgDisableUploadScriptChecks ) {
 483                         if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 484                                 $svgStatus = $this->detectScriptInSvg( $this->mTempPath, false );
 485                                 if ( $svgStatus !== false ) {
 486                                         return $svgStatus;
 487                                 }
 488                         }
 489                 }
 490
 491                 $handler = MediaHandler::getHandler( $mime );
 492                 if ( $handler ) {
 493                         $handlerStatus = $handler->verifyUpload( $this->mTempPath );
 494                         if ( !$handlerStatus->isOK() ) {
 495                                 $errors = $handlerStatus->getErrorsArray();
 496
 497                                 return reset( $errors );
 498                         }
 499                 }
 500
 501                 $error = true;
 502                 Hooks::run( 'UploadVerifyFile', [ $this, $mime, &$error ] );
 503                 if ( $error !== true ) {
 504                         if ( !is_array( $error ) ) {
 505                                 $error = [ $error ];
 506                         }
 507                         return $error;
 508                 }
 509
 510                 wfDebug( __METHOD__ . ": all clear; passing.\n" );
 511
 512                 return true;
 513         }
 514
 515         /**
 516          * A verification routine suitable for partial files
 517          *
 518          * Runs the blacklist checks, but not any checks that may
 519          * assume the entire file is present.
 520          *
 521          * @return mixed True for valid or array with error message key.
 522          */
 523         protected function verifyPartialFile() {
 524                 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
 525
 526                 # getTitle() sets some internal parameters like $this->mFinalExtension
 527                 $this->getTitle();
 528
 529                 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer() );
 530                 $this->mFileProps = $mwProps->getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 531
 532                 # check MIME type, if desired
 533                 $mime = $this->mFileProps['file-mime'];
 534                 $status = $this->verifyMimeType( $mime );
 535                 if ( $status !== true ) {
 536                         return $status;
 537                 }
 538
 539                 # check for htmlish code and javascript
 540                 if ( !$wgDisableUploadScriptChecks ) {
 541                         if ( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
 542                                 return [ 'uploadscripted' ];
 543                         }
 544                         if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 545                                 $svgStatus = $this->detectScriptInSvg( $this->mTempPath, true );
 546                                 if ( $svgStatus !== false ) {
 547                                         return $svgStatus;
 548                                 }
 549                         }
 550                 }
 551
 552                 # Check for Java applets, which if uploaded can bypass cross-site
 553                 # restrictions.
 554                 if ( !$wgAllowJavaUploads ) {
 555                         $this->mJavaDetected = false;
 556                         $zipStatus = ZipDirectoryReader::read( $this->mTempPath,
 557                                 [ $this, 'zipEntryCallback' ] );
 558                         if ( !$zipStatus->isOK() ) {
 559                                 $errors = $zipStatus->getErrorsArray();
 560                                 $error = reset( $errors );
 561                                 if ( $error[0] !== 'zip-wrong-format' ) {
 562                                         return $error;
 563                                 }
 564                         }
 565                         if ( $this->mJavaDetected ) {
 566                                 return [ 'uploadjava' ];
 567                         }
 568                 }
 569
 570                 # Scan the uploaded file for viruses
 571                 $virus = $this->detectVirus( $this->mTempPath );
 572                 if ( $virus ) {
 573                         return [ 'uploadvirus', $virus ];
 574                 }
 575
 576                 return true;
 577         }
 578
 579         /**
 580          * Callback for ZipDirectoryReader to detect Java class files.
 581          *
 582          * @param array $entry
 583          */
 584         public function zipEntryCallback( $entry ) {
 585                 $names = [ $entry['name'] ];
 586
 587                 // If there is a null character, cut off the name at it, because JDK's
 588                 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
 589                 // were constructed which had ".class\0" followed by a string chosen to
 590                 // make the hash collide with the truncated name, that file could be
 591                 // returned in response to a request for the .class file.
 592                 $nullPos = strpos( $entry['name'], "\000" );
 593                 if ( $nullPos !== false ) {
 594                         $names[] = substr( $entry['name'], 0, $nullPos );
 595                 }
 596
 597                 // If there is a trailing slash in the file name, we have to strip it,
 598                 // because that's what ZIP_GetEntry() does.
 599                 if ( preg_grep( '!\.class/?$!', $names ) ) {
 600                         $this->mJavaDetected = true;
 601                 }
 602         }
 603
 604         /**
 605          * Alias for verifyTitlePermissions. The function was originally
 606          * 'verifyPermissions', but that suggests it's checking the user, when it's
 607          * really checking the title + user combination.
 608          *
 609          * @param User $user User object to verify the permissions against
 610          * @return mixed An array as returned by getUserPermissionsErrors or true
 611          *   in case the user has proper permissions.
 612          */
 613         public function verifyPermissions( $user ) {
 614                 return $this->verifyTitlePermissions( $user );
 615         }
 616
 617         /**
 618          * Check whether the user can edit, upload and create the image. This
 619          * checks only against the current title; if it returns errors, it may
 620          * very well be that another title will not give errors. Therefore
 621          * isAllowed() should be called as well for generic is-user-blocked or
 622          * can-user-upload checking.
 623          *
 624          * @param User $user User object to verify the permissions against
 625          * @return mixed An array as returned by getUserPermissionsErrors or true
 626          *   in case the user has proper permissions.
 627          */
 628         public function verifyTitlePermissions( $user ) {
 629                 /**
 630                  * If the image is protected, non-sysop users won't be able
 631                  * to modify it by uploading a new revision.
 632                  */
 633                 $nt = $this->getTitle();
 634                 if ( is_null( $nt ) ) {
 635                         return true;
 636                 }
 637                 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
 638                 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
 639                 if ( !$nt->exists() ) {
 640                         $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
 641                 } else {
 642                         $permErrorsCreate = [];
 643                 }
 644                 if ( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
 645                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
 646                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
 647
 648                         return $permErrors;
 649                 }
 650
 651                 $overwriteError = $this->checkOverwrite( $user );
 652                 if ( $overwriteError !== true ) {
 653                         return [ $overwriteError ];
 654                 }
 655
 656                 return true;
 657         }
 658
 659         /**
 660          * Check for non fatal problems with the file.
 661          *
 662          * This should not assume that mTempPath is set.
 663          *
 664          * @return mixed[] Array of warnings
 665          */
 666         public function checkWarnings() {
 667                 $warnings = [];
 668
 669                 $localFile = $this->getLocalFile();
 670                 $localFile->load( File::READ_LATEST );
 671                 $filename = $localFile->getName();
 672                 $hash = $this->getTempFileSha1Base36();
 673
 674                 $badFileName = $this->checkBadFileName( $filename, $this->mDesiredDestName );
 675                 if ( $badFileName !== null ) {
 676                         $warnings['badfilename'] = $badFileName;
 677                 }
 678
 679                 $unwantedFileExtensionDetails = $this->checkUnwantedFileExtensions( $this->mFinalExtension );
 680                 if ( $unwantedFileExtensionDetails !== null ) {
 681                         $warnings['filetype-unwanted-type'] = $unwantedFileExtensionDetails;
 682                 }
 683
 684                 $fileSizeWarnings = $this->checkFileSize( $this->mFileSize );
 685                 if ( $fileSizeWarnings ) {
 686                         $warnings = array_merge( $warnings, $fileSizeWarnings );
 687                 }
 688
 689                 $localFileExistsWarnings = $this->checkLocalFileExists( $localFile, $hash );
 690                 if ( $localFileExistsWarnings ) {
 691                         $warnings = array_merge( $warnings, $localFileExistsWarnings );
 692                 }
 693
 694                 if ( $this->checkLocalFileWasDeleted( $localFile ) ) {
 695                         $warnings['was-deleted'] = $filename;
 696                 }
 697
 698                 // If a file with the same name exists locally then the local file has already been tested
 699                 // for duplication of content
 700                 $ignoreLocalDupes = isset( $warnings[ 'exists '] );
 701                 $dupes = $this->checkAgainstExistingDupes( $hash, $ignoreLocalDupes );
 702                 if ( $dupes ) {
 703                         $warnings['duplicate'] = $dupes;
 704                 }
 705
 706                 $archivedDupes = $this->checkAgainstArchiveDupes( $hash );
 707                 if ( $archivedDupes !== null ) {
 708                         $warnings['duplicate-archive'] = $archivedDupes;
 709                 }
 710
 711                 return $warnings;
 712         }
 713
 714         /**
 715          * Check whether the resulting filename is different from the desired one,
 716          * but ignore things like ucfirst() and spaces/underscore things
 717          *
 718          * @param string $filename
 719          * @param string $desiredFileName
 720          *
 721          * @return string|null String that was determined to be bad or null if the filename is okay
 722          */
 723         private function checkBadFileName( $filename, $desiredFileName ) {
 724                 $comparableName = str_replace( ' ', '_', $desiredFileName );
 725                 $comparableName = Title::capitalize( $comparableName, NS_FILE );
 726
 727                 if ( $desiredFileName != $filename && $comparableName != $filename ) {
 728                         return $filename;
 729                 }
 730
 731                 return null;
 732         }
 733
 734         /**
 735          * @param string $fileExtension The file extension to check
 736          *
 737          * @return array|null array with the following keys:
 738          *                    0 => string The final extension being used
 739          *                    1 => string[] The extensions that are allowed
 740          *                    2 => int The number of extensions that are allowed.
 741          */
 742         private function checkUnwantedFileExtensions( $fileExtension ) {
 743                 global $wgCheckFileExtensions, $wgFileExtensions, $wgLang;
 744
 745                 if ( $wgCheckFileExtensions ) {
 746                         $extensions = array_unique( $wgFileExtensions );
 747                         if ( !$this->checkFileExtension( $fileExtension, $extensions ) ) {
 748                                 return [
 749                                         $fileExtension,
 750                                         $wgLang->commaList( $extensions ),
 751                                         count( $extensions )
 752                                 ];
 753                         }
 754                 }
 755
 756                 return null;
 757         }
 758
 759         /**
 760          * @param int $fileSize
 761          *
 762          * @return array warnings
 763          */
 764         private function checkFileSize( $fileSize ) {
 765                 global $wgUploadSizeWarning;
 766
 767                 $warnings = [];
 768
 769                 if ( $wgUploadSizeWarning && ( $fileSize > $wgUploadSizeWarning ) ) {
 770                         $warnings['large-file'] = [ $wgUploadSizeWarning, $fileSize ];
 771                 }
 772
 773                 if ( $fileSize == 0 ) {
 774                         $warnings['empty-file'] = true;
 775                 }
 776
 777                 return $warnings;
 778         }
 779
 780         /**
 781          * @param LocalFile $localFile
 782          * @param string $hash sha1 hash of the file to check
 783          *
 784          * @return array warnings
 785          */
 786         private function checkLocalFileExists( LocalFile $localFile, $hash ) {
 787                 $warnings = [];
 788
 789                 $exists = self::getExistsWarning( $localFile );
 790                 if ( $exists !== false ) {
 791                         $warnings['exists'] = $exists;
 792
 793                         // check if file is an exact duplicate of current file version
 794                         if ( $hash === $localFile->getSha1() ) {
 795                                 $warnings['no-change'] = $localFile;
 796                         }
 797
 798                         // check if file is an exact duplicate of older versions of this file
 799                         $history = $localFile->getHistory();
 800                         foreach ( $history as $oldFile ) {
 801                                 if ( $hash === $oldFile->getSha1() ) {
 802                                         $warnings['duplicate-version'][] = $oldFile;
 803                                 }
 804                         }
 805                 }
 806
 807                 return $warnings;
 808         }
 809
 810         private function checkLocalFileWasDeleted( LocalFile $localFile ) {
 811                 return $localFile->wasDeleted() && !$localFile->exists();
 812         }
 813
 814         /**
 815          * @param string $hash sha1 hash of the file to check
 816          * @param bool $ignoreLocalDupes True to ignore local duplicates
 817          *
 818          * @return File[] Duplicate files, if found.
 819          */
 820         private function checkAgainstExistingDupes( $hash, $ignoreLocalDupes ) {
 821                 $dupes = RepoGroup::singleton()->findBySha1( $hash );
 822                 $title = $this->getTitle();
 823                 foreach ( $dupes as $key => $dupe ) {
 824                         if (
 825                                 ( $dupe instanceof LocalFile ) &&
 826                                 $ignoreLocalDupes &&
 827                                 $title->equals( $dupe->getTitle() )
 828                         ) {
 829                                 unset( $dupes[$key] );
 830                         }
 831                 }
 832
 833                 return $dupes;
 834         }
 835
 836         /**
 837          * @param string $hash sha1 hash of the file to check
 838          *
 839          * @return string|null Name of the dupe or empty string if discovered (depending on visibility)
 840          *                     null if the check discovered no dupes.
 841          */
 842         private function checkAgainstArchiveDupes( $hash ) {
 843                 $archivedFile = new ArchivedFile( null, 0, '', $hash );
 844                 if ( $archivedFile->getID() > 0 ) {
 845                         if ( $archivedFile->userCan( File::DELETED_FILE ) ) {
 846                                 return $archivedFile->getName();
 847                         } else {
 848                                 return '';
 849                         }
 850                 }
 851
 852                 return null;
 853         }
 854
 855         /**
 856          * Really perform the upload. Stores the file in the local repo, watches
 857          * if necessary and runs the UploadComplete hook.
 858          *
 859          * @param string $comment
 860          * @param string $pageText
 861          * @param bool $watch Whether the file page should be added to user's watchlist.
 862          *   (This doesn't check $user's permissions.)
 863          * @param User $user
 864          * @param string[] $tags Change tags to add to the log entry and page revision.
 865          *   (This doesn't check $user's permissions.)
 866          * @return Status Indicating the whether the upload succeeded.
 867          */
 868         public function performUpload( $comment, $pageText, $watch, $user, $tags = [] ) {
 869                 $this->getLocalFile()->load( File::READ_LATEST );
 870                 $props = $this->mFileProps;
 871
 872                 $error = null;
 873                 Hooks::run( 'UploadVerifyUpload', [ $this, $user, $props, $comment, $pageText, &$error ] );
 874                 if ( $error ) {
 875                         if ( !is_array( $error ) ) {
 876                                 $error = [ $error ];
 877                         }
 878                         return Status::newFatal( ...$error );
 879                 }
 880
 881                 $status = $this->getLocalFile()->upload(
 882                         $this->mTempPath,
 883                         $comment,
 884                         $pageText,
 885                         File::DELETE_SOURCE,
 886                         $props,
 887                         false,
 888                         $user,
 889                         $tags
 890                 );
 891
 892                 if ( $status->isGood() ) {
 893                         if ( $watch ) {
 894                                 WatchAction::doWatch(
 895                                         $this->getLocalFile()->getTitle(),
 896                                         $user,
 897                                         User::IGNORE_USER_RIGHTS
 898                                 );
 899                         }
 900                         // Avoid PHP 7.1 warning of passing $this by reference
 901                         $uploadBase = $this;
 902                         Hooks::run( 'UploadComplete', [ &$uploadBase ] );
 903
 904                         $this->postProcessUpload();
 905                 }
 906
 907                 return $status;
 908         }
 909
 910         /**
 911          * Perform extra steps after a successful upload.
 912          *
 913          * @since  1.25
 914          */
 915         public function postProcessUpload() {
 916         }
 917
 918         /**
 919          * Returns the title of the file to be uploaded. Sets mTitleError in case
 920          * the name was illegal.
 921          *
 922          * @return Title|null The title of the file or null in case the name was illegal
 923          */
 924         public function getTitle() {
 925                 if ( $this->mTitle !== false ) {
 926                         return $this->mTitle;
 927                 }
 928                 if ( !is_string( $this->mDesiredDestName ) ) {
 929                         $this->mTitleError = self::ILLEGAL_FILENAME;
 930                         $this->mTitle = null;
 931
 932                         return $this->mTitle;
 933                 }
 934                 /* Assume that if a user specified File:Something.jpg, this is an error
 935                  * and that the namespace prefix needs to be stripped of.
 936                  */
 937                 $title = Title::newFromText( $this->mDesiredDestName );
 938                 if ( $title && $title->getNamespace() == NS_FILE ) {
 939                         $this->mFilteredName = $title->getDBkey();
 940                 } else {
 941                         $this->mFilteredName = $this->mDesiredDestName;
 942                 }
 943
 944                 # oi_archive_name is max 255 bytes, which include a timestamp and an
 945                 # exclamation mark, so restrict file name to 240 bytes.
 946                 if ( strlen( $this->mFilteredName ) > 240 ) {
 947                         $this->mTitleError = self::FILENAME_TOO_LONG;
 948                         $this->mTitle = null;
 949
 950                         return $this->mTitle;
 951                 }
 952
 953                 /**
 954                  * Chop off any directories in the given filename. Then
 955                  * filter out illegal characters, and try to make a legible name
 956                  * out of it. We'll strip some silently that Title would die on.
 957                  */
 958                 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mFilteredName );
 959                 /* Normalize to title form before we do any further processing */
 960                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 961                 if ( is_null( $nt ) ) {
 962                         $this->mTitleError = self::ILLEGAL_FILENAME;
 963                         $this->mTitle = null;
 964
 965                         return $this->mTitle;
 966                 }
 967                 $this->mFilteredName = $nt->getDBkey();
 968
 969                 /**
 970                  * We'll want to blacklist against *any* 'extension', and use
 971                  * only the final one for the whitelist.
 972                  */
 973                 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
 974
 975                 if ( $ext !== [] ) {
 976                         $this->mFinalExtension = trim( end( $ext ) );
 977                 } else {
 978                         $this->mFinalExtension = '';
 979
 980                         # No extension, try guessing one
 981                         $magic = MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer();
 982                         $mime = $magic->guessMimeType( $this->mTempPath );
 983                         if ( $mime !== 'unknown/unknown' ) {
 984                                 # Get a space separated list of extensions
 985                                 $extList = $magic->getExtensionsForType( $mime );
 986                                 if ( $extList ) {
 987                                         # Set the extension to the canonical extension
 988                                         $this->mFinalExtension = strtok( $extList, ' ' );
 989
 990                                         # Fix up the other variables
 991                                         $this->mFilteredName .= ".{$this->mFinalExtension}";
 992                                         $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 993                                         $ext = [ $this->mFinalExtension ];
 994                                 }
 995                         }
 996                 }
 997
 998                 /* Don't allow users to override the blacklist (check file extension) */
 999                 global $wgCheckFileExtensions, $wgStrictFileExtensions;
1000                 global $wgFileExtensions, $wgFileBlacklist;
1001
1002                 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
1003
1004                 if ( $this->mFinalExtension == '' ) {
1005                         $this->mTitleError = self::FILETYPE_MISSING;
1006                         $this->mTitle = null;
1007
1008                         return $this->mTitle;
1009                 } elseif ( $blackListedExtensions ||
1010                         ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
1011                                 !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) )
1012                 ) {
1013                         $this->mBlackListedExtensions = $blackListedExtensions;
1014                         $this->mTitleError = self::FILETYPE_BADTYPE;
1015                         $this->mTitle = null;
1016
1017                         return $this->mTitle;
1018                 }
1019
1020                 // Windows may be broken with special characters, see T3780
1021                 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
1022                         && !RepoGroup::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
1023                 ) {
1024                         $this->mTitleError = self::WINDOWS_NONASCII_FILENAME;
1025                         $this->mTitle = null;
1026
1027                         return $this->mTitle;
1028                 }
1029
1030                 # If there was more than one "extension", reassemble the base
1031                 # filename to prevent bogus complaints about length
1032                 if ( count( $ext ) > 1 ) {
1033                         $iterations = count( $ext ) - 1;
1034                         for ( $i = 0; $i < $iterations; $i++ ) {
1035                                 $partname .= '.' . $ext[$i];
1036                         }
1037                 }
1038
1039                 if ( strlen( $partname ) < 1 ) {
1040                         $this->mTitleError = self::MIN_LENGTH_PARTNAME;
1041                         $this->mTitle = null;
1042
1043                         return $this->mTitle;
1044                 }
1045
1046                 $this->mTitle = $nt;
1047
1048                 return $this->mTitle;
1049         }
1050
1051         /**
1052          * Return the local file and initializes if necessary.
1053          *
1054          * @return LocalFile|null
1055          */
1056         public function getLocalFile() {
1057                 if ( is_null( $this->mLocalFile ) ) {
1058                         $nt = $this->getTitle();
1059                         $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
1060                 }
1061
1062                 return $this->mLocalFile;
1063         }
1064
1065         /**
1066          * @return UploadStashFile|null
1067          */
1068         public function getStashFile() {
1069                 return $this->mStashFile;
1070         }
1071
1072         /**
1073          * Like stashFile(), but respects extensions' wishes to prevent the stashing. verifyUpload() must
1074          * be called before calling this method (unless $isPartial is true).
1075          *
1076          * Upload stash exceptions are also caught and converted to an error status.
1077          *
1078          * @since 1.28
1079          * @param User $user
1080          * @param bool $isPartial Pass `true` if this is a part of a chunked upload (not a complete file).
1081          * @return Status If successful, value is an UploadStashFile instance
1082          */
1083         public function tryStashFile( User $user, $isPartial = false ) {
1084                 if ( !$isPartial ) {
1085                         $error = $this->runUploadStashFileHook( $user );
1086                         if ( $error ) {
1087                                 return Status::newFatal( ...$error );
1088                         }
1089                 }
1090                 try {
1091                         $file = $this->doStashFile( $user );
1092                         return Status::newGood( $file );
1093                 } catch ( UploadStashException $e ) {
1094                         return Status::newFatal( 'uploadstash-exception', get_class( $e ), $e->getMessage() );
1095                 }
1096         }
1097
1098         /**
1099          * @param User $user
1100          * @return array|null Error message and parameters, null if there's no error
1101          */
1102         protected function runUploadStashFileHook( User $user ) {
1103                 $props = $this->mFileProps;
1104                 $error = null;
1105                 Hooks::run( 'UploadStashFile', [ $this, $user, $props, &$error ] );
1106                 if ( $error && !is_array( $error ) ) {
1107                         $error = [ $error ];
1108                 }
1109                 return $error;
1110         }
1111
1112         /**
1113          * If the user does not supply all necessary information in the first upload
1114          * form submission (either by accident or by design) then we may want to
1115          * stash the file temporarily, get more information, and publish the file
1116          * later.
1117          *
1118          * This method will stash a file in a temporary directory for later
1119          * processing, and save the necessary descriptive info into the database.
1120          * This method returns the file object, which also has a 'fileKey' property
1121          * which can be passed through a form or API request to find this stashed
1122          * file again.
1123          *
1124          * @deprecated since 1.28 Use tryStashFile() instead
1125          * @param User|null $user
1126          * @return UploadStashFile Stashed file
1127          * @throws UploadStashBadPathException
1128          * @throws UploadStashFileException
1129          * @throws UploadStashNotLoggedInException
1130          */
1131         public function stashFile( User $user = null ) {
1132                 return $this->doStashFile( $user );
1133         }
1134
1135         /**
1136          * Implementation for stashFile() and tryStashFile().
1137          *
1138          * @param User|null $user
1139          * @return UploadStashFile Stashed file
1140          */
1141         protected function doStashFile( User $user = null ) {
1142                 $stash = RepoGroup::singleton()->getLocalRepo()->getUploadStash( $user );
1143                 $file = $stash->stashFile( $this->mTempPath, $this->getSourceType() );
1144                 $this->mStashFile = $file;
1145
1146                 return $file;
1147         }
1148
1149         /**
1150          * Stash a file in a temporary directory, returning a key which can be used
1151          * to find the file again. See stashFile().
1152          *
1153          * @deprecated since 1.28
1154          * @return string File key
1155          */
1156         public function stashFileGetKey() {
1157                 wfDeprecated( __METHOD__, '1.28' );
1158                 return $this->doStashFile()->getFileKey();
1159         }
1160
1161         /**
1162          * alias for stashFileGetKey, for backwards compatibility
1163          *
1164          * @deprecated since 1.28
1165          * @return string File key
1166          */
1167         public function stashSession() {
1168                 wfDeprecated( __METHOD__, '1.28' );
1169                 return $this->doStashFile()->getFileKey();
1170         }
1171
1172         /**
1173          * If we've modified the upload file we need to manually remove it
1174          * on exit to clean up.
1175          */
1176         public function cleanupTempFile() {
1177                 if ( $this->mRemoveTempFile && $this->tempFileObj ) {
1178                         // Delete when all relevant TempFSFile handles go out of scope
1179                         wfDebug( __METHOD__ . ": Marked temporary file '{$this->mTempPath}' for removal\n" );
1180                         $this->tempFileObj->autocollect();
1181                 }
1182         }
1183
1184         public function getTempPath() {
1185                 return $this->mTempPath;
1186         }
1187
1188         /**
1189          * Split a file into a base name and all dot-delimited 'extensions'
1190          * on the end. Some web server configurations will fall back to
1191          * earlier pseudo-'extensions' to determine type and execute
1192          * scripts, so the blacklist needs to check them all.
1193          *
1194          * @param string $filename
1195          * @return array [ string, string[] ]
1196          */
1197         public static function splitExtensions( $filename ) {
1198                 $bits = explode( '.', $filename );
1199                 $basename = array_shift( $bits );
1200
1201                 return [ $basename, $bits ];
1202         }
1203
1204         /**
1205          * Perform case-insensitive match against a list of file extensions.
1206          * Returns true if the extension is in the list.
1207          *
1208          * @param string $ext
1209          * @param array $list
1210          * @return bool
1211          */
1212         public static function checkFileExtension( $ext, $list ) {
1213                 return in_array( strtolower( $ext ), $list );
1214         }
1215
1216         /**
1217          * Perform case-insensitive match against a list of file extensions.
1218          * Returns an array of matching extensions.
1219          *
1220          * @param string[] $ext
1221          * @param string[] $list
1222          * @return string[]
1223          */
1224         public static function checkFileExtensionList( $ext, $list ) {
1225                 return array_intersect( array_map( 'strtolower', $ext ), $list );
1226         }
1227
1228         /**
1229          * Checks if the MIME type of the uploaded file matches the file extension.
1230          *
1231          * @param string $mime The MIME type of the uploaded file
1232          * @param string $extension The filename extension that the file is to be served with
1233          * @return bool
1234          */
1235         public static function verifyExtension( $mime, $extension ) {
1236                 $magic = MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer();
1237
1238                 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' ) {
1239                         if ( !$magic->isRecognizableExtension( $extension ) ) {
1240                                 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
1241                                         "unrecognized extension '$extension', can't verify\n" );
1242
1243                                 return true;
1244                         } else {
1245                                 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; " .
1246                                         "recognized extension '$extension', so probably invalid file\n" );
1247
1248                                 return false;
1249                         }
1250                 }
1251
1252                 $match = $magic->isMatchingExtension( $extension, $mime );
1253
1254                 if ( $match === null ) {
1255                         if ( $magic->getTypesForExtension( $extension ) !== null ) {
1256                                 wfDebug( __METHOD__ . ": No extension known for $mime, but we know a mime for $extension\n" );
1257
1258                                 return false;
1259                         } else {
1260                                 wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
1261
1262                                 return true;
1263                         }
1264                 } elseif ( $match === true ) {
1265                         wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
1266
1267                         /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1268                         return true;
1269                 } else {
1270                         wfDebug( __METHOD__
1271                                 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1272
1273                         return false;
1274                 }
1275         }
1276
1277         /**
1278          * Heuristic for detecting files that *could* contain JavaScript instructions or
1279          * things that may look like HTML to a browser and are thus
1280          * potentially harmful. The present implementation will produce false
1281          * positives in some situations.
1282          *
1283          * @param string $file Pathname to the temporary upload file
1284          * @param string $mime The MIME type of the file
1285          * @param string $extension The extension of the file
1286          * @return bool True if the file contains something looking like embedded scripts
1287          */
1288         public static function detectScript( $file, $mime, $extension ) {
1289                 # ugly hack: for text files, always look at the entire file.
1290                 # For binary field, just check the first K.
1291
1292                 $isText = strpos( $mime, 'text/' ) === 0;
1293                 if ( $isText ) {
1294                         $chunk = file_get_contents( $file );
1295                 } else {
1296                         $fp = fopen( $file, 'rb' );
1297                         $chunk = fread( $fp, 1024 );
1298                         fclose( $fp );
1299                 }
1300
1301                 $chunk = strtolower( $chunk );
1302
1303                 if ( !$chunk ) {
1304                         return false;
1305                 }
1306
1307                 # decode from UTF-16 if needed (could be used for obfuscation).
1308                 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1309                         $enc = 'UTF-16BE';
1310                 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1311                         $enc = 'UTF-16LE';
1312                 } else {
1313                         $enc = null;
1314                 }
1315
1316                 if ( $enc ) {
1317                         $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1318                 }
1319
1320                 $chunk = trim( $chunk );
1321
1322                 /** @todo FIXME: Convert from UTF-16 if necessary! */
1323                 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
1324
1325                 # check for HTML doctype
1326                 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1327                         return true;
1328                 }
1329
1330                 // Some browsers will interpret obscure xml encodings as UTF-8, while
1331                 // PHP/expat will interpret the given encoding in the xml declaration (T49304)
1332                 if ( $extension == 'svg' || strpos( $mime, 'image/svg' ) === 0 ) {
1333                         if ( self::checkXMLEncodingMissmatch( $file ) ) {
1334                                 return true;
1335                         }
1336                 }
1337
1338                 // Quick check for HTML heuristics in old IE and Safari.
1339                 //
1340                 // The exact heuristics IE uses are checked separately via verifyMimeType(), so we
1341                 // don't need them all here as it can cause many false positives.
1342                 //
1343                 // Check for `<script` and such still to forbid script tags and embedded HTML in SVG:
1344                 $tags = [
1345                         '<body',
1346                         '<head',
1347                         '<html', # also in safari
1348                         '<script', # also in safari
1349                 ];
1350
1351                 foreach ( $tags as $tag ) {
1352                         if ( strpos( $chunk, $tag ) !== false ) {
1353                                 wfDebug( __METHOD__ . ": found something that may make it be mistaken for html: $tag\n" );
1354
1355                                 return true;
1356                         }
1357                 }
1358
1359                 /*
1360                  * look for JavaScript
1361                  */
1362
1363                 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1364                 $chunk = Sanitizer::decodeCharReferences( $chunk );
1365
1366                 # look for script-types
1367                 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1368                         wfDebug( __METHOD__ . ": found script types\n" );
1369
1370                         return true;
1371                 }
1372
1373                 # look for html-style script-urls
1374                 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1375                         wfDebug( __METHOD__ . ": found html-style script urls\n" );
1376
1377                         return true;
1378                 }
1379
1380                 # look for css-style script-urls
1381                 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1382                         wfDebug( __METHOD__ . ": found css-style script urls\n" );
1383
1384                         return true;
1385                 }
1386
1387                 wfDebug( __METHOD__ . ": no scripts found\n" );
1388
1389                 return false;
1390         }
1391
1392         /**
1393          * Check a whitelist of xml encodings that are known not to be interpreted differently
1394          * by the server's xml parser (expat) and some common browsers.
1395          *
1396          * @param string $file Pathname to the temporary upload file
1397          * @return bool True if the file contains an encoding that could be misinterpreted
1398          */
1399         public static function checkXMLEncodingMissmatch( $file ) {
1400                 global $wgSVGMetadataCutoff;
1401                 $contents = file_get_contents( $file, false, null, 0, $wgSVGMetadataCutoff );
1402                 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1403
1404                 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1405                         if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1406                                 && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1407                         ) {
1408                                 wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1409
1410                                 return true;
1411                         }
1412                 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1413                         // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1414                         // bytes. There shouldn't be a legitimate reason for this to happen.
1415                         wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1416
1417                         return true;
1418                 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1419                         // EBCDIC encoded XML
1420                         wfDebug( __METHOD__ . ": EBCDIC Encoded XML\n" );
1421
1422                         return true;
1423                 }
1424
1425                 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1426                 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1427                 $attemptEncodings = [ 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' ];
1428                 foreach ( $attemptEncodings as $encoding ) {
1429                         Wikimedia\suppressWarnings();
1430                         $str = iconv( $encoding, 'UTF-8', $contents );
1431                         Wikimedia\restoreWarnings();
1432                         if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1433                                 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1434                                         && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1435                                 ) {
1436                                         wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1437
1438                                         return true;
1439                                 }
1440                         } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1441                                 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1442                                 // bytes. There shouldn't be a legitimate reason for this to happen.
1443                                 wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1444
1445                                 return true;
1446                         }
1447                 }
1448
1449                 return false;
1450         }
1451
1452         /**
1453          * @param string $filename
1454          * @param bool $partial
1455          * @return bool|array
1456          */
1457         protected function detectScriptInSvg( $filename, $partial ) {
1458                 $this->mSVGNSError = false;
1459                 $check = new XmlTypeCheck(
1460                         $filename,
1461                         [ $this, 'checkSvgScriptCallback' ],
1462                         true,
1463                         [
1464                                 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback',
1465                                 'external_dtd_handler' => 'UploadBase::checkSvgExternalDTD',
1466                         ]
1467                 );
1468                 if ( $check->wellFormed !== true ) {
1469                         // Invalid xml (T60553)
1470                         // But only when non-partial (T67724)
1471                         return $partial ? false : [ 'uploadinvalidxml' ];
1472                 } elseif ( $check->filterMatch ) {
1473                         if ( $this->mSVGNSError ) {
1474                                 return [ 'uploadscriptednamespace', $this->mSVGNSError ];
1475                         }
1476
1477                         return $check->filterMatchType;
1478                 }
1479
1480                 return false;
1481         }
1482
1483         /**
1484          * Callback to filter SVG Processing Instructions.
1485          * @param string $target Processing instruction name
1486          * @param string $data Processing instruction attribute and value
1487          * @return bool|array
1488          */
1489         public static function checkSvgPICallback( $target, $data ) {
1490                 // Don't allow external stylesheets (T59550)
1491                 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1492                         return [ 'upload-scripted-pi-callback' ];
1493                 }
1494
1495                 return false;
1496         }
1497
1498         /**
1499          * Verify that DTD urls referenced are only the standard dtds
1500          *
1501          * Browsers seem to ignore external dtds. However just to be on the
1502          * safe side, only allow dtds from the svg standard.
1503          *
1504          * @param string $type PUBLIC or SYSTEM
1505          * @param string $publicId The well-known public identifier for the dtd
1506          * @param string $systemId The url for the external dtd
1507          * @return bool|array
1508          */
1509         public static function checkSvgExternalDTD( $type, $publicId, $systemId ) {
1510                 // This doesn't include the XHTML+MathML+SVG doctype since we don't
1511                 // allow XHTML anyways.
1512                 $allowedDTDs = [
1513                         'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd',
1514                         'http://www.w3.org/TR/2001/REC-SVG-20010904/DTD/svg10.dtd',
1515                         'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-basic.dtd',
1516                         'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-tiny.dtd',
1517                         // https://phabricator.wikimedia.org/T168856
1518                         'http://www.w3.org/TR/2001/PR-SVG-20010719/DTD/svg10.dtd',
1519                 ];
1520                 if ( $type !== 'PUBLIC'
1521                         || !in_array( $systemId, $allowedDTDs )
1522                         || strpos( $publicId, "-//W3C//" ) !== 0
1523                 ) {
1524                         return [ 'upload-scripted-dtd' ];
1525                 }
1526                 return false;
1527         }
1528
1529         /**
1530          * @todo Replace this with a whitelist filter!
1531          * @param string $element
1532          * @param array $attribs
1533          * @param string|null $data
1534          * @return bool|array
1535          */
1536         public function checkSvgScriptCallback( $element, $attribs, $data = null ) {
1537                 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1538
1539                 // We specifically don't include:
1540                 // http://www.w3.org/1999/xhtml (T62771)
1541                 static $validNamespaces = [
1542                         '',
1543                         'adobe:ns:meta/',
1544                         'http://creativecommons.org/ns#',
1545                         'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1546                         'http://ns.adobe.com/adobeillustrator/10.0/',
1547                         'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1548                         'http://ns.adobe.com/extensibility/1.0/',
1549                         'http://ns.adobe.com/flows/1.0/',
1550                         'http://ns.adobe.com/illustrator/1.0/',
1551                         'http://ns.adobe.com/imagereplacement/1.0/',
1552                         'http://ns.adobe.com/pdf/1.3/',
1553                         'http://ns.adobe.com/photoshop/1.0/',
1554                         'http://ns.adobe.com/saveforweb/1.0/',
1555                         'http://ns.adobe.com/variables/1.0/',
1556                         'http://ns.adobe.com/xap/1.0/',
1557                         'http://ns.adobe.com/xap/1.0/g/',
1558                         'http://ns.adobe.com/xap/1.0/g/img/',
1559                         'http://ns.adobe.com/xap/1.0/mm/',
1560                         'http://ns.adobe.com/xap/1.0/rights/',
1561                         'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1562                         'http://ns.adobe.com/xap/1.0/stype/font#',
1563                         'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1564                         'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1565                         'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1566                         'http://ns.adobe.com/xap/1.0/t/pg/',
1567                         'http://purl.org/dc/elements/1.1/',
1568                         'http://purl.org/dc/elements/1.1',
1569                         'http://schemas.microsoft.com/visio/2003/svgextensions/',
1570                         'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1571                         'http://taptrix.com/inkpad/svg_extensions',
1572                         'http://web.resource.org/cc/',
1573                         'http://www.freesoftware.fsf.org/bkchem/cdml',
1574                         'http://www.inkscape.org/namespaces/inkscape',
1575                         'http://www.opengis.net/gml',
1576                         'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1577                         'http://www.w3.org/2000/svg',
1578                         'http://www.w3.org/tr/rec-rdf-syntax/',
1579                         'http://www.w3.org/2000/01/rdf-schema#',
1580                 ];
1581
1582                 // Inkscape mangles namespace definitions created by Adobe Illustrator.
1583                 // This is nasty but harmless. (T144827)
1584                 $isBuggyInkscape = preg_match( '/^&(#38;)*ns_[a-z_]+;$/', $namespace );
1585
1586                 if ( !( $isBuggyInkscape || in_array( $namespace, $validNamespaces ) ) ) {
1587                         wfDebug( __METHOD__ . ": Non-svg namespace '$namespace' in uploaded file.\n" );
1588                         /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1589                         $this->mSVGNSError = $namespace;
1590
1591                         return true;
1592                 }
1593
1594                 /*
1595                  * check for elements that can contain javascript
1596                  */
1597                 if ( $strippedElement == 'script' ) {
1598                         wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
1599
1600                         return [ 'uploaded-script-svg', $strippedElement ];
1601                 }
1602
1603                 # e.g., <svg xmlns="http://www.w3.org/2000/svg">
1604                 #  <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1605                 if ( $strippedElement == 'handler' ) {
1606                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1607
1608                         return [ 'uploaded-script-svg', $strippedElement ];
1609                 }
1610
1611                 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1612                 if ( $strippedElement == 'stylesheet' ) {
1613                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1614
1615                         return [ 'uploaded-script-svg', $strippedElement ];
1616                 }
1617
1618                 # Block iframes, in case they pass the namespace check
1619                 if ( $strippedElement == 'iframe' ) {
1620                         wfDebug( __METHOD__ . ": iframe in uploaded file.\n" );
1621
1622                         return [ 'uploaded-script-svg', $strippedElement ];
1623                 }
1624
1625                 # Check <style> css
1626                 if ( $strippedElement == 'style'
1627                         && self::checkCssFragment( Sanitizer::normalizeCss( $data ) )
1628                 ) {
1629                         wfDebug( __METHOD__ . ": hostile css in style element.\n" );
1630                         return [ 'uploaded-hostile-svg' ];
1631                 }
1632
1633                 foreach ( $attribs as $attrib => $value ) {
1634                         $stripped = $this->stripXmlNamespace( $attrib );
1635                         $value = strtolower( $value );
1636
1637                         if ( substr( $stripped, 0, 2 ) == 'on' ) {
1638                                 wfDebug( __METHOD__
1639                                         . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1640
1641                                 return [ 'uploaded-event-handler-on-svg', $attrib, $value ];
1642                         }
1643
1644                         # Do not allow relative links, or unsafe url schemas.
1645                         # For <a> tags, only data:, http: and https: and same-document
1646                         # fragment links are allowed. For all other tags, only data:
1647                         # and fragment are allowed.
1648                         if ( $stripped == 'href'
1649                                 && $value !== ''
1650                                 && strpos( $value, 'data:' ) !== 0
1651                                 && strpos( $value, '#' ) !== 0
1652                         ) {
1653                                 if ( !( $strippedElement === 'a'
1654                                         && preg_match( '!^https?://!i', $value ) )
1655                                 ) {
1656                                         wfDebug( __METHOD__ . ": Found href attribute <$strippedElement "
1657                                                 . "'$attrib'='$value' in uploaded file.\n" );
1658
1659                                         return [ 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value ];
1660                                 }
1661                         }
1662
1663                         # only allow data: targets that should be safe. This prevents vectors like,
1664                         # image/svg, text/xml, application/xml, and text/html, which can contain scripts
1665                         if ( $stripped == 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) {
1666                                 // rfc2397 parameters. This is only slightly slower than (;[\w;]+)*.
1667                                 // phpcs:ignore Generic.Files.LineLength
1668                                 $parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?';
1669
1670                                 if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|png)$parameters,!i", $value ) ) {
1671                                         wfDebug( __METHOD__ . ": Found href to unwhitelisted data: uri "
1672                                                 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1673                                         return [ 'uploaded-href-unsafe-target-svg', $strippedElement, $attrib, $value ];
1674                                 }
1675                         }
1676
1677                         # Change href with animate from (http://html5sec.org/#137).
1678                         if ( $stripped === 'attributename'
1679                                 && $strippedElement === 'animate'
1680                                 && $this->stripXmlNamespace( $value ) == 'href'
1681                         ) {
1682                                 wfDebug( __METHOD__ . ": Found animate that might be changing href using from "
1683                                         . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1684
1685                                 return [ 'uploaded-animate-svg', $strippedElement, $attrib, $value ];
1686                         }
1687
1688                         # use set/animate to add event-handler attribute to parent
1689                         if ( ( $strippedElement == 'set' || $strippedElement == 'animate' )
1690                                 && $stripped == 'attributename'
1691                                 && substr( $value, 0, 2 ) == 'on'
1692                         ) {
1693                                 wfDebug( __METHOD__ . ": Found svg setting event-handler attribute with "
1694                                         . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1695
1696                                 return [ 'uploaded-setting-event-handler-svg', $strippedElement, $stripped, $value ];
1697                         }
1698
1699                         # use set to add href attribute to parent element
1700                         if ( $strippedElement == 'set'
1701                                 && $stripped == 'attributename'
1702                                 && strpos( $value, 'href' ) !== false
1703                         ) {
1704                                 wfDebug( __METHOD__ . ": Found svg setting href attribute '$value' in uploaded file.\n" );
1705
1706                                 return [ 'uploaded-setting-href-svg' ];
1707                         }
1708
1709                         # use set to add a remote / data / script target to an element
1710                         if ( $strippedElement == 'set'
1711                                 && $stripped == 'to'
1712                                 && preg_match( '!(http|https|data|script):!sim', $value )
1713                         ) {
1714                                 wfDebug( __METHOD__ . ": Found svg setting attribute to '$value' in uploaded file.\n" );
1715
1716                                 return [ 'uploaded-wrong-setting-svg', $value ];
1717                         }
1718
1719                         # use handler attribute with remote / data / script
1720                         if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1721                                 wfDebug( __METHOD__ . ": Found svg setting handler with remote/data/script "
1722                                         . "'$attrib'='$value' in uploaded file.\n" );
1723
1724                                 return [ 'uploaded-setting-handler-svg', $attrib, $value ];
1725                         }
1726
1727                         # use CSS styles to bring in remote code
1728                         if ( $stripped == 'style'
1729                                 && self::checkCssFragment( Sanitizer::normalizeCss( $value ) )
1730                         ) {
1731                                 wfDebug( __METHOD__ . ": Found svg setting a style with "
1732                                         . "remote url '$attrib'='$value' in uploaded file.\n" );
1733                                 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1734                         }
1735
1736                         # Several attributes can include css, css character escaping isn't allowed
1737                         $cssAttrs = [ 'font', 'clip-path', 'fill', 'filter', 'marker',
1738                                 'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke' ];
1739                         if ( in_array( $stripped, $cssAttrs )
1740                                 && self::checkCssFragment( $value )
1741                         ) {
1742                                 wfDebug( __METHOD__ . ": Found svg setting a style with "
1743                                         . "remote url '$attrib'='$value' in uploaded file.\n" );
1744                                 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1745                         }
1746
1747                         # image filters can pull in url, which could be svg that executes scripts
1748                         # Only allow url( "#foo" ). Do not allow url( http://example.com )
1749                         if ( $strippedElement == 'image'
1750                                 && $stripped == 'filter'
1751                                 && preg_match( '!url\s*\(\s*["\']?[^#]!sim', $value )
1752                         ) {
1753                                 wfDebug( __METHOD__ . ": Found image filter with url: "
1754                                         . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1755
1756                                 return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ];
1757                         }
1758                 }
1759
1760                 return false; // No scripts detected
1761         }
1762
1763         /**
1764          * Check a block of CSS or CSS fragment for anything that looks like
1765          * it is bringing in remote code.
1766          * @param string $value a string of CSS
1767          * @param bool $propOnly only check css properties (start regex with :)
1768          * @return bool true if the CSS contains an illegal string, false if otherwise
1769          */
1770         private static function checkCssFragment( $value ) {
1771                 # Forbid external stylesheets, for both reliability and to protect viewer's privacy
1772                 if ( stripos( $value, '@import' ) !== false ) {
1773                         return true;
1774                 }
1775
1776                 # We allow @font-face to embed fonts with data: urls, so we snip the string
1777                 # 'url' out so this case won't match when we check for urls below
1778                 $pattern = '!(@font-face\s*{[^}]*src:)url(\("data:;base64,)!im';
1779                 $value = preg_replace( $pattern, '$1$2', $value );
1780
1781                 # Check for remote and executable CSS. Unlike in Sanitizer::checkCss, the CSS
1782                 # properties filter and accelerator don't seem to be useful for xss in SVG files.
1783                 # Expression and -o-link don't seem to work either, but filtering them here in case.
1784                 # Additionally, we catch remote urls like url("http:..., url('http:..., url(http:...,
1785                 # but not local ones such as url("#..., url('#..., url(#....
1786                 if ( preg_match( '!expression
1787                                 | -o-link\s*:
1788                                 | -o-link-source\s*:
1789                                 | -o-replace\s*:!imx', $value ) ) {
1790                         return true;
1791                 }
1792
1793                 if ( preg_match_all(
1794                                 "!(\s*(url|image|image-set)\s*\(\s*[\"']?\s*[^#]+.*?\))!sim",
1795                                 $value,
1796                                 $matches
1797                         ) !== 0
1798                 ) {
1799                         # TODO: redo this in one regex. Until then, url("#whatever") matches the first
1800                         foreach ( $matches[1] as $match ) {
1801                                 if ( !preg_match( "!\s*(url|image|image-set)\s*\(\s*(#|'#|\"#)!im", $match ) ) {
1802                                         return true;
1803                                 }
1804                         }
1805                 }
1806
1807                 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
1808                         return true;
1809                 }
1810
1811                 return false;
1812         }
1813
1814         /**
1815          * Divide the element name passed by the xml parser to the callback into URI and prifix.
1816          * @param string $element
1817          * @return array Containing the namespace URI and prefix
1818          */
1819         private static function splitXmlNamespace( $element ) {
1820                 // 'http://www.w3.org/2000/svg:script' -> [ 'http://www.w3.org/2000/svg', 'script' ]
1821                 $parts = explode( ':', strtolower( $element ) );
1822                 $name = array_pop( $parts );
1823                 $ns = implode( ':', $parts );
1824
1825                 return [ $ns, $name ];
1826         }
1827
1828         /**
1829          * @param string $name
1830          * @return string
1831          */
1832         private function stripXmlNamespace( $name ) {
1833                 // 'http://www.w3.org/2000/svg:script' -> 'script'
1834                 $parts = explode( ':', strtolower( $name ) );
1835
1836                 return array_pop( $parts );
1837         }
1838
1839         /**
1840          * Generic wrapper function for a virus scanner program.
1841          * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1842          * $wgAntivirusRequired may be used to deny upload if the scan fails.
1843          *
1844          * @param string $file Pathname to the temporary upload file
1845          * @return bool|null|string False if not virus is found, null if the scan fails or is disabled,
1846          *   or a string containing feedback from the virus scanner if a virus was found.
1847          *   If textual feedback is missing but a virus was found, this function returns true.
1848          */
1849         public static function detectVirus( $file ) {
1850                 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1851
1852                 if ( !$wgAntivirus ) {
1853                         wfDebug( __METHOD__ . ": virus scanner disabled\n" );
1854
1855                         return null;
1856                 }
1857
1858                 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1859                         wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
1860                         $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1861                                 [ 'virus-badscanner', $wgAntivirus ] );
1862
1863                         return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1864                 }
1865
1866                 # look up scanner configuration
1867                 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1868                 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1869                 $msgPattern = $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ?? null;
1870
1871                 if ( strpos( $command, "%f" ) === false ) {
1872                         # simple pattern: append file to scan
1873                         $command .= " " . Shell::escape( $file );
1874                 } else {
1875                         # complex pattern: replace "%f" with file to scan
1876                         $command = str_replace( "%f", Shell::escape( $file ), $command );
1877                 }
1878
1879                 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
1880
1881                 # execute virus scanner
1882                 $exitCode = false;
1883
1884                 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1885                 #      that does not seem to be worth the pain.
1886                 #      Ask me (Duesentrieb) about it if it's ever needed.
1887                 $output = wfShellExecWithStderr( $command, $exitCode );
1888
1889                 # map exit code to AV_xxx constants.
1890                 $mappedCode = $exitCode;
1891                 if ( $exitCodeMap ) {
1892                         if ( isset( $exitCodeMap[$exitCode] ) ) {
1893                                 $mappedCode = $exitCodeMap[$exitCode];
1894                         } elseif ( isset( $exitCodeMap["*"] ) ) {
1895                                 $mappedCode = $exitCodeMap["*"];
1896                         }
1897                 }
1898
1899                 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1900                  * so we need the strict equalities === and thus can't use a switch here
1901                  */
1902                 if ( $mappedCode === AV_SCAN_FAILED ) {
1903                         # scan failed (code was mapped to false by $exitCodeMap)
1904                         wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
1905
1906                         $output = $wgAntivirusRequired
1907                                 ? wfMessage( 'virus-scanfailed', [ $exitCode ] )->text()
1908                                 : null;
1909                 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
1910                         # scan failed because filetype is unknown (probably imune)
1911                         wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
1912                         $output = null;
1913                 } elseif ( $mappedCode === AV_NO_VIRUS ) {
1914                         # no virus found
1915                         wfDebug( __METHOD__ . ": file passed virus scan.\n" );
1916                         $output = false;
1917                 } else {
1918                         $output = trim( $output );
1919
1920                         if ( !$output ) {
1921                                 $output = true; # if there's no output, return true
1922                         } elseif ( $msgPattern ) {
1923                                 $groups = [];
1924                                 if ( preg_match( $msgPattern, $output, $groups ) && $groups[1] ) {
1925                                         $output = $groups[1];
1926                                 }
1927                         }
1928
1929                         wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
1930                 }
1931
1932                 return $output;
1933         }
1934
1935         /**
1936          * Check if there's an overwrite conflict and, if so, if restrictions
1937          * forbid this user from performing the upload.
1938          *
1939          * @param User $user
1940          *
1941          * @return bool|array
1942          */
1943         private function checkOverwrite( $user ) {
1944                 // First check whether the local file can be overwritten
1945                 $file = $this->getLocalFile();
1946                 $file->load( File::READ_LATEST );
1947                 if ( $file->exists() ) {
1948                         if ( !self::userCanReUpload( $user, $file ) ) {
1949                                 return [ 'fileexists-forbidden', $file->getName() ];
1950                         } else {
1951                                 return true;
1952                         }
1953                 }
1954
1955                 /* Check shared conflicts: if the local file does not exist, but
1956                  * wfFindFile finds a file, it exists in a shared repository.
1957                  */
1958                 $file = wfFindFile( $this->getTitle(), [ 'latest' => true ] );
1959                 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1960                         return [ 'fileexists-shared-forbidden', $file->getName() ];
1961                 }
1962
1963                 return true;
1964         }
1965
1966         /**
1967          * Check if a user is the last uploader
1968          *
1969          * @param User $user
1970          * @param File $img
1971          * @return bool
1972          */
1973         public static function userCanReUpload( User $user, File $img ) {
1974                 if ( $user->isAllowed( 'reupload' ) ) {
1975                         return true; // non-conditional
1976                 } elseif ( !$user->isAllowed( 'reupload-own' ) ) {
1977                         return false;
1978                 }
1979
1980                 if ( !( $img instanceof LocalFile ) ) {
1981                         return false;
1982                 }
1983
1984                 $img->load();
1985
1986                 return $user->getId() == $img->getUser( 'id' );
1987         }
1988
1989         /**
1990          * Helper function that does various existence checks for a file.
1991          * The following checks are performed:
1992          * - The file exists
1993          * - Article with the same name as the file exists
1994          * - File exists with normalized extension
1995          * - The file looks like a thumbnail and the original exists
1996          *
1997          * @param File $file The File object to check
1998          * @return mixed False if the file does not exists, else an array
1999          */
2000         public static function getExistsWarning( $file ) {
2001                 if ( $file->exists() ) {
2002                         return [ 'warning' => 'exists', 'file' => $file ];
2003                 }
2004
2005                 if ( $file->getTitle()->getArticleID() ) {
2006                         return [ 'warning' => 'page-exists', 'file' => $file ];
2007                 }
2008
2009                 if ( strpos( $file->getName(), '.' ) == false ) {
2010                         $partname = $file->getName();
2011                         $extension = '';
2012                 } else {
2013                         $n = strrpos( $file->getName(), '.' );
2014                         $extension = substr( $file->getName(), $n + 1 );
2015                         $partname = substr( $file->getName(), 0, $n );
2016                 }
2017                 $normalizedExtension = File::normalizeExtension( $extension );
2018
2019                 if ( $normalizedExtension != $extension ) {
2020                         // We're not using the normalized form of the extension.
2021                         // Normal form is lowercase, using most common of alternate
2022                         // extensions (eg 'jpg' rather than 'JPEG').
2023
2024                         // Check for another file using the normalized form...
2025                         $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
2026                         $file_lc = wfLocalFile( $nt_lc );
2027
2028                         if ( $file_lc->exists() ) {
2029                                 return [
2030                                         'warning' => 'exists-normalized',
2031                                         'file' => $file,
2032                                         'normalizedFile' => $file_lc
2033                                 ];
2034                         }
2035                 }
2036
2037                 // Check for files with the same name but a different extension
2038                 $similarFiles = RepoGroup::singleton()->getLocalRepo()->findFilesByPrefix(
2039                         "{$partname}.", 1 );
2040                 if ( count( $similarFiles ) ) {
2041                         return [
2042                                 'warning' => 'exists-normalized',
2043                                 'file' => $file,
2044                                 'normalizedFile' => $similarFiles[0],
2045                         ];
2046                 }
2047
2048                 if ( self::isThumbName( $file->getName() ) ) {
2049                         # Check for filenames like 50px- or 180px-, these are mostly thumbnails
2050                         $nt_thb = Title::newFromText(
2051                                 substr( $partname, strpos( $partname, '-' ) + 1 ) . '.' . $extension,
2052                                 NS_FILE
2053                         );
2054                         $file_thb = wfLocalFile( $nt_thb );
2055                         if ( $file_thb->exists() ) {
2056                                 return [
2057                                         'warning' => 'thumb',
2058                                         'file' => $file,
2059                                         'thumbFile' => $file_thb
2060                                 ];
2061                         } else {
2062                                 // File does not exist, but we just don't like the name
2063                                 return [
2064                                         'warning' => 'thumb-name',
2065                                         'file' => $file,
2066                                         'thumbFile' => $file_thb
2067                                 ];
2068                         }
2069                 }
2070
2071                 foreach ( self::getFilenamePrefixBlacklist() as $prefix ) {
2072                         if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
2073                                 return [
2074                                         'warning' => 'bad-prefix',
2075                                         'file' => $file,
2076                                         'prefix' => $prefix
2077                                 ];
2078                         }
2079                 }
2080
2081                 return false;
2082         }
2083
2084         /**
2085          * Helper function that checks whether the filename looks like a thumbnail
2086          * @param string $filename
2087          * @return bool
2088          */
2089         public static function isThumbName( $filename ) {
2090                 $n = strrpos( $filename, '.' );
2091                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
2092
2093                 return (
2094                                 substr( $partname, 3, 3 ) == 'px-' ||
2095                                 substr( $partname, 2, 3 ) == 'px-'
2096                         ) &&
2097                         preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
2098         }
2099
2100         /**
2101          * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
2102          *
2103          * @return array List of prefixes
2104          */
2105         public static function getFilenamePrefixBlacklist() {
2106                 $blacklist = [];
2107                 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
2108                 if ( !$message->isDisabled() ) {
2109                         $lines = explode( "\n", $message->plain() );
2110                         foreach ( $lines as $line ) {
2111                                 // Remove comment lines
2112                                 $comment = substr( trim( $line ), 0, 1 );
2113                                 if ( $comment == '#' || $comment == '' ) {
2114                                         continue;
2115                                 }
2116                                 // Remove additional comments after a prefix
2117                                 $comment = strpos( $line, '#' );
2118                                 if ( $comment > 0 ) {
2119                                         $line = substr( $line, 0, $comment - 1 );
2120                                 }
2121                                 $blacklist[] = trim( $line );
2122                         }
2123                 }
2124
2125                 return $blacklist;
2126         }
2127
2128         /**
2129          * Gets image info about the file just uploaded.
2130          *
2131          * Also has the effect of setting metadata to be an 'indexed tag name' in
2132          * returned API result if 'metadata' was requested. Oddly, we have to pass
2133          * the "result" object down just so it can do that with the appropriate
2134          * format, presumably.
2135          *
2136          * @param ApiResult $result
2137          * @return array Image info
2138          */
2139         public function getImageInfo( $result ) {
2140                 $localFile = $this->getLocalFile();
2141                 $stashFile = $this->getStashFile();
2142                 // Calling a different API module depending on whether the file was stashed is less than optimal.
2143                 // In fact, calling API modules here at all is less than optimal. Maybe it should be refactored.
2144                 if ( $stashFile ) {
2145                         $imParam = ApiQueryStashImageInfo::getPropertyNames();
2146                         $info = ApiQueryStashImageInfo::getInfo( $stashFile, array_flip( $imParam ), $result );
2147                 } else {
2148                         $imParam = ApiQueryImageInfo::getPropertyNames();
2149                         $info = ApiQueryImageInfo::getInfo( $localFile, array_flip( $imParam ), $result );
2150                 }
2151
2152                 return $info;
2153         }
2154
2155         /**
2156          * @param array $error
2157          * @return Status
2158          */
2159         public function convertVerifyErrorToStatus( $error ) {
2160                 $code = $error['status'];
2161                 unset( $code['status'] );
2162
2163                 return Status::newFatal( $this->getVerificationErrorCode( $code ), $error );
2164         }
2165
2166         /**
2167          * Get the MediaWiki maximum uploaded file size for given type of upload, based on
2168          * $wgMaxUploadSize.
2169          *
2170          * @param null|string $forType
2171          * @return int
2172          */
2173         public static function getMaxUploadSize( $forType = null ) {
2174                 global $wgMaxUploadSize;
2175
2176                 if ( is_array( $wgMaxUploadSize ) ) {
2177                         if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
2178                                 return $wgMaxUploadSize[$forType];
2179                         } else {
2180                                 return $wgMaxUploadSize['*'];
2181                         }
2182                 } else {
2183                         return intval( $wgMaxUploadSize );
2184                 }
2185         }
2186
2187         /**
2188          * Get the PHP maximum uploaded file size, based on ini settings. If there is no limit or the
2189          * limit can't be guessed, returns a very large number (PHP_INT_MAX).
2190          *
2191          * @since 1.27
2192          * @return int
2193          */
2194         public static function getMaxPhpUploadSize() {
2195                 $phpMaxFileSize = wfShorthandToInteger(
2196                         ini_get( 'upload_max_filesize' ) ?: ini_get( 'hhvm.server.upload.upload_max_file_size' ),
2197                         PHP_INT_MAX
2198                 );
2199                 $phpMaxPostSize = wfShorthandToInteger(
2200                         ini_get( 'post_max_size' ) ?: ini_get( 'hhvm.server.max_post_size' ),
2201                         PHP_INT_MAX
2202                 ) ?: PHP_INT_MAX;
2203                 return min( $phpMaxFileSize, $phpMaxPostSize );
2204         }
2205
2206         /**
2207          * Get the current status of a chunked upload (used for polling)
2208          *
2209          * The value will be read from cache.
2210          *
2211          * @param User $user
2212          * @param string $statusKey
2213          * @return Status[]|bool
2214          */
2215         public static function getSessionStatus( User $user, $statusKey ) {
2216                 $store = self::getUploadSessionStore();
2217                 $key = self::getUploadSessionKey( $store, $user, $statusKey );
2218
2219                 return $store->get( $key );
2220         }
2221
2222         /**
2223          * Set the current status of a chunked upload (used for polling)
2224          *
2225          * The value will be set in cache for 1 day
2226          *
2227          * Avoid triggering this method on HTTP GET/HEAD requests
2228          *
2229          * @param User $user
2230          * @param string $statusKey
2231          * @param array|bool $value
2232          * @return void
2233          */
2234         public static function setSessionStatus( User $user, $statusKey, $value ) {
2235                 $store = self::getUploadSessionStore();
2236                 $key = self::getUploadSessionKey( $store, $user, $statusKey );
2237
2238                 if ( $value === false ) {
2239                         $store->delete( $key );
2240                 } else {
2241                         $store->set( $key, $value, $store::TTL_DAY );
2242                 }
2243         }
2244
2245         /**
2246          * @param BagOStuff $store
2247          * @param User $user
2248          * @param string $statusKey
2249          * @return string
2250          */
2251         private static function getUploadSessionKey( BagOStuff $store, User $user, $statusKey ) {
2252                 return $store->makeKey(
2253                         'uploadstatus',
2254                         $user->getId() ?: md5( $user->getName() ),
2255                         $statusKey
2256                 );
2257         }
2258
2259         /**
2260          * @return BagOStuff
2261          */
2262         private static function getUploadSessionStore() {
2263                 return ObjectCache::getInstance( 'db-replicated' );
2264         }
2265 }