includes/upload/UploadBase.php

   1 <?php
   2 /**
   3  * Base class for the backend of file upload.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Upload
  22  */
  23
  24 use MediaWiki\MediaWikiServices;
  25 use MediaWiki\Shell\Shell;
  26 use MediaWiki\User\UserIdentity;
  27
  28 /**
  29  * @defgroup Upload Upload related
  30  */
  31
  32 /**
  33  * @ingroup Upload
  34  *
  35  * UploadBase and subclasses are the backend of MediaWiki's file uploads.
  36  * The frontends are formed by ApiUpload and SpecialUpload.
  37  *
  38  * @author Brion Vibber
  39  * @author Bryan Tong Minh
  40  * @author Michael Dale
  41  */
  42 abstract class UploadBase {
  43         /** @var string Local file system path to the file to upload (or a local copy) */
  44         protected $mTempPath;
  45         /** @var TempFSFile|null Wrapper to handle deleting the temp file */
  46         protected $tempFileObj;
  47         /** @var string|null */
  48         protected $mDesiredDestName;
  49         /** @var string|null */
  50         protected $mDestName;
  51         /** @var string|null */
  52         protected $mRemoveTempFile;
  53         /** @var string|null */
  54         protected $mSourceType;
  55         /** @var Title|bool */
  56         protected $mTitle = false;
  57         /** @var int */
  58         protected $mTitleError = 0;
  59         /** @var string|null */
  60         protected $mFilteredName;
  61         /** @var string|null */
  62         protected $mFinalExtension;
  63         /** @var LocalFile */
  64         protected $mLocalFile;
  65         /** @var UploadStashFile */
  66         protected $mStashFile;
  67         /** @var int|null */
  68         protected $mFileSize;
  69         /** @var array|null */
  70         protected $mFileProps;
  71         /** @var string[] */
  72         protected $mBlackListedExtensions;
  73         /** @var bool|null */
  74         protected $mJavaDetected;
  75         /** @var string|null */
  76         protected $mSVGNSError;
  77
  78         protected static $safeXmlEncodings = [
  79                 'UTF-8',
  80                 'ISO-8859-1',
  81                 'ISO-8859-2',
  82                 'UTF-16',
  83                 'UTF-32',
  84                 'WINDOWS-1250',
  85                 'WINDOWS-1251',
  86                 'WINDOWS-1252',
  87                 'WINDOWS-1253',
  88                 'WINDOWS-1254',
  89                 'WINDOWS-1255',
  90                 'WINDOWS-1256',
  91                 'WINDOWS-1257',
  92                 'WINDOWS-1258',
  93         ];
  94
  95         const SUCCESS = 0;
  96         const OK = 0;
  97         const EMPTY_FILE = 3;
  98         const MIN_LENGTH_PARTNAME = 4;
  99         const ILLEGAL_FILENAME = 5;
 100         const OVERWRITE_EXISTING_FILE = 7; # Not used anymore; handled by verifyTitlePermissions()
 101         const FILETYPE_MISSING = 8;
 102         const FILETYPE_BADTYPE = 9;
 103         const VERIFICATION_ERROR = 10;
 104         const HOOK_ABORTED = 11;
 105         const FILE_TOO_LARGE = 12;
 106         const WINDOWS_NONASCII_FILENAME = 13;
 107         const FILENAME_TOO_LONG = 14;
 108
 109         /**
 110          * @param int $error
 111          * @return string
 112          */
 113         public function getVerificationErrorCode( $error ) {
 114                 $code_to_status = [
 115                         self::EMPTY_FILE => 'empty-file',
 116                         self::FILE_TOO_LARGE => 'file-too-large',
 117                         self::FILETYPE_MISSING => 'filetype-missing',
 118                         self::FILETYPE_BADTYPE => 'filetype-banned',
 119                         self::MIN_LENGTH_PARTNAME => 'filename-tooshort',
 120                         self::ILLEGAL_FILENAME => 'illegal-filename',
 121                         self::OVERWRITE_EXISTING_FILE => 'overwrite',
 122                         self::VERIFICATION_ERROR => 'verification-error',
 123                         self::HOOK_ABORTED => 'hookaborted',
 124                         self::WINDOWS_NONASCII_FILENAME => 'windows-nonascii-filename',
 125                         self::FILENAME_TOO_LONG => 'filename-toolong',
 126                 ];
 127                 return $code_to_status[$error] ?? 'unknown-error';
 128         }
 129
 130         /**
 131          * Returns true if uploads are enabled.
 132          * Can be override by subclasses.
 133          * @return bool
 134          */
 135         public static function isEnabled() {
 136                 global $wgEnableUploads;
 137
 138                 if ( !$wgEnableUploads ) {
 139                         return false;
 140                 }
 141
 142                 # Check php's file_uploads setting
 143                 return wfIsHHVM() || wfIniGetBool( 'file_uploads' );
 144         }
 145
 146         /**
 147          * Returns true if the user can use this upload module or else a string
 148          * identifying the missing permission.
 149          * Can be overridden by subclasses.
 150          *
 151          * @param UserIdentity $user
 152          * @return bool|string
 153          */
 154         public static function isAllowed( UserIdentity $user ) {
 155                 $permissionManager = MediaWikiServices::getInstance()->getPermissionManager();
 156                 foreach ( [ 'upload', 'edit' ] as $permission ) {
 157                         if ( !$permissionManager->userHasRight( $user, $permission ) ) {
 158                                 return $permission;
 159                         }
 160                 }
 161
 162                 return true;
 163         }
 164
 165         /**
 166          * Returns true if the user has surpassed the upload rate limit, false otherwise.
 167          *
 168          * @param User $user
 169          * @return bool
 170          */
 171         public static function isThrottled( $user ) {
 172                 return $user->pingLimiter( 'upload' );
 173         }
 174
 175         // Upload handlers. Should probably just be a global.
 176         private static $uploadHandlers = [ 'Stash', 'File', 'Url' ];
 177
 178         /**
 179          * Create a form of UploadBase depending on wpSourceType and initializes it
 180          *
 181          * @param WebRequest &$request
 182          * @param string|null $type
 183          * @return null|UploadBase
 184          */
 185         public static function createFromRequest( &$request, $type = null ) {
 186                 $type = $type ?: $request->getVal( 'wpSourceType', 'File' );
 187
 188                 if ( !$type ) {
 189                         return null;
 190                 }
 191
 192                 // Get the upload class
 193                 $type = ucfirst( $type );
 194
 195                 // Give hooks the chance to handle this request
 196                 $className = null;
 197                 Hooks::run( 'UploadCreateFromRequest', [ $type, &$className ] );
 198                 if ( is_null( $className ) ) {
 199                         $className = 'UploadFrom' . $type;
 200                         wfDebug( __METHOD__ . ": class name: $className\n" );
 201                         if ( !in_array( $type, self::$uploadHandlers ) ) {
 202                                 return null;
 203                         }
 204                 }
 205
 206                 // Check whether this upload class is enabled
 207                 if ( !call_user_func( [ $className, 'isEnabled' ] ) ) {
 208                         return null;
 209                 }
 210
 211                 // Check whether the request is valid
 212                 if ( !call_user_func( [ $className, 'isValidRequest' ], $request ) ) {
 213                         return null;
 214                 }
 215
 216                 /** @var UploadBase $handler */
 217                 $handler = new $className;
 218
 219                 $handler->initializeFromRequest( $request );
 220
 221                 return $handler;
 222         }
 223
 224         /**
 225          * Check whether a request if valid for this handler
 226          * @param WebRequest $request
 227          * @return bool
 228          */
 229         public static function isValidRequest( $request ) {
 230                 return false;
 231         }
 232
 233         public function __construct() {
 234         }
 235
 236         /**
 237          * Returns the upload type. Should be overridden by child classes
 238          *
 239          * @since 1.18
 240          * @return string
 241          */
 242         public function getSourceType() {
 243                 return null;
 244         }
 245
 246         /**
 247          * Initialize the path information
 248          * @param string $name The desired destination name
 249          * @param string $tempPath The temporary path
 250          * @param int $fileSize The file size
 251          * @param bool $removeTempFile (false) remove the temporary file?
 252          * @throws MWException
 253          */
 254         public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
 255                 $this->mDesiredDestName = $name;
 256                 if ( FileBackend::isStoragePath( $tempPath ) ) {
 257                         throw new MWException( __METHOD__ . " given storage path `$tempPath`." );
 258                 }
 259
 260                 $this->setTempFile( $tempPath, $fileSize );
 261                 $this->mRemoveTempFile = $removeTempFile;
 262         }
 263
 264         /**
 265          * Initialize from a WebRequest. Override this in a subclass.
 266          *
 267          * @param WebRequest &$request
 268          */
 269         abstract public function initializeFromRequest( &$request );
 270
 271         /**
 272          * @param string $tempPath File system path to temporary file containing the upload
 273          * @param int|null $fileSize
 274          */
 275         protected function setTempFile( $tempPath, $fileSize = null ) {
 276                 $this->mTempPath = $tempPath;
 277                 $this->mFileSize = $fileSize ?: null;
 278                 if ( strlen( $this->mTempPath ) && file_exists( $this->mTempPath ) ) {
 279                         $this->tempFileObj = new TempFSFile( $this->mTempPath );
 280                         if ( !$fileSize ) {
 281                                 $this->mFileSize = filesize( $this->mTempPath );
 282                         }
 283                 } else {
 284                         $this->tempFileObj = null;
 285                 }
 286         }
 287
 288         /**
 289          * Fetch the file. Usually a no-op
 290          * @return Status
 291          */
 292         public function fetchFile() {
 293                 return Status::newGood();
 294         }
 295
 296         /**
 297          * Return true if the file is empty
 298          * @return bool
 299          */
 300         public function isEmptyFile() {
 301                 return empty( $this->mFileSize );
 302         }
 303
 304         /**
 305          * Return the file size
 306          * @return int
 307          */
 308         public function getFileSize() {
 309                 return $this->mFileSize;
 310         }
 311
 312         /**
 313          * Get the base 36 SHA1 of the file
 314          * @return string
 315          */
 316         public function getTempFileSha1Base36() {
 317                 return FSFile::getSha1Base36FromPath( $this->mTempPath );
 318         }
 319
 320         /**
 321          * @param string $srcPath The source path
 322          * @return string|bool The real path if it was a virtual URL Returns false on failure
 323          */
 324         public function getRealPath( $srcPath ) {
 325                 $repo = RepoGroup::singleton()->getLocalRepo();
 326                 if ( FileRepo::isVirtualUrl( $srcPath ) ) {
 327                         /** @todo Just make uploads work with storage paths UploadFromStash
 328                          *  loads files via virtual URLs.
 329                          */
 330                         $tmpFile = $repo->getLocalCopy( $srcPath );
 331                         if ( $tmpFile ) {
 332                                 $tmpFile->bind( $this ); // keep alive with $this
 333                         }
 334                         $path = $tmpFile ? $tmpFile->getPath() : false;
 335                 } else {
 336                         $path = $srcPath;
 337                 }
 338
 339                 return $path;
 340         }
 341
 342         /**
 343          * Verify whether the upload is sane.
 344          * @return mixed Const self::OK or else an array with error information
 345          */
 346         public function verifyUpload() {
 347                 /**
 348                  * If there was no filename or a zero size given, give up quick.
 349                  */
 350                 if ( $this->isEmptyFile() ) {
 351                         return [ 'status' => self::EMPTY_FILE ];
 352                 }
 353
 354                 /**
 355                  * Honor $wgMaxUploadSize
 356                  */
 357                 $maxSize = self::getMaxUploadSize( $this->getSourceType() );
 358                 if ( $this->mFileSize > $maxSize ) {
 359                         return [
 360                                 'status' => self::FILE_TOO_LARGE,
 361                                 'max' => $maxSize,
 362                         ];
 363                 }
 364
 365                 /**
 366                  * Look at the contents of the file; if we can recognize the
 367                  * type but it's corrupt or data of the wrong type, we should
 368                  * probably not accept it.
 369                  */
 370                 $verification = $this->verifyFile();
 371                 if ( $verification !== true ) {
 372                         return [
 373                                 'status' => self::VERIFICATION_ERROR,
 374                                 'details' => $verification
 375                         ];
 376                 }
 377
 378                 /**
 379                  * Make sure this file can be created
 380                  */
 381                 $result = $this->validateName();
 382                 if ( $result !== true ) {
 383                         return $result;
 384                 }
 385
 386                 return [ 'status' => self::OK ];
 387         }
 388
 389         /**
 390          * Verify that the name is valid and, if necessary, that we can overwrite
 391          *
 392          * @return mixed True if valid, otherwise and array with 'status'
 393          * and other keys
 394          */
 395         public function validateName() {
 396                 $nt = $this->getTitle();
 397                 if ( is_null( $nt ) ) {
 398                         $result = [ 'status' => $this->mTitleError ];
 399                         if ( $this->mTitleError == self::ILLEGAL_FILENAME ) {
 400                                 $result['filtered'] = $this->mFilteredName;
 401                         }
 402                         if ( $this->mTitleError == self::FILETYPE_BADTYPE ) {
 403                                 $result['finalExt'] = $this->mFinalExtension;
 404                                 if ( count( $this->mBlackListedExtensions ) ) {
 405                                         $result['blacklistedExt'] = $this->mBlackListedExtensions;
 406                                 }
 407                         }
 408
 409                         return $result;
 410                 }
 411                 $this->mDestName = $this->getLocalFile()->getName();
 412
 413                 return true;
 414         }
 415
 416         /**
 417          * Verify the MIME type.
 418          *
 419          * @note Only checks that it is not an evil MIME. The "does it have
 420          *  correct extension given its MIME type?" check is in verifyFile.
 421          *  in `verifyFile()` that MIME type and file extension correlate.
 422          * @param string $mime Representing the MIME
 423          * @return mixed True if the file is verified, an array otherwise
 424          */
 425         protected function verifyMimeType( $mime ) {
 426                 global $wgVerifyMimeType, $wgVerifyMimeTypeIE;
 427                 if ( $wgVerifyMimeType ) {
 428                         wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>\n" );
 429                         global $wgMimeTypeBlacklist;
 430                         if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
 431                                 return [ 'filetype-badmime', $mime ];
 432                         }
 433
 434                         if ( $wgVerifyMimeTypeIE ) {
 435                                 # Check what Internet Explorer would detect
 436                                 $fp = fopen( $this->mTempPath, 'rb' );
 437                                 $chunk = fread( $fp, 256 );
 438                                 fclose( $fp );
 439
 440                                 $magic = MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer();
 441                                 $extMime = $magic->guessTypesForExtension( $this->mFinalExtension );
 442                                 $ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime );
 443                                 foreach ( $ieTypes as $ieType ) {
 444                                         if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
 445                                                 return [ 'filetype-bad-ie-mime', $ieType ];
 446                                         }
 447                                 }
 448                         }
 449                 }
 450
 451                 return true;
 452         }
 453
 454         /**
 455          * Verifies that it's ok to include the uploaded file
 456          *
 457          * @return mixed True of the file is verified, array otherwise.
 458          */
 459         protected function verifyFile() {
 460                 global $wgVerifyMimeType, $wgDisableUploadScriptChecks;
 461
 462                 $status = $this->verifyPartialFile();
 463                 if ( $status !== true ) {
 464                         return $status;
 465                 }
 466
 467                 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer() );
 468                 $this->mFileProps = $mwProps->getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 469                 $mime = $this->mFileProps['mime'];
 470
 471                 if ( $wgVerifyMimeType ) {
 472                         # XXX: Missing extension will be caught by validateName() via getTitle()
 473                         if ( $this->mFinalExtension != '' && !$this->verifyExtension( $mime, $this->mFinalExtension ) ) {
 474                                 return [ 'filetype-mime-mismatch', $this->mFinalExtension, $mime ];
 475                         }
 476                 }
 477
 478                 # check for htmlish code and javascript
 479                 if ( !$wgDisableUploadScriptChecks ) {
 480                         if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 481                                 $svgStatus = $this->detectScriptInSvg( $this->mTempPath, false );
 482                                 if ( $svgStatus !== false ) {
 483                                         return $svgStatus;
 484                                 }
 485                         }
 486                 }
 487
 488                 $handler = MediaHandler::getHandler( $mime );
 489                 if ( $handler ) {
 490                         $handlerStatus = $handler->verifyUpload( $this->mTempPath );
 491                         if ( !$handlerStatus->isOK() ) {
 492                                 $errors = $handlerStatus->getErrorsArray();
 493
 494                                 return reset( $errors );
 495                         }
 496                 }
 497
 498                 $error = true;
 499                 Hooks::run( 'UploadVerifyFile', [ $this, $mime, &$error ] );
 500                 if ( $error !== true ) {
 501                         if ( !is_array( $error ) ) {
 502                                 $error = [ $error ];
 503                         }
 504                         return $error;
 505                 }
 506
 507                 wfDebug( __METHOD__ . ": all clear; passing.\n" );
 508
 509                 return true;
 510         }
 511
 512         /**
 513          * A verification routine suitable for partial files
 514          *
 515          * Runs the blacklist checks, but not any checks that may
 516          * assume the entire file is present.
 517          *
 518          * @return mixed True for valid or array with error message key.
 519          */
 520         protected function verifyPartialFile() {
 521                 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
 522
 523                 # getTitle() sets some internal parameters like $this->mFinalExtension
 524                 $this->getTitle();
 525
 526                 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer() );
 527                 $this->mFileProps = $mwProps->getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 528
 529                 # check MIME type, if desired
 530                 $mime = $this->mFileProps['file-mime'];
 531                 $status = $this->verifyMimeType( $mime );
 532                 if ( $status !== true ) {
 533                         return $status;
 534                 }
 535
 536                 # check for htmlish code and javascript
 537                 if ( !$wgDisableUploadScriptChecks ) {
 538                         if ( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
 539                                 return [ 'uploadscripted' ];
 540                         }
 541                         if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 542                                 $svgStatus = $this->detectScriptInSvg( $this->mTempPath, true );
 543                                 if ( $svgStatus !== false ) {
 544                                         return $svgStatus;
 545                                 }
 546                         }
 547                 }
 548
 549                 # Check for Java applets, which if uploaded can bypass cross-site
 550                 # restrictions.
 551                 if ( !$wgAllowJavaUploads ) {
 552                         $this->mJavaDetected = false;
 553                         $zipStatus = ZipDirectoryReader::read( $this->mTempPath,
 554                                 [ $this, 'zipEntryCallback' ] );
 555                         if ( !$zipStatus->isOK() ) {
 556                                 $errors = $zipStatus->getErrorsArray();
 557                                 $error = reset( $errors );
 558                                 if ( $error[0] !== 'zip-wrong-format' ) {
 559                                         return $error;
 560                                 }
 561                         }
 562                         if ( $this->mJavaDetected ) {
 563                                 return [ 'uploadjava' ];
 564                         }
 565                 }
 566
 567                 # Scan the uploaded file for viruses
 568                 $virus = $this->detectVirus( $this->mTempPath );
 569                 if ( $virus ) {
 570                         return [ 'uploadvirus', $virus ];
 571                 }
 572
 573                 return true;
 574         }
 575
 576         /**
 577          * Callback for ZipDirectoryReader to detect Java class files.
 578          *
 579          * @param array $entry
 580          */
 581         public function zipEntryCallback( $entry ) {
 582                 $names = [ $entry['name'] ];
 583
 584                 // If there is a null character, cut off the name at it, because JDK's
 585                 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
 586                 // were constructed which had ".class\0" followed by a string chosen to
 587                 // make the hash collide with the truncated name, that file could be
 588                 // returned in response to a request for the .class file.
 589                 $nullPos = strpos( $entry['name'], "\000" );
 590                 if ( $nullPos !== false ) {
 591                         $names[] = substr( $entry['name'], 0, $nullPos );
 592                 }
 593
 594                 // If there is a trailing slash in the file name, we have to strip it,
 595                 // because that's what ZIP_GetEntry() does.
 596                 if ( preg_grep( '!\.class/?$!', $names ) ) {
 597                         $this->mJavaDetected = true;
 598                 }
 599         }
 600
 601         /**
 602          * Alias for verifyTitlePermissions. The function was originally
 603          * 'verifyPermissions', but that suggests it's checking the user, when it's
 604          * really checking the title + user combination.
 605          *
 606          * @param User $user User object to verify the permissions against
 607          * @return mixed An array as returned by getUserPermissionsErrors or true
 608          *   in case the user has proper permissions.
 609          */
 610         public function verifyPermissions( $user ) {
 611                 return $this->verifyTitlePermissions( $user );
 612         }
 613
 614         /**
 615          * Check whether the user can edit, upload and create the image. This
 616          * checks only against the current title; if it returns errors, it may
 617          * very well be that another title will not give errors. Therefore
 618          * isAllowed() should be called as well for generic is-user-blocked or
 619          * can-user-upload checking.
 620          *
 621          * @param User $user User object to verify the permissions against
 622          * @return mixed An array as returned by getUserPermissionsErrors or true
 623          *   in case the user has proper permissions.
 624          */
 625         public function verifyTitlePermissions( $user ) {
 626                 /**
 627                  * If the image is protected, non-sysop users won't be able
 628                  * to modify it by uploading a new revision.
 629                  */
 630                 $nt = $this->getTitle();
 631                 if ( is_null( $nt ) ) {
 632                         return true;
 633                 }
 634                 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
 635                 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
 636                 if ( !$nt->exists() ) {
 637                         $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
 638                 } else {
 639                         $permErrorsCreate = [];
 640                 }
 641                 if ( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
 642                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
 643                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
 644
 645                         return $permErrors;
 646                 }
 647
 648                 $overwriteError = $this->checkOverwrite( $user );
 649                 if ( $overwriteError !== true ) {
 650                         return [ $overwriteError ];
 651                 }
 652
 653                 return true;
 654         }
 655
 656         /**
 657          * Check for non fatal problems with the file.
 658          *
 659          * This should not assume that mTempPath is set.
 660          *
 661          * @return mixed[] Array of warnings
 662          */
 663         public function checkWarnings() {
 664                 $warnings = [];
 665
 666                 $localFile = $this->getLocalFile();
 667                 $localFile->load( File::READ_LATEST );
 668                 $filename = $localFile->getName();
 669                 $hash = $this->getTempFileSha1Base36();
 670
 671                 $badFileName = $this->checkBadFileName( $filename, $this->mDesiredDestName );
 672                 if ( $badFileName !== null ) {
 673                         $warnings['badfilename'] = $badFileName;
 674                 }
 675
 676                 $unwantedFileExtensionDetails = $this->checkUnwantedFileExtensions( $this->mFinalExtension );
 677                 if ( $unwantedFileExtensionDetails !== null ) {
 678                         $warnings['filetype-unwanted-type'] = $unwantedFileExtensionDetails;
 679                 }
 680
 681                 $fileSizeWarnings = $this->checkFileSize( $this->mFileSize );
 682                 if ( $fileSizeWarnings ) {
 683                         $warnings = array_merge( $warnings, $fileSizeWarnings );
 684                 }
 685
 686                 $localFileExistsWarnings = $this->checkLocalFileExists( $localFile, $hash );
 687                 if ( $localFileExistsWarnings ) {
 688                         $warnings = array_merge( $warnings, $localFileExistsWarnings );
 689                 }
 690
 691                 if ( $this->checkLocalFileWasDeleted( $localFile ) ) {
 692                         $warnings['was-deleted'] = $filename;
 693                 }
 694
 695                 // If a file with the same name exists locally then the local file has already been tested
 696                 // for duplication of content
 697                 $ignoreLocalDupes = isset( $warnings[ 'exists '] );
 698                 $dupes = $this->checkAgainstExistingDupes( $hash, $ignoreLocalDupes );
 699                 if ( $dupes ) {
 700                         $warnings['duplicate'] = $dupes;
 701                 }
 702
 703                 $archivedDupes = $this->checkAgainstArchiveDupes( $hash );
 704                 if ( $archivedDupes !== null ) {
 705                         $warnings['duplicate-archive'] = $archivedDupes;
 706                 }
 707
 708                 return $warnings;
 709         }
 710
 711         /**
 712          * Convert the warnings array returned by checkWarnings() to something that
 713          * can be serialized. File objects will be converted to an associative array
 714          * with the following keys:
 715          *
 716          *   - fileName: The name of the file
 717          *   - timestamp: The upload timestamp
 718          *
 719          * @param mixed[] $warnings
 720          * @return mixed[]
 721          */
 722         public static function makeWarningsSerializable( $warnings ) {
 723                 array_walk_recursive( $warnings, function ( &$param, $key ) {
 724                         if ( $param instanceof File ) {
 725                                 $param = [
 726                                         'fileName' => $param->getName(),
 727                                         'timestamp' => $param->getTimestamp()
 728                                 ];
 729                         } elseif ( is_object( $param ) ) {
 730                                 throw new InvalidArgumentException(
 731                                         'UploadBase::makeWarningsSerializable: ' .
 732                                         'Unexpected object of class ' . get_class( $param ) );
 733                         }
 734                 } );
 735                 return $warnings;
 736         }
 737
 738         /**
 739          * Check whether the resulting filename is different from the desired one,
 740          * but ignore things like ucfirst() and spaces/underscore things
 741          *
 742          * @param string $filename
 743          * @param string $desiredFileName
 744          *
 745          * @return string|null String that was determined to be bad or null if the filename is okay
 746          */
 747         private function checkBadFileName( $filename, $desiredFileName ) {
 748                 $comparableName = str_replace( ' ', '_', $desiredFileName );
 749                 $comparableName = Title::capitalize( $comparableName, NS_FILE );
 750
 751                 if ( $desiredFileName != $filename && $comparableName != $filename ) {
 752                         return $filename;
 753                 }
 754
 755                 return null;
 756         }
 757
 758         /**
 759          * @param string $fileExtension The file extension to check
 760          *
 761          * @return array|null array with the following keys:
 762          *                    0 => string The final extension being used
 763          *                    1 => string[] The extensions that are allowed
 764          *                    2 => int The number of extensions that are allowed.
 765          */
 766         private function checkUnwantedFileExtensions( $fileExtension ) {
 767                 global $wgCheckFileExtensions, $wgFileExtensions, $wgLang;
 768
 769                 if ( $wgCheckFileExtensions ) {
 770                         $extensions = array_unique( $wgFileExtensions );
 771                         if ( !$this->checkFileExtension( $fileExtension, $extensions ) ) {
 772                                 return [
 773                                         $fileExtension,
 774                                         $wgLang->commaList( $extensions ),
 775                                         count( $extensions )
 776                                 ];
 777                         }
 778                 }
 779
 780                 return null;
 781         }
 782
 783         /**
 784          * @param int $fileSize
 785          *
 786          * @return array warnings
 787          */
 788         private function checkFileSize( $fileSize ) {
 789                 global $wgUploadSizeWarning;
 790
 791                 $warnings = [];
 792
 793                 if ( $wgUploadSizeWarning && ( $fileSize > $wgUploadSizeWarning ) ) {
 794                         $warnings['large-file'] = [ $wgUploadSizeWarning, $fileSize ];
 795                 }
 796
 797                 if ( $fileSize == 0 ) {
 798                         $warnings['empty-file'] = true;
 799                 }
 800
 801                 return $warnings;
 802         }
 803
 804         /**
 805          * @param LocalFile $localFile
 806          * @param string $hash sha1 hash of the file to check
 807          *
 808          * @return array warnings
 809          */
 810         private function checkLocalFileExists( LocalFile $localFile, $hash ) {
 811                 $warnings = [];
 812
 813                 $exists = self::getExistsWarning( $localFile );
 814                 if ( $exists !== false ) {
 815                         $warnings['exists'] = $exists;
 816
 817                         // check if file is an exact duplicate of current file version
 818                         if ( $hash === $localFile->getSha1() ) {
 819                                 $warnings['no-change'] = $localFile;
 820                         }
 821
 822                         // check if file is an exact duplicate of older versions of this file
 823                         $history = $localFile->getHistory();
 824                         foreach ( $history as $oldFile ) {
 825                                 if ( $hash === $oldFile->getSha1() ) {
 826                                         $warnings['duplicate-version'][] = $oldFile;
 827                                 }
 828                         }
 829                 }
 830
 831                 return $warnings;
 832         }
 833
 834         private function checkLocalFileWasDeleted( LocalFile $localFile ) {
 835                 return $localFile->wasDeleted() && !$localFile->exists();
 836         }
 837
 838         /**
 839          * @param string $hash sha1 hash of the file to check
 840          * @param bool $ignoreLocalDupes True to ignore local duplicates
 841          *
 842          * @return File[] Duplicate files, if found.
 843          */
 844         private function checkAgainstExistingDupes( $hash, $ignoreLocalDupes ) {
 845                 $dupes = RepoGroup::singleton()->findBySha1( $hash );
 846                 $title = $this->getTitle();
 847                 foreach ( $dupes as $key => $dupe ) {
 848                         if (
 849                                 ( $dupe instanceof LocalFile ) &&
 850                                 $ignoreLocalDupes &&
 851                                 $title->equals( $dupe->getTitle() )
 852                         ) {
 853                                 unset( $dupes[$key] );
 854                         }
 855                 }
 856
 857                 return $dupes;
 858         }
 859
 860         /**
 861          * @param string $hash sha1 hash of the file to check
 862          *
 863          * @return string|null Name of the dupe or empty string if discovered (depending on visibility)
 864          *                     null if the check discovered no dupes.
 865          */
 866         private function checkAgainstArchiveDupes( $hash ) {
 867                 $archivedFile = new ArchivedFile( null, 0, '', $hash );
 868                 if ( $archivedFile->getID() > 0 ) {
 869                         if ( $archivedFile->userCan( File::DELETED_FILE ) ) {
 870                                 return $archivedFile->getName();
 871                         } else {
 872                                 return '';
 873                         }
 874                 }
 875
 876                 return null;
 877         }
 878
 879         /**
 880          * Really perform the upload. Stores the file in the local repo, watches
 881          * if necessary and runs the UploadComplete hook.
 882          *
 883          * @param string $comment
 884          * @param string $pageText
 885          * @param bool $watch Whether the file page should be added to user's watchlist.
 886          *   (This doesn't check $user's permissions.)
 887          * @param User $user
 888          * @param string[] $tags Change tags to add to the log entry and page revision.
 889          *   (This doesn't check $user's permissions.)
 890          * @return Status Indicating the whether the upload succeeded.
 891          */
 892         public function performUpload( $comment, $pageText, $watch, $user, $tags = [] ) {
 893                 $this->getLocalFile()->load( File::READ_LATEST );
 894                 $props = $this->mFileProps;
 895
 896                 $error = null;
 897                 Hooks::run( 'UploadVerifyUpload', [ $this, $user, $props, $comment, $pageText, &$error ] );
 898                 if ( $error ) {
 899                         if ( !is_array( $error ) ) {
 900                                 $error = [ $error ];
 901                         }
 902                         return Status::newFatal( ...$error );
 903                 }
 904
 905                 $status = $this->getLocalFile()->upload(
 906                         $this->mTempPath,
 907                         $comment,
 908                         $pageText,
 909                         File::DELETE_SOURCE,
 910                         $props,
 911                         false,
 912                         $user,
 913                         $tags
 914                 );
 915
 916                 if ( $status->isGood() ) {
 917                         if ( $watch ) {
 918                                 WatchAction::doWatch(
 919                                         $this->getLocalFile()->getTitle(),
 920                                         $user,
 921                                         User::IGNORE_USER_RIGHTS
 922                                 );
 923                         }
 924                         // Avoid PHP 7.1 warning of passing $this by reference
 925                         $uploadBase = $this;
 926                         Hooks::run( 'UploadComplete', [ &$uploadBase ] );
 927
 928                         $this->postProcessUpload();
 929                 }
 930
 931                 return $status;
 932         }
 933
 934         /**
 935          * Perform extra steps after a successful upload.
 936          *
 937          * @since  1.25
 938          */
 939         public function postProcessUpload() {
 940         }
 941
 942         /**
 943          * Returns the title of the file to be uploaded. Sets mTitleError in case
 944          * the name was illegal.
 945          *
 946          * @return Title|null The title of the file or null in case the name was illegal
 947          */
 948         public function getTitle() {
 949                 if ( $this->mTitle !== false ) {
 950                         return $this->mTitle;
 951                 }
 952                 if ( !is_string( $this->mDesiredDestName ) ) {
 953                         $this->mTitleError = self::ILLEGAL_FILENAME;
 954                         $this->mTitle = null;
 955
 956                         return $this->mTitle;
 957                 }
 958                 /* Assume that if a user specified File:Something.jpg, this is an error
 959                  * and that the namespace prefix needs to be stripped of.
 960                  */
 961                 $title = Title::newFromText( $this->mDesiredDestName );
 962                 if ( $title && $title->getNamespace() == NS_FILE ) {
 963                         $this->mFilteredName = $title->getDBkey();
 964                 } else {
 965                         $this->mFilteredName = $this->mDesiredDestName;
 966                 }
 967
 968                 # oi_archive_name is max 255 bytes, which include a timestamp and an
 969                 # exclamation mark, so restrict file name to 240 bytes.
 970                 if ( strlen( $this->mFilteredName ) > 240 ) {
 971                         $this->mTitleError = self::FILENAME_TOO_LONG;
 972                         $this->mTitle = null;
 973
 974                         return $this->mTitle;
 975                 }
 976
 977                 /**
 978                  * Chop off any directories in the given filename. Then
 979                  * filter out illegal characters, and try to make a legible name
 980                  * out of it. We'll strip some silently that Title would die on.
 981                  */
 982                 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mFilteredName );
 983                 /* Normalize to title form before we do any further processing */
 984                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 985                 if ( is_null( $nt ) ) {
 986                         $this->mTitleError = self::ILLEGAL_FILENAME;
 987                         $this->mTitle = null;
 988
 989                         return $this->mTitle;
 990                 }
 991                 $this->mFilteredName = $nt->getDBkey();
 992
 993                 /**
 994                  * We'll want to blacklist against *any* 'extension', and use
 995                  * only the final one for the whitelist.
 996                  */
 997                 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
 998
 999                 if ( $ext !== [] ) {
1000                         $this->mFinalExtension = trim( end( $ext ) );
1001                 } else {
1002                         $this->mFinalExtension = '';
1003
1004                         # No extension, try guessing one
1005                         $magic = MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer();
1006                         $mime = $magic->guessMimeType( $this->mTempPath );
1007                         if ( $mime !== 'unknown/unknown' ) {
1008                                 # Get a space separated list of extensions
1009                                 $extList = $magic->getExtensionsForType( $mime );
1010                                 if ( $extList ) {
1011                                         # Set the extension to the canonical extension
1012                                         $this->mFinalExtension = strtok( $extList, ' ' );
1013
1014                                         # Fix up the other variables
1015                                         $this->mFilteredName .= ".{$this->mFinalExtension}";
1016                                         $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
1017                                         $ext = [ $this->mFinalExtension ];
1018                                 }
1019                         }
1020                 }
1021
1022                 /* Don't allow users to override the blacklist (check file extension) */
1023                 global $wgCheckFileExtensions, $wgStrictFileExtensions;
1024                 global $wgFileExtensions, $wgFileBlacklist;
1025
1026                 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
1027
1028                 if ( $this->mFinalExtension == '' ) {
1029                         $this->mTitleError = self::FILETYPE_MISSING;
1030                         $this->mTitle = null;
1031
1032                         return $this->mTitle;
1033                 } elseif ( $blackListedExtensions ||
1034                         ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
1035                                 !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) )
1036                 ) {
1037                         $this->mBlackListedExtensions = $blackListedExtensions;
1038                         $this->mTitleError = self::FILETYPE_BADTYPE;
1039                         $this->mTitle = null;
1040
1041                         return $this->mTitle;
1042                 }
1043
1044                 // Windows may be broken with special characters, see T3780
1045                 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
1046                         && !RepoGroup::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
1047                 ) {
1048                         $this->mTitleError = self::WINDOWS_NONASCII_FILENAME;
1049                         $this->mTitle = null;
1050
1051                         return $this->mTitle;
1052                 }
1053
1054                 # If there was more than one "extension", reassemble the base
1055                 # filename to prevent bogus complaints about length
1056                 if ( count( $ext ) > 1 ) {
1057                         $iterations = count( $ext ) - 1;
1058                         for ( $i = 0; $i < $iterations; $i++ ) {
1059                                 $partname .= '.' . $ext[$i];
1060                         }
1061                 }
1062
1063                 if ( strlen( $partname ) < 1 ) {
1064                         $this->mTitleError = self::MIN_LENGTH_PARTNAME;
1065                         $this->mTitle = null;
1066
1067                         return $this->mTitle;
1068                 }
1069
1070                 $this->mTitle = $nt;
1071
1072                 return $this->mTitle;
1073         }
1074
1075         /**
1076          * Return the local file and initializes if necessary.
1077          *
1078          * @return LocalFile|null
1079          */
1080         public function getLocalFile() {
1081                 if ( is_null( $this->mLocalFile ) ) {
1082                         $nt = $this->getTitle();
1083                         $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
1084                 }
1085
1086                 return $this->mLocalFile;
1087         }
1088
1089         /**
1090          * @return UploadStashFile|null
1091          */
1092         public function getStashFile() {
1093                 return $this->mStashFile;
1094         }
1095
1096         /**
1097          * Like stashFile(), but respects extensions' wishes to prevent the stashing. verifyUpload() must
1098          * be called before calling this method (unless $isPartial is true).
1099          *
1100          * Upload stash exceptions are also caught and converted to an error status.
1101          *
1102          * @since 1.28
1103          * @param User $user
1104          * @param bool $isPartial Pass `true` if this is a part of a chunked upload (not a complete file).
1105          * @return Status If successful, value is an UploadStashFile instance
1106          */
1107         public function tryStashFile( User $user, $isPartial = false ) {
1108                 if ( !$isPartial ) {
1109                         $error = $this->runUploadStashFileHook( $user );
1110                         if ( $error ) {
1111                                 return Status::newFatal( ...$error );
1112                         }
1113                 }
1114                 try {
1115                         $file = $this->doStashFile( $user );
1116                         return Status::newGood( $file );
1117                 } catch ( UploadStashException $e ) {
1118                         return Status::newFatal( 'uploadstash-exception', get_class( $e ), $e->getMessage() );
1119                 }
1120         }
1121
1122         /**
1123          * @param User $user
1124          * @return array|null Error message and parameters, null if there's no error
1125          */
1126         protected function runUploadStashFileHook( User $user ) {
1127                 $props = $this->mFileProps;
1128                 $error = null;
1129                 Hooks::run( 'UploadStashFile', [ $this, $user, $props, &$error ] );
1130                 if ( $error && !is_array( $error ) ) {
1131                         $error = [ $error ];
1132                 }
1133                 return $error;
1134         }
1135
1136         /**
1137          * If the user does not supply all necessary information in the first upload
1138          * form submission (either by accident or by design) then we may want to
1139          * stash the file temporarily, get more information, and publish the file
1140          * later.
1141          *
1142          * This method will stash a file in a temporary directory for later
1143          * processing, and save the necessary descriptive info into the database.
1144          * This method returns the file object, which also has a 'fileKey' property
1145          * which can be passed through a form or API request to find this stashed
1146          * file again.
1147          *
1148          * @deprecated since 1.28 Use tryStashFile() instead
1149          * @param User|null $user
1150          * @return UploadStashFile Stashed file
1151          * @throws UploadStashBadPathException
1152          * @throws UploadStashFileException
1153          * @throws UploadStashNotLoggedInException
1154          */
1155         public function stashFile( User $user = null ) {
1156                 wfDeprecated( __METHOD__, '1.28' );
1157
1158                 return $this->doStashFile( $user );
1159         }
1160
1161         /**
1162          * Implementation for stashFile() and tryStashFile().
1163          *
1164          * @param User|null $user
1165          * @return UploadStashFile Stashed file
1166          */
1167         protected function doStashFile( User $user = null ) {
1168                 $stash = RepoGroup::singleton()->getLocalRepo()->getUploadStash( $user );
1169                 $file = $stash->stashFile( $this->mTempPath, $this->getSourceType() );
1170                 $this->mStashFile = $file;
1171
1172                 return $file;
1173         }
1174
1175         /**
1176          * If we've modified the upload file we need to manually remove it
1177          * on exit to clean up.
1178          */
1179         public function cleanupTempFile() {
1180                 if ( $this->mRemoveTempFile && $this->tempFileObj ) {
1181                         // Delete when all relevant TempFSFile handles go out of scope
1182                         wfDebug( __METHOD__ . ": Marked temporary file '{$this->mTempPath}' for removal\n" );
1183                         $this->tempFileObj->autocollect();
1184                 }
1185         }
1186
1187         public function getTempPath() {
1188                 return $this->mTempPath;
1189         }
1190
1191         /**
1192          * Split a file into a base name and all dot-delimited 'extensions'
1193          * on the end. Some web server configurations will fall back to
1194          * earlier pseudo-'extensions' to determine type and execute
1195          * scripts, so the blacklist needs to check them all.
1196          *
1197          * @param string $filename
1198          * @return array [ string, string[] ]
1199          */
1200         public static function splitExtensions( $filename ) {
1201                 $bits = explode( '.', $filename );
1202                 $basename = array_shift( $bits );
1203
1204                 return [ $basename, $bits ];
1205         }
1206
1207         /**
1208          * Perform case-insensitive match against a list of file extensions.
1209          * Returns true if the extension is in the list.
1210          *
1211          * @param string $ext
1212          * @param array $list
1213          * @return bool
1214          */
1215         public static function checkFileExtension( $ext, $list ) {
1216                 return in_array( strtolower( $ext ), $list );
1217         }
1218
1219         /**
1220          * Perform case-insensitive match against a list of file extensions.
1221          * Returns an array of matching extensions.
1222          *
1223          * @param string[] $ext
1224          * @param string[] $list
1225          * @return string[]
1226          */
1227         public static function checkFileExtensionList( $ext, $list ) {
1228                 return array_intersect( array_map( 'strtolower', $ext ), $list );
1229         }
1230
1231         /**
1232          * Checks if the MIME type of the uploaded file matches the file extension.
1233          *
1234          * @param string $mime The MIME type of the uploaded file
1235          * @param string $extension The filename extension that the file is to be served with
1236          * @return bool
1237          */
1238         public static function verifyExtension( $mime, $extension ) {
1239                 $magic = MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer();
1240
1241                 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' ) {
1242                         if ( !$magic->isRecognizableExtension( $extension ) ) {
1243                                 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
1244                                         "unrecognized extension '$extension', can't verify\n" );
1245
1246                                 return true;
1247                         } else {
1248                                 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; " .
1249                                         "recognized extension '$extension', so probably invalid file\n" );
1250
1251                                 return false;
1252                         }
1253                 }
1254
1255                 $match = $magic->isMatchingExtension( $extension, $mime );
1256
1257                 if ( $match === null ) {
1258                         if ( $magic->getTypesForExtension( $extension ) !== null ) {
1259                                 wfDebug( __METHOD__ . ": No extension known for $mime, but we know a mime for $extension\n" );
1260
1261                                 return false;
1262                         } else {
1263                                 wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
1264
1265                                 return true;
1266                         }
1267                 } elseif ( $match === true ) {
1268                         wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
1269
1270                         /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1271                         return true;
1272                 } else {
1273                         wfDebug( __METHOD__
1274                                 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1275
1276                         return false;
1277                 }
1278         }
1279
1280         /**
1281          * Heuristic for detecting files that *could* contain JavaScript instructions or
1282          * things that may look like HTML to a browser and are thus
1283          * potentially harmful. The present implementation will produce false
1284          * positives in some situations.
1285          *
1286          * @param string $file Pathname to the temporary upload file
1287          * @param string $mime The MIME type of the file
1288          * @param string $extension The extension of the file
1289          * @return bool True if the file contains something looking like embedded scripts
1290          */
1291         public static function detectScript( $file, $mime, $extension ) {
1292                 # ugly hack: for text files, always look at the entire file.
1293                 # For binary field, just check the first K.
1294
1295                 $isText = strpos( $mime, 'text/' ) === 0;
1296                 if ( $isText ) {
1297                         $chunk = file_get_contents( $file );
1298                 } else {
1299                         $fp = fopen( $file, 'rb' );
1300                         $chunk = fread( $fp, 1024 );
1301                         fclose( $fp );
1302                 }
1303
1304                 $chunk = strtolower( $chunk );
1305
1306                 if ( !$chunk ) {
1307                         return false;
1308                 }
1309
1310                 # decode from UTF-16 if needed (could be used for obfuscation).
1311                 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1312                         $enc = 'UTF-16BE';
1313                 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1314                         $enc = 'UTF-16LE';
1315                 } else {
1316                         $enc = null;
1317                 }
1318
1319                 if ( $enc !== null ) {
1320                         $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1321                 }
1322
1323                 $chunk = trim( $chunk );
1324
1325                 /** @todo FIXME: Convert from UTF-16 if necessary! */
1326                 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
1327
1328                 # check for HTML doctype
1329                 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1330                         return true;
1331                 }
1332
1333                 // Some browsers will interpret obscure xml encodings as UTF-8, while
1334                 // PHP/expat will interpret the given encoding in the xml declaration (T49304)
1335                 if ( $extension == 'svg' || strpos( $mime, 'image/svg' ) === 0 ) {
1336                         if ( self::checkXMLEncodingMissmatch( $file ) ) {
1337                                 return true;
1338                         }
1339                 }
1340
1341                 // Quick check for HTML heuristics in old IE and Safari.
1342                 //
1343                 // The exact heuristics IE uses are checked separately via verifyMimeType(), so we
1344                 // don't need them all here as it can cause many false positives.
1345                 //
1346                 // Check for `<script` and such still to forbid script tags and embedded HTML in SVG:
1347                 $tags = [
1348                         '<body',
1349                         '<head',
1350                         '<html', # also in safari
1351                         '<script', # also in safari
1352                 ];
1353
1354                 foreach ( $tags as $tag ) {
1355                         if ( strpos( $chunk, $tag ) !== false ) {
1356                                 wfDebug( __METHOD__ . ": found something that may make it be mistaken for html: $tag\n" );
1357
1358                                 return true;
1359                         }
1360                 }
1361
1362                 /*
1363                  * look for JavaScript
1364                  */
1365
1366                 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1367                 $chunk = Sanitizer::decodeCharReferences( $chunk );
1368
1369                 # look for script-types
1370                 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1371                         wfDebug( __METHOD__ . ": found script types\n" );
1372
1373                         return true;
1374                 }
1375
1376                 # look for html-style script-urls
1377                 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1378                         wfDebug( __METHOD__ . ": found html-style script urls\n" );
1379
1380                         return true;
1381                 }
1382
1383                 # look for css-style script-urls
1384                 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1385                         wfDebug( __METHOD__ . ": found css-style script urls\n" );
1386
1387                         return true;
1388                 }
1389
1390                 wfDebug( __METHOD__ . ": no scripts found\n" );
1391
1392                 return false;
1393         }
1394
1395         /**
1396          * Check a whitelist of xml encodings that are known not to be interpreted differently
1397          * by the server's xml parser (expat) and some common browsers.
1398          *
1399          * @param string $file Pathname to the temporary upload file
1400          * @return bool True if the file contains an encoding that could be misinterpreted
1401          */
1402         public static function checkXMLEncodingMissmatch( $file ) {
1403                 global $wgSVGMetadataCutoff;
1404                 $contents = file_get_contents( $file, false, null, 0, $wgSVGMetadataCutoff );
1405                 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1406
1407                 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1408                         if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1409                                 && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1410                         ) {
1411                                 wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1412
1413                                 return true;
1414                         }
1415                 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1416                         // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1417                         // bytes. There shouldn't be a legitimate reason for this to happen.
1418                         wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1419
1420                         return true;
1421                 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1422                         // EBCDIC encoded XML
1423                         wfDebug( __METHOD__ . ": EBCDIC Encoded XML\n" );
1424
1425                         return true;
1426                 }
1427
1428                 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1429                 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1430                 $attemptEncodings = [ 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' ];
1431                 foreach ( $attemptEncodings as $encoding ) {
1432                         Wikimedia\suppressWarnings();
1433                         $str = iconv( $encoding, 'UTF-8', $contents );
1434                         Wikimedia\restoreWarnings();
1435                         if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1436                                 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1437                                         && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1438                                 ) {
1439                                         wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1440
1441                                         return true;
1442                                 }
1443                         } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1444                                 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1445                                 // bytes. There shouldn't be a legitimate reason for this to happen.
1446                                 wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1447
1448                                 return true;
1449                         }
1450                 }
1451
1452                 return false;
1453         }
1454
1455         /**
1456          * @param string $filename
1457          * @param bool $partial
1458          * @return bool|array
1459          */
1460         protected function detectScriptInSvg( $filename, $partial ) {
1461                 $this->mSVGNSError = false;
1462                 $check = new XmlTypeCheck(
1463                         $filename,
1464                         [ $this, 'checkSvgScriptCallback' ],
1465                         true,
1466                         [
1467                                 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback',
1468                                 'external_dtd_handler' => 'UploadBase::checkSvgExternalDTD',
1469                         ]
1470                 );
1471                 if ( $check->wellFormed !== true ) {
1472                         // Invalid xml (T60553)
1473                         // But only when non-partial (T67724)
1474                         return $partial ? false : [ 'uploadinvalidxml' ];
1475                 } elseif ( $check->filterMatch ) {
1476                         if ( $this->mSVGNSError ) {
1477                                 return [ 'uploadscriptednamespace', $this->mSVGNSError ];
1478                         }
1479
1480                         return $check->filterMatchType;
1481                 }
1482
1483                 return false;
1484         }
1485
1486         /**
1487          * Callback to filter SVG Processing Instructions.
1488          * @param string $target Processing instruction name
1489          * @param string $data Processing instruction attribute and value
1490          * @return bool|array
1491          */
1492         public static function checkSvgPICallback( $target, $data ) {
1493                 // Don't allow external stylesheets (T59550)
1494                 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1495                         return [ 'upload-scripted-pi-callback' ];
1496                 }
1497
1498                 return false;
1499         }
1500
1501         /**
1502          * Verify that DTD urls referenced are only the standard dtds
1503          *
1504          * Browsers seem to ignore external dtds. However just to be on the
1505          * safe side, only allow dtds from the svg standard.
1506          *
1507          * @param string $type PUBLIC or SYSTEM
1508          * @param string $publicId The well-known public identifier for the dtd
1509          * @param string $systemId The url for the external dtd
1510          * @return bool|array
1511          */
1512         public static function checkSvgExternalDTD( $type, $publicId, $systemId ) {
1513                 // This doesn't include the XHTML+MathML+SVG doctype since we don't
1514                 // allow XHTML anyways.
1515                 $allowedDTDs = [
1516                         'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd',
1517                         'http://www.w3.org/TR/2001/REC-SVG-20010904/DTD/svg10.dtd',
1518                         'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-basic.dtd',
1519                         'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-tiny.dtd',
1520                         // https://phabricator.wikimedia.org/T168856
1521                         'http://www.w3.org/TR/2001/PR-SVG-20010719/DTD/svg10.dtd',
1522                 ];
1523                 if ( $type !== 'PUBLIC'
1524                         || !in_array( $systemId, $allowedDTDs )
1525                         || strpos( $publicId, "-//W3C//" ) !== 0
1526                 ) {
1527                         return [ 'upload-scripted-dtd' ];
1528                 }
1529                 return false;
1530         }
1531
1532         /**
1533          * @todo Replace this with a whitelist filter!
1534          * @param string $element
1535          * @param array $attribs
1536          * @param string|null $data
1537          * @return bool|array
1538          */
1539         public function checkSvgScriptCallback( $element, $attribs, $data = null ) {
1540                 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1541
1542                 // We specifically don't include:
1543                 // http://www.w3.org/1999/xhtml (T62771)
1544                 static $validNamespaces = [
1545                         '',
1546                         'adobe:ns:meta/',
1547                         'http://creativecommons.org/ns#',
1548                         'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1549                         'http://ns.adobe.com/adobeillustrator/10.0/',
1550                         'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1551                         'http://ns.adobe.com/extensibility/1.0/',
1552                         'http://ns.adobe.com/flows/1.0/',
1553                         'http://ns.adobe.com/illustrator/1.0/',
1554                         'http://ns.adobe.com/imagereplacement/1.0/',
1555                         'http://ns.adobe.com/pdf/1.3/',
1556                         'http://ns.adobe.com/photoshop/1.0/',
1557                         'http://ns.adobe.com/saveforweb/1.0/',
1558                         'http://ns.adobe.com/variables/1.0/',
1559                         'http://ns.adobe.com/xap/1.0/',
1560                         'http://ns.adobe.com/xap/1.0/g/',
1561                         'http://ns.adobe.com/xap/1.0/g/img/',
1562                         'http://ns.adobe.com/xap/1.0/mm/',
1563                         'http://ns.adobe.com/xap/1.0/rights/',
1564                         'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1565                         'http://ns.adobe.com/xap/1.0/stype/font#',
1566                         'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1567                         'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1568                         'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1569                         'http://ns.adobe.com/xap/1.0/t/pg/',
1570                         'http://purl.org/dc/elements/1.1/',
1571                         'http://purl.org/dc/elements/1.1',
1572                         'http://schemas.microsoft.com/visio/2003/svgextensions/',
1573                         'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1574                         'http://taptrix.com/inkpad/svg_extensions',
1575                         'http://web.resource.org/cc/',
1576                         'http://www.freesoftware.fsf.org/bkchem/cdml',
1577                         'http://www.inkscape.org/namespaces/inkscape',
1578                         'http://www.opengis.net/gml',
1579                         'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1580                         'http://www.w3.org/2000/svg',
1581                         'http://www.w3.org/tr/rec-rdf-syntax/',
1582                         'http://www.w3.org/2000/01/rdf-schema#',
1583                 ];
1584
1585                 // Inkscape mangles namespace definitions created by Adobe Illustrator.
1586                 // This is nasty but harmless. (T144827)
1587                 $isBuggyInkscape = preg_match( '/^&(#38;)*ns_[a-z_]+;$/', $namespace );
1588
1589                 if ( !( $isBuggyInkscape || in_array( $namespace, $validNamespaces ) ) ) {
1590                         wfDebug( __METHOD__ . ": Non-svg namespace '$namespace' in uploaded file.\n" );
1591                         /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1592                         $this->mSVGNSError = $namespace;
1593
1594                         return true;
1595                 }
1596
1597                 /*
1598                  * check for elements that can contain javascript
1599                  */
1600                 if ( $strippedElement == 'script' ) {
1601                         wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
1602
1603                         return [ 'uploaded-script-svg', $strippedElement ];
1604                 }
1605
1606                 # e.g., <svg xmlns="http://www.w3.org/2000/svg">
1607                 #  <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1608                 if ( $strippedElement == 'handler' ) {
1609                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1610
1611                         return [ 'uploaded-script-svg', $strippedElement ];
1612                 }
1613
1614                 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1615                 if ( $strippedElement == 'stylesheet' ) {
1616                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1617
1618                         return [ 'uploaded-script-svg', $strippedElement ];
1619                 }
1620
1621                 # Block iframes, in case they pass the namespace check
1622                 if ( $strippedElement == 'iframe' ) {
1623                         wfDebug( __METHOD__ . ": iframe in uploaded file.\n" );
1624
1625                         return [ 'uploaded-script-svg', $strippedElement ];
1626                 }
1627
1628                 # Check <style> css
1629                 if ( $strippedElement == 'style'
1630                         && self::checkCssFragment( Sanitizer::normalizeCss( $data ) )
1631                 ) {
1632                         wfDebug( __METHOD__ . ": hostile css in style element.\n" );
1633                         return [ 'uploaded-hostile-svg' ];
1634                 }
1635
1636                 foreach ( $attribs as $attrib => $value ) {
1637                         $stripped = $this->stripXmlNamespace( $attrib );
1638                         $value = strtolower( $value );
1639
1640                         if ( substr( $stripped, 0, 2 ) == 'on' ) {
1641                                 wfDebug( __METHOD__
1642                                         . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1643
1644                                 return [ 'uploaded-event-handler-on-svg', $attrib, $value ];
1645                         }
1646
1647                         # Do not allow relative links, or unsafe url schemas.
1648                         # For <a> tags, only data:, http: and https: and same-document
1649                         # fragment links are allowed. For all other tags, only data:
1650                         # and fragment are allowed.
1651                         if ( $stripped == 'href'
1652                                 && $value !== ''
1653                                 && strpos( $value, 'data:' ) !== 0
1654                                 && strpos( $value, '#' ) !== 0
1655                         ) {
1656                                 if ( !( $strippedElement === 'a'
1657                                         && preg_match( '!^https?://!i', $value ) )
1658                                 ) {
1659                                         wfDebug( __METHOD__ . ": Found href attribute <$strippedElement "
1660                                                 . "'$attrib'='$value' in uploaded file.\n" );
1661
1662                                         return [ 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value ];
1663                                 }
1664                         }
1665
1666                         # only allow data: targets that should be safe. This prevents vectors like,
1667                         # image/svg, text/xml, application/xml, and text/html, which can contain scripts
1668                         if ( $stripped == 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) {
1669                                 // rfc2397 parameters. This is only slightly slower than (;[\w;]+)*.
1670                                 // phpcs:ignore Generic.Files.LineLength
1671                                 $parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?';
1672
1673                                 if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|png)$parameters,!i", $value ) ) {
1674                                         wfDebug( __METHOD__ . ": Found href to unwhitelisted data: uri "
1675                                                 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1676                                         return [ 'uploaded-href-unsafe-target-svg', $strippedElement, $attrib, $value ];
1677                                 }
1678                         }
1679
1680                         # Change href with animate from (http://html5sec.org/#137).
1681                         if ( $stripped === 'attributename'
1682                                 && $strippedElement === 'animate'
1683                                 && $this->stripXmlNamespace( $value ) == 'href'
1684                         ) {
1685                                 wfDebug( __METHOD__ . ": Found animate that might be changing href using from "
1686                                         . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1687
1688                                 return [ 'uploaded-animate-svg', $strippedElement, $attrib, $value ];
1689                         }
1690
1691                         # use set/animate to add event-handler attribute to parent
1692                         if ( ( $strippedElement == 'set' || $strippedElement == 'animate' )
1693                                 && $stripped == 'attributename'
1694                                 && substr( $value, 0, 2 ) == 'on'
1695                         ) {
1696                                 wfDebug( __METHOD__ . ": Found svg setting event-handler attribute with "
1697                                         . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1698
1699                                 return [ 'uploaded-setting-event-handler-svg', $strippedElement, $stripped, $value ];
1700                         }
1701
1702                         # use set to add href attribute to parent element
1703                         if ( $strippedElement == 'set'
1704                                 && $stripped == 'attributename'
1705                                 && strpos( $value, 'href' ) !== false
1706                         ) {
1707                                 wfDebug( __METHOD__ . ": Found svg setting href attribute '$value' in uploaded file.\n" );
1708
1709                                 return [ 'uploaded-setting-href-svg' ];
1710                         }
1711
1712                         # use set to add a remote / data / script target to an element
1713                         if ( $strippedElement == 'set'
1714                                 && $stripped == 'to'
1715                                 && preg_match( '!(http|https|data|script):!sim', $value )
1716                         ) {
1717                                 wfDebug( __METHOD__ . ": Found svg setting attribute to '$value' in uploaded file.\n" );
1718
1719                                 return [ 'uploaded-wrong-setting-svg', $value ];
1720                         }
1721
1722                         # use handler attribute with remote / data / script
1723                         if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1724                                 wfDebug( __METHOD__ . ": Found svg setting handler with remote/data/script "
1725                                         . "'$attrib'='$value' in uploaded file.\n" );
1726
1727                                 return [ 'uploaded-setting-handler-svg', $attrib, $value ];
1728                         }
1729
1730                         # use CSS styles to bring in remote code
1731                         if ( $stripped == 'style'
1732                                 && self::checkCssFragment( Sanitizer::normalizeCss( $value ) )
1733                         ) {
1734                                 wfDebug( __METHOD__ . ": Found svg setting a style with "
1735                                         . "remote url '$attrib'='$value' in uploaded file.\n" );
1736                                 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1737                         }
1738
1739                         # Several attributes can include css, css character escaping isn't allowed
1740                         $cssAttrs = [ 'font', 'clip-path', 'fill', 'filter', 'marker',
1741                                 'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke' ];
1742                         if ( in_array( $stripped, $cssAttrs )
1743                                 && self::checkCssFragment( $value )
1744                         ) {
1745                                 wfDebug( __METHOD__ . ": Found svg setting a style with "
1746                                         . "remote url '$attrib'='$value' in uploaded file.\n" );
1747                                 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1748                         }
1749
1750                         # image filters can pull in url, which could be svg that executes scripts
1751                         # Only allow url( "#foo" ). Do not allow url( http://example.com )
1752                         if ( $strippedElement == 'image'
1753                                 && $stripped == 'filter'
1754                                 && preg_match( '!url\s*\(\s*["\']?[^#]!sim', $value )
1755                         ) {
1756                                 wfDebug( __METHOD__ . ": Found image filter with url: "
1757                                         . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1758
1759                                 return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ];
1760                         }
1761                 }
1762
1763                 return false; // No scripts detected
1764         }
1765
1766         /**
1767          * Check a block of CSS or CSS fragment for anything that looks like
1768          * it is bringing in remote code.
1769          * @param string $value a string of CSS
1770          * @return bool true if the CSS contains an illegal string, false if otherwise
1771          */
1772         private static function checkCssFragment( $value ) {
1773                 # Forbid external stylesheets, for both reliability and to protect viewer's privacy
1774                 if ( stripos( $value, '@import' ) !== false ) {
1775                         return true;
1776                 }
1777
1778                 # We allow @font-face to embed fonts with data: urls, so we snip the string
1779                 # 'url' out so this case won't match when we check for urls below
1780                 $pattern = '!(@font-face\s*{[^}]*src:)url(\("data:;base64,)!im';
1781                 $value = preg_replace( $pattern, '$1$2', $value );
1782
1783                 # Check for remote and executable CSS. Unlike in Sanitizer::checkCss, the CSS
1784                 # properties filter and accelerator don't seem to be useful for xss in SVG files.
1785                 # Expression and -o-link don't seem to work either, but filtering them here in case.
1786                 # Additionally, we catch remote urls like url("http:..., url('http:..., url(http:...,
1787                 # but not local ones such as url("#..., url('#..., url(#....
1788                 if ( preg_match( '!expression
1789                                 | -o-link\s*:
1790                                 | -o-link-source\s*:
1791                                 | -o-replace\s*:!imx', $value ) ) {
1792                         return true;
1793                 }
1794
1795                 if ( preg_match_all(
1796                                 "!(\s*(url|image|image-set)\s*\(\s*[\"']?\s*[^#]+.*?\))!sim",
1797                                 $value,
1798                                 $matches
1799                         ) !== 0
1800                 ) {
1801                         # TODO: redo this in one regex. Until then, url("#whatever") matches the first
1802                         foreach ( $matches[1] as $match ) {
1803                                 if ( !preg_match( "!\s*(url|image|image-set)\s*\(\s*(#|'#|\"#)!im", $match ) ) {
1804                                         return true;
1805                                 }
1806                         }
1807                 }
1808
1809                 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
1810                         return true;
1811                 }
1812
1813                 return false;
1814         }
1815
1816         /**
1817          * Divide the element name passed by the xml parser to the callback into URI and prifix.
1818          * @param string $element
1819          * @return array Containing the namespace URI and prefix
1820          */
1821         private static function splitXmlNamespace( $element ) {
1822                 // 'http://www.w3.org/2000/svg:script' -> [ 'http://www.w3.org/2000/svg', 'script' ]
1823                 $parts = explode( ':', strtolower( $element ) );
1824                 $name = array_pop( $parts );
1825                 $ns = implode( ':', $parts );
1826
1827                 return [ $ns, $name ];
1828         }
1829
1830         /**
1831          * @param string $name
1832          * @return string
1833          */
1834         private function stripXmlNamespace( $name ) {
1835                 // 'http://www.w3.org/2000/svg:script' -> 'script'
1836                 $parts = explode( ':', strtolower( $name ) );
1837
1838                 return array_pop( $parts );
1839         }
1840
1841         /**
1842          * Generic wrapper function for a virus scanner program.
1843          * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1844          * $wgAntivirusRequired may be used to deny upload if the scan fails.
1845          *
1846          * @param string $file Pathname to the temporary upload file
1847          * @return bool|null|string False if not virus is found, null if the scan fails or is disabled,
1848          *   or a string containing feedback from the virus scanner if a virus was found.
1849          *   If textual feedback is missing but a virus was found, this function returns true.
1850          */
1851         public static function detectVirus( $file ) {
1852                 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1853
1854                 if ( !$wgAntivirus ) {
1855                         wfDebug( __METHOD__ . ": virus scanner disabled\n" );
1856
1857                         return null;
1858                 }
1859
1860                 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1861                         wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
1862                         $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1863                                 [ 'virus-badscanner', $wgAntivirus ] );
1864
1865                         return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1866                 }
1867
1868                 # look up scanner configuration
1869                 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1870                 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1871                 $msgPattern = $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ?? null;
1872
1873                 if ( strpos( $command, "%f" ) === false ) {
1874                         # simple pattern: append file to scan
1875                         $command .= " " . Shell::escape( $file );
1876                 } else {
1877                         # complex pattern: replace "%f" with file to scan
1878                         $command = str_replace( "%f", Shell::escape( $file ), $command );
1879                 }
1880
1881                 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
1882
1883                 # execute virus scanner
1884                 $exitCode = false;
1885
1886                 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1887                 #      that does not seem to be worth the pain.
1888                 #      Ask me (Duesentrieb) about it if it's ever needed.
1889                 $output = wfShellExecWithStderr( $command, $exitCode );
1890
1891                 # map exit code to AV_xxx constants.
1892                 $mappedCode = $exitCode;
1893                 if ( $exitCodeMap ) {
1894                         if ( isset( $exitCodeMap[$exitCode] ) ) {
1895                                 $mappedCode = $exitCodeMap[$exitCode];
1896                         } elseif ( isset( $exitCodeMap["*"] ) ) {
1897                                 $mappedCode = $exitCodeMap["*"];
1898                         }
1899                 }
1900
1901                 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1902                  * so we need the strict equalities === and thus can't use a switch here
1903                  */
1904                 if ( $mappedCode === AV_SCAN_FAILED ) {
1905                         # scan failed (code was mapped to false by $exitCodeMap)
1906                         wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
1907
1908                         $output = $wgAntivirusRequired
1909                                 ? wfMessage( 'virus-scanfailed', [ $exitCode ] )->text()
1910                                 : null;
1911                 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
1912                         # scan failed because filetype is unknown (probably imune)
1913                         wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
1914                         $output = null;
1915                 } elseif ( $mappedCode === AV_NO_VIRUS ) {
1916                         # no virus found
1917                         wfDebug( __METHOD__ . ": file passed virus scan.\n" );
1918                         $output = false;
1919                 } else {
1920                         $output = trim( $output );
1921
1922                         if ( !$output ) {
1923                                 $output = true; # if there's no output, return true
1924                         } elseif ( $msgPattern ) {
1925                                 $groups = [];
1926                                 if ( preg_match( $msgPattern, $output, $groups ) && $groups[1] ) {
1927                                         $output = $groups[1];
1928                                 }
1929                         }
1930
1931                         wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
1932                 }
1933
1934                 return $output;
1935         }
1936
1937         /**
1938          * Check if there's an overwrite conflict and, if so, if restrictions
1939          * forbid this user from performing the upload.
1940          *
1941          * @param User $user
1942          *
1943          * @return bool|array
1944          */
1945         private function checkOverwrite( $user ) {
1946                 // First check whether the local file can be overwritten
1947                 $file = $this->getLocalFile();
1948                 $file->load( File::READ_LATEST );
1949                 if ( $file->exists() ) {
1950                         if ( !self::userCanReUpload( $user, $file ) ) {
1951                                 return [ 'fileexists-forbidden', $file->getName() ];
1952                         } else {
1953                                 return true;
1954                         }
1955                 }
1956
1957                 /* Check shared conflicts: if the local file does not exist, but
1958                  * wfFindFile finds a file, it exists in a shared repository.
1959                  */
1960                 $file = wfFindFile( $this->getTitle(), [ 'latest' => true ] );
1961                 if ( $file && !MediaWikiServices::getInstance()
1962                                 ->getPermissionManager()
1963                                 ->userHasRight( $user, 'reupload-shared' )
1964                 ) {
1965                         return [ 'fileexists-shared-forbidden', $file->getName() ];
1966                 }
1967
1968                 return true;
1969         }
1970
1971         /**
1972          * Check if a user is the last uploader
1973          *
1974          * @param User $user
1975          * @param File $img
1976          * @return bool
1977          */
1978         public static function userCanReUpload( User $user, File $img ) {
1979                 $permissionManager = MediaWikiServices::getInstance()->getPermissionManager();
1980                 if ( $permissionManager->userHasRight( $user, 'reupload' ) ) {
1981                         return true; // non-conditional
1982                 } elseif ( !$permissionManager->userHasRight( $user, 'reupload-own' ) ) {
1983                         return false;
1984                 }
1985
1986                 if ( !( $img instanceof LocalFile ) ) {
1987                         return false;
1988                 }
1989
1990                 $img->load();
1991
1992                 return $user->getId() == $img->getUser( 'id' );
1993         }
1994
1995         /**
1996          * Helper function that does various existence checks for a file.
1997          * The following checks are performed:
1998          * - The file exists
1999          * - Article with the same name as the file exists
2000          * - File exists with normalized extension
2001          * - The file looks like a thumbnail and the original exists
2002          *
2003          * @param File $file The File object to check
2004          * @return mixed False if the file does not exists, else an array
2005          */
2006         public static function getExistsWarning( $file ) {
2007                 if ( $file->exists() ) {
2008                         return [ 'warning' => 'exists', 'file' => $file ];
2009                 }
2010
2011                 if ( $file->getTitle()->getArticleID() ) {
2012                         return [ 'warning' => 'page-exists', 'file' => $file ];
2013                 }
2014
2015                 if ( strpos( $file->getName(), '.' ) == false ) {
2016                         $partname = $file->getName();
2017                         $extension = '';
2018                 } else {
2019                         $n = strrpos( $file->getName(), '.' );
2020                         $extension = substr( $file->getName(), $n + 1 );
2021                         $partname = substr( $file->getName(), 0, $n );
2022                 }
2023                 $normalizedExtension = File::normalizeExtension( $extension );
2024
2025                 if ( $normalizedExtension != $extension ) {
2026                         // We're not using the normalized form of the extension.
2027                         // Normal form is lowercase, using most common of alternate
2028                         // extensions (eg 'jpg' rather than 'JPEG').
2029
2030                         // Check for another file using the normalized form...
2031                         $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
2032                         $file_lc = wfLocalFile( $nt_lc );
2033
2034                         if ( $file_lc->exists() ) {
2035                                 return [
2036                                         'warning' => 'exists-normalized',
2037                                         'file' => $file,
2038                                         'normalizedFile' => $file_lc
2039                                 ];
2040                         }
2041                 }
2042
2043                 // Check for files with the same name but a different extension
2044                 $similarFiles = RepoGroup::singleton()->getLocalRepo()->findFilesByPrefix(
2045                         "{$partname}.", 1 );
2046                 if ( count( $similarFiles ) ) {
2047                         return [
2048                                 'warning' => 'exists-normalized',
2049                                 'file' => $file,
2050                                 'normalizedFile' => $similarFiles[0],
2051                         ];
2052                 }
2053
2054                 if ( self::isThumbName( $file->getName() ) ) {
2055                         # Check for filenames like 50px- or 180px-, these are mostly thumbnails
2056                         $nt_thb = Title::newFromText(
2057                                 substr( $partname, strpos( $partname, '-' ) + 1 ) . '.' . $extension,
2058                                 NS_FILE
2059                         );
2060                         $file_thb = wfLocalFile( $nt_thb );
2061                         if ( $file_thb->exists() ) {
2062                                 return [
2063                                         'warning' => 'thumb',
2064                                         'file' => $file,
2065                                         'thumbFile' => $file_thb
2066                                 ];
2067                         } else {
2068                                 // File does not exist, but we just don't like the name
2069                                 return [
2070                                         'warning' => 'thumb-name',
2071                                         'file' => $file,
2072                                         'thumbFile' => $file_thb
2073                                 ];
2074                         }
2075                 }
2076
2077                 foreach ( self::getFilenamePrefixBlacklist() as $prefix ) {
2078                         if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
2079                                 return [
2080                                         'warning' => 'bad-prefix',
2081                                         'file' => $file,
2082                                         'prefix' => $prefix
2083                                 ];
2084                         }
2085                 }
2086
2087                 return false;
2088         }
2089
2090         /**
2091          * Helper function that checks whether the filename looks like a thumbnail
2092          * @param string $filename
2093          * @return bool
2094          */
2095         public static function isThumbName( $filename ) {
2096                 $n = strrpos( $filename, '.' );
2097                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
2098
2099                 return (
2100                                 substr( $partname, 3, 3 ) == 'px-' ||
2101                                 substr( $partname, 2, 3 ) == 'px-'
2102                         ) &&
2103                         preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
2104         }
2105
2106         /**
2107          * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
2108          *
2109          * @return array List of prefixes
2110          */
2111         public static function getFilenamePrefixBlacklist() {
2112                 $blacklist = [];
2113                 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
2114                 if ( !$message->isDisabled() ) {
2115                         $lines = explode( "\n", $message->plain() );
2116                         foreach ( $lines as $line ) {
2117                                 // Remove comment lines
2118                                 $comment = substr( trim( $line ), 0, 1 );
2119                                 if ( $comment == '#' || $comment == '' ) {
2120                                         continue;
2121                                 }
2122                                 // Remove additional comments after a prefix
2123                                 $comment = strpos( $line, '#' );
2124                                 if ( $comment > 0 ) {
2125                                         $line = substr( $line, 0, $comment - 1 );
2126                                 }
2127                                 $blacklist[] = trim( $line );
2128                         }
2129                 }
2130
2131                 return $blacklist;
2132         }
2133
2134         /**
2135          * Gets image info about the file just uploaded.
2136          *
2137          * Also has the effect of setting metadata to be an 'indexed tag name' in
2138          * returned API result if 'metadata' was requested. Oddly, we have to pass
2139          * the "result" object down just so it can do that with the appropriate
2140          * format, presumably.
2141          *
2142          * @param ApiResult $result
2143          * @return array Image info
2144          */
2145         public function getImageInfo( $result ) {
2146                 $localFile = $this->getLocalFile();
2147                 $stashFile = $this->getStashFile();
2148                 // Calling a different API module depending on whether the file was stashed is less than optimal.
2149                 // In fact, calling API modules here at all is less than optimal. Maybe it should be refactored.
2150                 if ( $stashFile ) {
2151                         $imParam = ApiQueryStashImageInfo::getPropertyNames();
2152                         $info = ApiQueryStashImageInfo::getInfo( $stashFile, array_flip( $imParam ), $result );
2153                 } else {
2154                         $imParam = ApiQueryImageInfo::getPropertyNames();
2155                         $info = ApiQueryImageInfo::getInfo( $localFile, array_flip( $imParam ), $result );
2156                 }
2157
2158                 return $info;
2159         }
2160
2161         /**
2162          * @param array $error
2163          * @return Status
2164          */
2165         public function convertVerifyErrorToStatus( $error ) {
2166                 $code = $error['status'];
2167                 unset( $code['status'] );
2168
2169                 return Status::newFatal( $this->getVerificationErrorCode( $code ), $error );
2170         }
2171
2172         /**
2173          * Get the MediaWiki maximum uploaded file size for given type of upload, based on
2174          * $wgMaxUploadSize.
2175          *
2176          * @param null|string $forType
2177          * @return int
2178          */
2179         public static function getMaxUploadSize( $forType = null ) {
2180                 global $wgMaxUploadSize;
2181
2182                 if ( is_array( $wgMaxUploadSize ) ) {
2183                         if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
2184                                 return $wgMaxUploadSize[$forType];
2185                         } else {
2186                                 return $wgMaxUploadSize['*'];
2187                         }
2188                 } else {
2189                         return intval( $wgMaxUploadSize );
2190                 }
2191         }
2192
2193         /**
2194          * Get the PHP maximum uploaded file size, based on ini settings. If there is no limit or the
2195          * limit can't be guessed, returns a very large number (PHP_INT_MAX).
2196          *
2197          * @since 1.27
2198          * @return int
2199          */
2200         public static function getMaxPhpUploadSize() {
2201                 $phpMaxFileSize = wfShorthandToInteger(
2202                         ini_get( 'upload_max_filesize' ) ?: ini_get( 'hhvm.server.upload.upload_max_file_size' ),
2203                         PHP_INT_MAX
2204                 );
2205                 $phpMaxPostSize = wfShorthandToInteger(
2206                         ini_get( 'post_max_size' ) ?: ini_get( 'hhvm.server.max_post_size' ),
2207                         PHP_INT_MAX
2208                 ) ?: PHP_INT_MAX;
2209                 return min( $phpMaxFileSize, $phpMaxPostSize );
2210         }
2211
2212         /**
2213          * Get the current status of a chunked upload (used for polling)
2214          *
2215          * The value will be read from cache.
2216          *
2217          * @param User $user
2218          * @param string $statusKey
2219          * @return Status[]|bool
2220          */
2221         public static function getSessionStatus( User $user, $statusKey ) {
2222                 $store = self::getUploadSessionStore();
2223                 $key = self::getUploadSessionKey( $store, $user, $statusKey );
2224
2225                 return $store->get( $key );
2226         }
2227
2228         /**
2229          * Set the current status of a chunked upload (used for polling)
2230          *
2231          * The value will be set in cache for 1 day
2232          *
2233          * Avoid triggering this method on HTTP GET/HEAD requests
2234          *
2235          * @param User $user
2236          * @param string $statusKey
2237          * @param array|bool $value
2238          * @return void
2239          */
2240         public static function setSessionStatus( User $user, $statusKey, $value ) {
2241                 $store = self::getUploadSessionStore();
2242                 $key = self::getUploadSessionKey( $store, $user, $statusKey );
2243
2244                 if ( $value === false ) {
2245                         $store->delete( $key );
2246                 } else {
2247                         $store->set( $key, $value, $store::TTL_DAY );
2248                 }
2249         }
2250
2251         /**
2252          * @param BagOStuff $store
2253          * @param User $user
2254          * @param string $statusKey
2255          * @return string
2256          */
2257         private static function getUploadSessionKey( BagOStuff $store, User $user, $statusKey ) {
2258                 return $store->makeKey(
2259                         'uploadstatus',
2260                         $user->getId() ?: md5( $user->getName() ),
2261                         $statusKey
2262                 );
2263         }
2264
2265         /**
2266          * @return BagOStuff
2267          */
2268         private static function getUploadSessionStore() {
2269                 return ObjectCache::getInstance( 'db-replicated' );
2270         }
2271 }