includes/upload/UploadBase.php

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