includes/upload/UploadBase.php

   1 <?php
   2 /**
   3  * Base class for the backend of file upload.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Upload
  22  */
  23 use MediaWiki\MediaWikiServices;
  24
  25 /**
  26  * @defgroup Upload Upload related
  27  */
  28
  29 /**
  30  * @ingroup Upload
  31  *
  32  * UploadBase and subclasses are the backend of MediaWiki's file uploads.
  33  * The frontends are formed by ApiUpload and SpecialUpload.
  34  *
  35  * @author Brion Vibber
  36  * @author Bryan Tong Minh
  37  * @author Michael Dale
  38  */
  39 abstract class UploadBase {
  40         /** @var string Local file system path to the file to upload (or a local copy) */
  41         protected $mTempPath;
  42         /** @var TempFSFile|null Wrapper to handle deleting the temp file */
  43         protected $tempFileObj;
  44
  45         protected $mDesiredDestName, $mDestName, $mRemoveTempFile, $mSourceType;
  46         protected $mTitle = false, $mTitleError = 0;
  47         protected $mFilteredName, $mFinalExtension;
  48         protected $mLocalFile, $mStashFile, $mFileSize, $mFileProps;
  49         protected $mBlackListedExtensions;
  50         protected $mJavaDetected, $mSVGNSError;
  51
  52         protected static $safeXmlEncodings = [
  53                 'UTF-8',
  54                 'ISO-8859-1',
  55                 'ISO-8859-2',
  56                 'UTF-16',
  57                 'UTF-32',
  58                 'WINDOWS-1250',
  59                 'WINDOWS-1251',
  60                 'WINDOWS-1252',
  61                 'WINDOWS-1253',
  62                 'WINDOWS-1254',
  63                 'WINDOWS-1255',
  64                 'WINDOWS-1256',
  65                 'WINDOWS-1257',
  66                 'WINDOWS-1258',
  67         ];
  68
  69         const SUCCESS = 0;
  70         const OK = 0;
  71         const EMPTY_FILE = 3;
  72         const MIN_LENGTH_PARTNAME = 4;
  73         const ILLEGAL_FILENAME = 5;
  74         const OVERWRITE_EXISTING_FILE = 7; # Not used anymore; handled by verifyTitlePermissions()
  75         const FILETYPE_MISSING = 8;
  76         const FILETYPE_BADTYPE = 9;
  77         const VERIFICATION_ERROR = 10;
  78         const HOOK_ABORTED = 11;
  79         const FILE_TOO_LARGE = 12;
  80         const WINDOWS_NONASCII_FILENAME = 13;
  81         const FILENAME_TOO_LONG = 14;
  82
  83         /**
  84          * @param int $error
  85          * @return string
  86          */
  87         public function getVerificationErrorCode( $error ) {
  88                 $code_to_status = [
  89                         self::EMPTY_FILE => 'empty-file',
  90                         self::FILE_TOO_LARGE => 'file-too-large',
  91                         self::FILETYPE_MISSING => 'filetype-missing',
  92                         self::FILETYPE_BADTYPE => 'filetype-banned',
  93                         self::MIN_LENGTH_PARTNAME => 'filename-tooshort',
  94                         self::ILLEGAL_FILENAME => 'illegal-filename',
  95                         self::OVERWRITE_EXISTING_FILE => 'overwrite',
  96                         self::VERIFICATION_ERROR => 'verification-error',
  97                         self::HOOK_ABORTED => 'hookaborted',
  98                         self::WINDOWS_NONASCII_FILENAME => 'windows-nonascii-filename',
  99                         self::FILENAME_TOO_LONG => 'filename-toolong',
 100                 ];
 101                 return $code_to_status[$error] ?? 'unknown-error';
 102         }
 103
 104         /**
 105          * Returns true if uploads are enabled.
 106          * Can be override by subclasses.
 107          * @return bool
 108          */
 109         public static function isEnabled() {
 110                 global $wgEnableUploads;
 111
 112                 if ( !$wgEnableUploads ) {
 113                         return false;
 114                 }
 115
 116                 # Check php's file_uploads setting
 117                 return wfIsHHVM() || wfIniGetBool( 'file_uploads' );
 118         }
 119
 120         /**
 121          * Returns true if the user can use this upload module or else a string
 122          * identifying the missing permission.
 123          * Can be overridden by subclasses.
 124          *
 125          * @param User $user
 126          * @return bool|string
 127          */
 128         public static function isAllowed( $user ) {
 129                 foreach ( [ 'upload', 'edit' ] as $permission ) {
 130                         if ( !$user->isAllowed( $permission ) ) {
 131                                 return $permission;
 132                         }
 133                 }
 134
 135                 return true;
 136         }
 137
 138         /**
 139          * Returns true if the user has surpassed the upload rate limit, false otherwise.
 140          *
 141          * @param User $user
 142          * @return bool
 143          */
 144         public static function isThrottled( $user ) {
 145                 return $user->pingLimiter( 'upload' );
 146         }
 147
 148         // Upload handlers. Should probably just be a global.
 149         private static $uploadHandlers = [ 'Stash', 'File', 'Url' ];
 150
 151         /**
 152          * Create a form of UploadBase depending on wpSourceType and initializes it
 153          *
 154          * @param WebRequest &$request
 155          * @param string|null $type
 156          * @return null|UploadBase
 157          */
 158         public static function createFromRequest( &$request, $type = null ) {
 159                 $type = $type ?: $request->getVal( 'wpSourceType', 'File' );
 160
 161                 if ( !$type ) {
 162                         return null;
 163                 }
 164
 165                 // Get the upload class
 166                 $type = ucfirst( $type );
 167
 168                 // Give hooks the chance to handle this request
 169                 $className = null;
 170                 Hooks::run( 'UploadCreateFromRequest', [ $type, &$className ] );
 171                 if ( is_null( $className ) ) {
 172                         $className = 'UploadFrom' . $type;
 173                         wfDebug( __METHOD__ . ": class name: $className\n" );
 174                         if ( !in_array( $type, self::$uploadHandlers ) ) {
 175                                 return null;
 176                         }
 177                 }
 178
 179                 // Check whether this upload class is enabled
 180                 if ( !call_user_func( [ $className, 'isEnabled' ] ) ) {
 181                         return null;
 182                 }
 183
 184                 // Check whether the request is valid
 185                 if ( !call_user_func( [ $className, 'isValidRequest' ], $request ) ) {
 186                         return null;
 187                 }
 188
 189                 /** @var UploadBase $handler */
 190                 $handler = new $className;
 191
 192                 $handler->initializeFromRequest( $request );
 193
 194                 return $handler;
 195         }
 196
 197         /**
 198          * Check whether a request if valid for this handler
 199          * @param WebRequest $request
 200          * @return bool
 201          */
 202         public static function isValidRequest( $request ) {
 203                 return false;
 204         }
 205
 206         public function __construct() {
 207         }
 208
 209         /**
 210          * Returns the upload type. Should be overridden by child classes
 211          *
 212          * @since 1.18
 213          * @return string
 214          */
 215         public function getSourceType() {
 216                 return null;
 217         }
 218
 219         /**
 220          * Initialize the path information
 221          * @param string $name The desired destination name
 222          * @param string $tempPath The temporary path
 223          * @param int $fileSize The file size
 224          * @param bool $removeTempFile (false) remove the temporary file?
 225          * @throws MWException
 226          */
 227         public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
 228                 $this->mDesiredDestName = $name;
 229                 if ( FileBackend::isStoragePath( $tempPath ) ) {
 230                         throw new MWException( __METHOD__ . " given storage path `$tempPath`." );
 231                 }
 232
 233                 $this->setTempFile( $tempPath, $fileSize );
 234                 $this->mRemoveTempFile = $removeTempFile;
 235         }
 236
 237         /**
 238          * Initialize from a WebRequest. Override this in a subclass.
 239          *
 240          * @param WebRequest &$request
 241          */
 242         abstract public function initializeFromRequest( &$request );
 243
 244         /**
 245          * @param string $tempPath File system path to temporary file containing the upload
 246          * @param int|null $fileSize
 247          */
 248         protected function setTempFile( $tempPath, $fileSize = null ) {
 249                 $this->mTempPath = $tempPath;
 250                 $this->mFileSize = $fileSize ?: null;
 251                 if ( strlen( $this->mTempPath ) && file_exists( $this->mTempPath ) ) {
 252                         $this->tempFileObj = new TempFSFile( $this->mTempPath );
 253                         if ( !$fileSize ) {
 254                                 $this->mFileSize = filesize( $this->mTempPath );
 255                         }
 256                 } else {
 257                         $this->tempFileObj = null;
 258                 }
 259         }
 260
 261         /**
 262          * Fetch the file. Usually a no-op
 263          * @return Status
 264          */
 265         public function fetchFile() {
 266                 return Status::newGood();
 267         }
 268
 269         /**
 270          * Return true if the file is empty
 271          * @return bool
 272          */
 273         public function isEmptyFile() {
 274                 return empty( $this->mFileSize );
 275         }
 276
 277         /**
 278          * Return the file size
 279          * @return int
 280          */
 281         public function getFileSize() {
 282                 return $this->mFileSize;
 283         }
 284
 285         /**
 286          * Get the base 36 SHA1 of the file
 287          * @return string
 288          */
 289         public function getTempFileSha1Base36() {
 290                 return FSFile::getSha1Base36FromPath( $this->mTempPath );
 291         }
 292
 293         /**
 294          * @param string $srcPath The source path
 295          * @return string|bool The real path if it was a virtual URL Returns false on failure
 296          */
 297         public function getRealPath( $srcPath ) {
 298                 $repo = RepoGroup::singleton()->getLocalRepo();
 299                 if ( FileRepo::isVirtualUrl( $srcPath ) ) {
 300                         /** @todo Just make uploads work with storage paths UploadFromStash
 301                          *  loads files via virtual URLs.
 302                          */
 303                         $tmpFile = $repo->getLocalCopy( $srcPath );
 304                         if ( $tmpFile ) {
 305                                 $tmpFile->bind( $this ); // keep alive with $this
 306                         }
 307                         $path = $tmpFile ? $tmpFile->getPath() : false;
 308                 } else {
 309                         $path = $srcPath;
 310                 }
 311
 312                 return $path;
 313         }
 314
 315         /**
 316          * Verify whether the upload is sane.
 317          * @return mixed Const self::OK or else an array with error information
 318          */
 319         public function verifyUpload() {
 320                 /**
 321                  * If there was no filename or a zero size given, give up quick.
 322                  */
 323                 if ( $this->isEmptyFile() ) {
 324                         return [ 'status' => self::EMPTY_FILE ];
 325                 }
 326
 327                 /**
 328                  * Honor $wgMaxUploadSize
 329                  */
 330                 $maxSize = self::getMaxUploadSize( $this->getSourceType() );
 331                 if ( $this->mFileSize > $maxSize ) {
 332                         return [
 333                                 'status' => self::FILE_TOO_LARGE,
 334                                 'max' => $maxSize,
 335                         ];
 336                 }
 337
 338                 /**
 339                  * Look at the contents of the file; if we can recognize the
 340                  * type but it's corrupt or data of the wrong type, we should
 341                  * probably not accept it.
 342                  */
 343                 $verification = $this->verifyFile();
 344                 if ( $verification !== true ) {
 345                         return [
 346                                 'status' => self::VERIFICATION_ERROR,
 347                                 'details' => $verification
 348                         ];
 349                 }
 350
 351                 /**
 352                  * Make sure this file can be created
 353                  */
 354                 $result = $this->validateName();
 355                 if ( $result !== true ) {
 356                         return $result;
 357                 }
 358
 359                 $error = '';
 360                 if ( !Hooks::run( 'UploadVerification',
 361                         [ $this->mDestName, $this->mTempPath, &$error ], '1.28' )
 362                 ) {
 363                         return [ 'status' => self::HOOK_ABORTED, 'error' => $error ];
 364                 }
 365
 366                 return [ 'status' => self::OK ];
 367         }
 368
 369         /**
 370          * Verify that the name is valid and, if necessary, that we can overwrite
 371          *
 372          * @return mixed True if valid, otherwise and array with 'status'
 373          * and other keys
 374          */
 375         public function validateName() {
 376                 $nt = $this->getTitle();
 377                 if ( is_null( $nt ) ) {
 378                         $result = [ 'status' => $this->mTitleError ];
 379                         if ( $this->mTitleError == self::ILLEGAL_FILENAME ) {
 380                                 $result['filtered'] = $this->mFilteredName;
 381                         }
 382                         if ( $this->mTitleError == self::FILETYPE_BADTYPE ) {
 383                                 $result['finalExt'] = $this->mFinalExtension;
 384                                 if ( count( $this->mBlackListedExtensions ) ) {
 385                                         $result['blacklistedExt'] = $this->mBlackListedExtensions;
 386                                 }
 387                         }
 388
 389                         return $result;
 390                 }
 391                 $this->mDestName = $this->getLocalFile()->getName();
 392
 393                 return true;
 394         }
 395
 396         /**
 397          * Verify the MIME type.
 398          *
 399          * @note Only checks that it is not an evil MIME. The "does it have
 400          *  correct extension given its MIME type?" check is in verifyFile.
 401          *  in `verifyFile()` that MIME type and file extension correlate.
 402          * @param string $mime Representing the MIME
 403          * @return mixed True if the file is verified, an array otherwise
 404          */
 405         protected function verifyMimeType( $mime ) {
 406                 global $wgVerifyMimeType;
 407                 if ( $wgVerifyMimeType ) {
 408                         wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>\n" );
 409                         global $wgMimeTypeBlacklist;
 410                         if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
 411                                 return [ 'filetype-badmime', $mime ];
 412                         }
 413
 414                         # Check what Internet Explorer would detect
 415                         $fp = fopen( $this->mTempPath, 'rb' );
 416                         $chunk = fread( $fp, 256 );
 417                         fclose( $fp );
 418
 419                         $magic = MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer();
 420                         $extMime = $magic->guessTypesForExtension( $this->mFinalExtension );
 421                         $ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime );
 422                         foreach ( $ieTypes as $ieType ) {
 423                                 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
 424                                         return [ 'filetype-bad-ie-mime', $ieType ];
 425                                 }
 426                         }
 427                 }
 428
 429                 return true;
 430         }
 431
 432         /**
 433          * Verifies that it's ok to include the uploaded file
 434          *
 435          * @return mixed True of the file is verified, array otherwise.
 436          */
 437         protected function verifyFile() {
 438                 global $wgVerifyMimeType, $wgDisableUploadScriptChecks;
 439
 440                 $status = $this->verifyPartialFile();
 441                 if ( $status !== true ) {
 442                         return $status;
 443                 }
 444
 445                 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer() );
 446                 $this->mFileProps = $mwProps->getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 447                 $mime = $this->mFileProps['mime'];
 448
 449                 if ( $wgVerifyMimeType ) {
 450                         # XXX: Missing extension will be caught by validateName() via getTitle()
 451                         if ( $this->mFinalExtension != '' && !$this->verifyExtension( $mime, $this->mFinalExtension ) ) {
 452                                 return [ 'filetype-mime-mismatch', $this->mFinalExtension, $mime ];
 453                         }
 454                 }
 455
 456                 # check for htmlish code and javascript
 457                 if ( !$wgDisableUploadScriptChecks ) {
 458                         if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 459                                 $svgStatus = $this->detectScriptInSvg( $this->mTempPath, false );
 460                                 if ( $svgStatus !== false ) {
 461                                         return $svgStatus;
 462                                 }
 463                         }
 464                 }
 465
 466                 $handler = MediaHandler::getHandler( $mime );
 467                 if ( $handler ) {
 468                         $handlerStatus = $handler->verifyUpload( $this->mTempPath );
 469                         if ( !$handlerStatus->isOK() ) {
 470                                 $errors = $handlerStatus->getErrorsArray();
 471
 472                                 return reset( $errors );
 473                         }
 474                 }
 475
 476                 $error = true;
 477                 Hooks::run( 'UploadVerifyFile', [ $this, $mime, &$error ] );
 478                 if ( $error !== true ) {
 479                         if ( !is_array( $error ) ) {
 480                                 $error = [ $error ];
 481                         }
 482                         return $error;
 483                 }
 484
 485                 wfDebug( __METHOD__ . ": all clear; passing.\n" );
 486
 487                 return true;
 488         }
 489
 490         /**
 491          * A verification routine suitable for partial files
 492          *
 493          * Runs the blacklist checks, but not any checks that may
 494          * assume the entire file is present.
 495          *
 496          * @return mixed True for valid or array with error message key.
 497          */
 498         protected function verifyPartialFile() {
 499                 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
 500
 501                 # getTitle() sets some internal parameters like $this->mFinalExtension
 502                 $this->getTitle();
 503
 504                 $mwProps = new MWFileProps( MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer() );
 505                 $this->mFileProps = $mwProps->getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 506
 507                 # check MIME type, if desired
 508                 $mime = $this->mFileProps['file-mime'];
 509                 $status = $this->verifyMimeType( $mime );
 510                 if ( $status !== true ) {
 511                         return $status;
 512                 }
 513
 514                 # check for htmlish code and javascript
 515                 if ( !$wgDisableUploadScriptChecks ) {
 516                         if ( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
 517                                 return [ 'uploadscripted' ];
 518                         }
 519                         if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 520                                 $svgStatus = $this->detectScriptInSvg( $this->mTempPath, true );
 521                                 if ( $svgStatus !== false ) {
 522                                         return $svgStatus;
 523                                 }
 524                         }
 525                 }
 526
 527                 # Check for Java applets, which if uploaded can bypass cross-site
 528                 # restrictions.
 529                 if ( !$wgAllowJavaUploads ) {
 530                         $this->mJavaDetected = false;
 531                         $zipStatus = ZipDirectoryReader::read( $this->mTempPath,
 532                                 [ $this, 'zipEntryCallback' ] );
 533                         if ( !$zipStatus->isOK() ) {
 534                                 $errors = $zipStatus->getErrorsArray();
 535                                 $error = reset( $errors );
 536                                 if ( $error[0] !== 'zip-wrong-format' ) {
 537                                         return $error;
 538                                 }
 539                         }
 540                         if ( $this->mJavaDetected ) {
 541                                 return [ 'uploadjava' ];
 542                         }
 543                 }
 544
 545                 # Scan the uploaded file for viruses
 546                 $virus = $this->detectVirus( $this->mTempPath );
 547                 if ( $virus ) {
 548                         return [ 'uploadvirus', $virus ];
 549                 }
 550
 551                 return true;
 552         }
 553
 554         /**
 555          * Callback for ZipDirectoryReader to detect Java class files.
 556          *
 557          * @param array $entry
 558          */
 559         public function zipEntryCallback( $entry ) {
 560                 $names = [ $entry['name'] ];
 561
 562                 // If there is a null character, cut off the name at it, because JDK's
 563                 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
 564                 // were constructed which had ".class\0" followed by a string chosen to
 565                 // make the hash collide with the truncated name, that file could be
 566                 // returned in response to a request for the .class file.
 567                 $nullPos = strpos( $entry['name'], "\000" );
 568                 if ( $nullPos !== false ) {
 569                         $names[] = substr( $entry['name'], 0, $nullPos );
 570                 }
 571
 572                 // If there is a trailing slash in the file name, we have to strip it,
 573                 // because that's what ZIP_GetEntry() does.
 574                 if ( preg_grep( '!\.class/?$!', $names ) ) {
 575                         $this->mJavaDetected = true;
 576                 }
 577         }
 578
 579         /**
 580          * Alias for verifyTitlePermissions. The function was originally
 581          * 'verifyPermissions', but that suggests it's checking the user, when it's
 582          * really checking the title + user combination.
 583          *
 584          * @param User $user User object to verify the permissions against
 585          * @return mixed An array as returned by getUserPermissionsErrors or true
 586          *   in case the user has proper permissions.
 587          */
 588         public function verifyPermissions( $user ) {
 589                 return $this->verifyTitlePermissions( $user );
 590         }
 591
 592         /**
 593          * Check whether the user can edit, upload and create the image. This
 594          * checks only against the current title; if it returns errors, it may
 595          * very well be that another title will not give errors. Therefore
 596          * isAllowed() should be called as well for generic is-user-blocked or
 597          * can-user-upload checking.
 598          *
 599          * @param User $user User object to verify the permissions against
 600          * @return mixed An array as returned by getUserPermissionsErrors or true
 601          *   in case the user has proper permissions.
 602          */
 603         public function verifyTitlePermissions( $user ) {
 604                 /**
 605                  * If the image is protected, non-sysop users won't be able
 606                  * to modify it by uploading a new revision.
 607                  */
 608                 $nt = $this->getTitle();
 609                 if ( is_null( $nt ) ) {
 610                         return true;
 611                 }
 612                 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
 613                 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
 614                 if ( !$nt->exists() ) {
 615                         $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
 616                 } else {
 617                         $permErrorsCreate = [];
 618                 }
 619                 if ( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
 620                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
 621                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
 622
 623                         return $permErrors;
 624                 }
 625
 626                 $overwriteError = $this->checkOverwrite( $user );
 627                 if ( $overwriteError !== true ) {
 628                         return [ $overwriteError ];
 629                 }
 630
 631                 return true;
 632         }
 633
 634         /**
 635          * Check for non fatal problems with the file.
 636          *
 637          * This should not assume that mTempPath is set.
 638          *
 639          * @return mixed[] Array of warnings
 640          */
 641         public function checkWarnings() {
 642                 $warnings = [];
 643
 644                 $localFile = $this->getLocalFile();
 645                 $localFile->load( File::READ_LATEST );
 646                 $filename = $localFile->getName();
 647                 $hash = $this->getTempFileSha1Base36();
 648
 649                 $badFileName = $this->checkBadFileName( $filename, $this->mDesiredDestName );
 650                 if ( $badFileName !== null ) {
 651                         $warnings['badfilename'] = $badFileName;
 652                 }
 653
 654                 $unwantedFileExtensionDetails = $this->checkUnwantedFileExtensions( $this->mFinalExtension );
 655                 if ( $unwantedFileExtensionDetails !== null ) {
 656                         $warnings['filetype-unwanted-type'] = $unwantedFileExtensionDetails;
 657                 }
 658
 659                 $fileSizeWarnings = $this->checkFileSize( $this->mFileSize );
 660                 if ( $fileSizeWarnings ) {
 661                         $warnings = array_merge( $warnings, $fileSizeWarnings );
 662                 }
 663
 664                 $localFileExistsWarnings = $this->checkLocalFileExists( $localFile, $hash );
 665                 if ( $localFileExistsWarnings ) {
 666                         $warnings = array_merge( $warnings, $localFileExistsWarnings );
 667                 }
 668
 669                 if ( $this->checkLocalFileWasDeleted( $localFile ) ) {
 670                         $warnings['was-deleted'] = $filename;
 671                 }
 672
 673                 // If a file with the same name exists locally then the local file has already been tested
 674                 // for duplication of content
 675                 $ignoreLocalDupes = isset( $warnings[ 'exists '] );
 676                 $dupes = $this->checkAgainstExistingDupes( $hash, $ignoreLocalDupes );
 677                 if ( $dupes ) {
 678                         $warnings['duplicate'] = $dupes;
 679                 }
 680
 681                 $archivedDupes = $this->checkAgainstArchiveDupes( $hash );
 682                 if ( $archivedDupes !== null ) {
 683                         $warnings['duplicate-archive'] = $archivedDupes;
 684                 }
 685
 686                 return $warnings;
 687         }
 688
 689         /**
 690          * Check whether the resulting filename is different from the desired one,
 691          * but ignore things like ucfirst() and spaces/underscore things
 692          *
 693          * @param string $filename
 694          * @param string $desiredFileName
 695          *
 696          * @return string|null String that was determined to be bad or null if the filename is okay
 697          */
 698         private function checkBadFileName( $filename, $desiredFileName ) {
 699                 $comparableName = str_replace( ' ', '_', $desiredFileName );
 700                 $comparableName = Title::capitalize( $comparableName, NS_FILE );
 701
 702                 if ( $desiredFileName != $filename && $comparableName != $filename ) {
 703                         return $filename;
 704                 }
 705
 706                 return null;
 707         }
 708
 709         /**
 710          * @param string $fileExtension The file extension to check
 711          *
 712          * @return array|null array with the following keys:
 713          *                    0 => string The final extension being used
 714          *                    1 => string[] The extensions that are allowed
 715          *                    2 => int The number of extensions that are allowed.
 716          */
 717         private function checkUnwantedFileExtensions( $fileExtension ) {
 718                 global $wgCheckFileExtensions, $wgFileExtensions, $wgLang;
 719
 720                 if ( $wgCheckFileExtensions ) {
 721                         $extensions = array_unique( $wgFileExtensions );
 722                         if ( !$this->checkFileExtension( $fileExtension, $extensions ) ) {
 723                                 return [
 724                                         $fileExtension,
 725                                         $wgLang->commaList( $extensions ),
 726                                         count( $extensions )
 727                                 ];
 728                         }
 729                 }
 730
 731                 return null;
 732         }
 733
 734         /**
 735          * @param int $fileSize
 736          *
 737          * @return array warnings
 738          */
 739         private function checkFileSize( $fileSize ) {
 740                 global $wgUploadSizeWarning;
 741
 742                 $warnings = [];
 743
 744                 if ( $wgUploadSizeWarning && ( $fileSize > $wgUploadSizeWarning ) ) {
 745                         $warnings['large-file'] = [ $wgUploadSizeWarning, $fileSize ];
 746                 }
 747
 748                 if ( $fileSize == 0 ) {
 749                         $warnings['empty-file'] = true;
 750                 }
 751
 752                 return $warnings;
 753         }
 754
 755         /**
 756          * @param LocalFile $localFile
 757          * @param string $hash sha1 hash of the file to check
 758          *
 759          * @return array warnings
 760          */
 761         private function checkLocalFileExists( LocalFile $localFile, $hash ) {
 762                 $warnings = [];
 763
 764                 $exists = self::getExistsWarning( $localFile );
 765                 if ( $exists !== false ) {
 766                         $warnings['exists'] = $exists;
 767
 768                         // check if file is an exact duplicate of current file version
 769                         if ( $hash === $localFile->getSha1() ) {
 770                                 $warnings['no-change'] = $localFile;
 771                         }
 772
 773                         // check if file is an exact duplicate of older versions of this file
 774                         $history = $localFile->getHistory();
 775                         foreach ( $history as $oldFile ) {
 776                                 if ( $hash === $oldFile->getSha1() ) {
 777                                         $warnings['duplicate-version'][] = $oldFile;
 778                                 }
 779                         }
 780                 }
 781
 782                 return $warnings;
 783         }
 784
 785         private function checkLocalFileWasDeleted( LocalFile $localFile ) {
 786                 return $localFile->wasDeleted() && !$localFile->exists();
 787         }
 788
 789         /**
 790          * @param string $hash sha1 hash of the file to check
 791          * @param bool $ignoreLocalDupes True to ignore local duplicates
 792          *
 793          * @return File[] Duplicate files, if found.
 794          */
 795         private function checkAgainstExistingDupes( $hash, $ignoreLocalDupes ) {
 796                 $dupes = RepoGroup::singleton()->findBySha1( $hash );
 797                 $title = $this->getTitle();
 798                 foreach ( $dupes as $key => $dupe ) {
 799                         if (
 800                                 ( $dupe instanceof LocalFile ) &&
 801                                 $ignoreLocalDupes &&
 802                                 $title->equals( $dupe->getTitle() )
 803                         ) {
 804                                 unset( $dupes[$key] );
 805                         }
 806                 }
 807
 808                 return $dupes;
 809         }
 810
 811         /**
 812          * @param string $hash sha1 hash of the file to check
 813          *
 814          * @return string|null Name of the dupe or empty string if discovered (depending on visibility)
 815          *                     null if the check discovered no dupes.
 816          */
 817         private function checkAgainstArchiveDupes( $hash ) {
 818                 $archivedFile = new ArchivedFile( null, 0, '', $hash );
 819                 if ( $archivedFile->getID() > 0 ) {
 820                         if ( $archivedFile->userCan( File::DELETED_FILE ) ) {
 821                                 return $archivedFile->getName();
 822                         } else {
 823                                 return '';
 824                         }
 825                 }
 826
 827                 return null;
 828         }
 829
 830         /**
 831          * Really perform the upload. Stores the file in the local repo, watches
 832          * if necessary and runs the UploadComplete hook.
 833          *
 834          * @param string $comment
 835          * @param string $pageText
 836          * @param bool $watch Whether the file page should be added to user's watchlist.
 837          *   (This doesn't check $user's permissions.)
 838          * @param User $user
 839          * @param string[] $tags Change tags to add to the log entry and page revision.
 840          *   (This doesn't check $user's permissions.)
 841          * @return Status Indicating the whether the upload succeeded.
 842          */
 843         public function performUpload( $comment, $pageText, $watch, $user, $tags = [] ) {
 844                 $this->getLocalFile()->load( File::READ_LATEST );
 845                 $props = $this->mFileProps;
 846
 847                 $error = null;
 848                 Hooks::run( 'UploadVerifyUpload', [ $this, $user, $props, $comment, $pageText, &$error ] );
 849                 if ( $error ) {
 850                         if ( !is_array( $error ) ) {
 851                                 $error = [ $error ];
 852                         }
 853                         return Status::newFatal( ...$error );
 854                 }
 855
 856                 $status = $this->getLocalFile()->upload(
 857                         $this->mTempPath,
 858                         $comment,
 859                         $pageText,
 860                         File::DELETE_SOURCE,
 861                         $props,
 862                         false,
 863                         $user,
 864                         $tags
 865                 );
 866
 867                 if ( $status->isGood() ) {
 868                         if ( $watch ) {
 869                                 WatchAction::doWatch(
 870                                         $this->getLocalFile()->getTitle(),
 871                                         $user,
 872                                         User::IGNORE_USER_RIGHTS
 873                                 );
 874                         }
 875                         // Avoid PHP 7.1 warning of passing $this by reference
 876                         $uploadBase = $this;
 877                         Hooks::run( 'UploadComplete', [ &$uploadBase ] );
 878
 879                         $this->postProcessUpload();
 880                 }
 881
 882                 return $status;
 883         }
 884
 885         /**
 886          * Perform extra steps after a successful upload.
 887          *
 888          * @since  1.25
 889          */
 890         public function postProcessUpload() {
 891         }
 892
 893         /**
 894          * Returns the title of the file to be uploaded. Sets mTitleError in case
 895          * the name was illegal.
 896          *
 897          * @return Title|null The title of the file or null in case the name was illegal
 898          */
 899         public function getTitle() {
 900                 if ( $this->mTitle !== false ) {
 901                         return $this->mTitle;
 902                 }
 903                 if ( !is_string( $this->mDesiredDestName ) ) {
 904                         $this->mTitleError = self::ILLEGAL_FILENAME;
 905                         $this->mTitle = null;
 906
 907                         return $this->mTitle;
 908                 }
 909                 /* Assume that if a user specified File:Something.jpg, this is an error
 910                  * and that the namespace prefix needs to be stripped of.
 911                  */
 912                 $title = Title::newFromText( $this->mDesiredDestName );
 913                 if ( $title && $title->getNamespace() == NS_FILE ) {
 914                         $this->mFilteredName = $title->getDBkey();
 915                 } else {
 916                         $this->mFilteredName = $this->mDesiredDestName;
 917                 }
 918
 919                 # oi_archive_name is max 255 bytes, which include a timestamp and an
 920                 # exclamation mark, so restrict file name to 240 bytes.
 921                 if ( strlen( $this->mFilteredName ) > 240 ) {
 922                         $this->mTitleError = self::FILENAME_TOO_LONG;
 923                         $this->mTitle = null;
 924
 925                         return $this->mTitle;
 926                 }
 927
 928                 /**
 929                  * Chop off any directories in the given filename. Then
 930                  * filter out illegal characters, and try to make a legible name
 931                  * out of it. We'll strip some silently that Title would die on.
 932                  */
 933                 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mFilteredName );
 934                 /* Normalize to title form before we do any further processing */
 935                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 936                 if ( is_null( $nt ) ) {
 937                         $this->mTitleError = self::ILLEGAL_FILENAME;
 938                         $this->mTitle = null;
 939
 940                         return $this->mTitle;
 941                 }
 942                 $this->mFilteredName = $nt->getDBkey();
 943
 944                 /**
 945                  * We'll want to blacklist against *any* 'extension', and use
 946                  * only the final one for the whitelist.
 947                  */
 948                 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
 949
 950                 if ( $ext !== [] ) {
 951                         $this->mFinalExtension = trim( end( $ext ) );
 952                 } else {
 953                         $this->mFinalExtension = '';
 954
 955                         # No extension, try guessing one
 956                         $magic = MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer();
 957                         $mime = $magic->guessMimeType( $this->mTempPath );
 958                         if ( $mime !== 'unknown/unknown' ) {
 959                                 # Get a space separated list of extensions
 960                                 $extList = $magic->getExtensionsForType( $mime );
 961                                 if ( $extList ) {
 962                                         # Set the extension to the canonical extension
 963                                         $this->mFinalExtension = strtok( $extList, ' ' );
 964
 965                                         # Fix up the other variables
 966                                         $this->mFilteredName .= ".{$this->mFinalExtension}";
 967                                         $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 968                                         $ext = [ $this->mFinalExtension ];
 969                                 }
 970                         }
 971                 }
 972
 973                 /* Don't allow users to override the blacklist (check file extension) */
 974                 global $wgCheckFileExtensions, $wgStrictFileExtensions;
 975                 global $wgFileExtensions, $wgFileBlacklist;
 976
 977                 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
 978
 979                 if ( $this->mFinalExtension == '' ) {
 980                         $this->mTitleError = self::FILETYPE_MISSING;
 981                         $this->mTitle = null;
 982
 983                         return $this->mTitle;
 984                 } elseif ( $blackListedExtensions ||
 985                         ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
 986                                 !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) )
 987                 ) {
 988                         $this->mBlackListedExtensions = $blackListedExtensions;
 989                         $this->mTitleError = self::FILETYPE_BADTYPE;
 990                         $this->mTitle = null;
 991
 992                         return $this->mTitle;
 993                 }
 994
 995                 // Windows may be broken with special characters, see T3780
 996                 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
 997                         && !RepoGroup::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
 998                 ) {
 999                         $this->mTitleError = self::WINDOWS_NONASCII_FILENAME;
1000                         $this->mTitle = null;
1001
1002                         return $this->mTitle;
1003                 }
1004
1005                 # If there was more than one "extension", reassemble the base
1006                 # filename to prevent bogus complaints about length
1007                 if ( count( $ext ) > 1 ) {
1008                         $iterations = count( $ext ) - 1;
1009                         for ( $i = 0; $i < $iterations; $i++ ) {
1010                                 $partname .= '.' . $ext[$i];
1011                         }
1012                 }
1013
1014                 if ( strlen( $partname ) < 1 ) {
1015                         $this->mTitleError = self::MIN_LENGTH_PARTNAME;
1016                         $this->mTitle = null;
1017
1018                         return $this->mTitle;
1019                 }
1020
1021                 $this->mTitle = $nt;
1022
1023                 return $this->mTitle;
1024         }
1025
1026         /**
1027          * Return the local file and initializes if necessary.
1028          *
1029          * @return LocalFile|null
1030          */
1031         public function getLocalFile() {
1032                 if ( is_null( $this->mLocalFile ) ) {
1033                         $nt = $this->getTitle();
1034                         $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
1035                 }
1036
1037                 return $this->mLocalFile;
1038         }
1039
1040         /**
1041          * @return UploadStashFile|null
1042          */
1043         public function getStashFile() {
1044                 return $this->mStashFile;
1045         }
1046
1047         /**
1048          * Like stashFile(), but respects extensions' wishes to prevent the stashing. verifyUpload() must
1049          * be called before calling this method (unless $isPartial is true).
1050          *
1051          * Upload stash exceptions are also caught and converted to an error status.
1052          *
1053          * @since 1.28
1054          * @param User $user
1055          * @param bool $isPartial Pass `true` if this is a part of a chunked upload (not a complete file).
1056          * @return Status If successful, value is an UploadStashFile instance
1057          */
1058         public function tryStashFile( User $user, $isPartial = false ) {
1059                 if ( !$isPartial ) {
1060                         $error = $this->runUploadStashFileHook( $user );
1061                         if ( $error ) {
1062                                 return Status::newFatal( ...$error );
1063                         }
1064                 }
1065                 try {
1066                         $file = $this->doStashFile( $user );
1067                         return Status::newGood( $file );
1068                 } catch ( UploadStashException $e ) {
1069                         return Status::newFatal( 'uploadstash-exception', get_class( $e ), $e->getMessage() );
1070                 }
1071         }
1072
1073         /**
1074          * @param User $user
1075          * @return array|null Error message and parameters, null if there's no error
1076          */
1077         protected function runUploadStashFileHook( User $user ) {
1078                 $props = $this->mFileProps;
1079                 $error = null;
1080                 Hooks::run( 'UploadStashFile', [ $this, $user, $props, &$error ] );
1081                 if ( $error && !is_array( $error ) ) {
1082                         $error = [ $error ];
1083                 }
1084                 return $error;
1085         }
1086
1087         /**
1088          * If the user does not supply all necessary information in the first upload
1089          * form submission (either by accident or by design) then we may want to
1090          * stash the file temporarily, get more information, and publish the file
1091          * later.
1092          *
1093          * This method will stash a file in a temporary directory for later
1094          * processing, and save the necessary descriptive info into the database.
1095          * This method returns the file object, which also has a 'fileKey' property
1096          * which can be passed through a form or API request to find this stashed
1097          * file again.
1098          *
1099          * @deprecated since 1.28 Use tryStashFile() instead
1100          * @param User|null $user
1101          * @return UploadStashFile Stashed file
1102          * @throws UploadStashBadPathException
1103          * @throws UploadStashFileException
1104          * @throws UploadStashNotLoggedInException
1105          */
1106         public function stashFile( User $user = null ) {
1107                 return $this->doStashFile( $user );
1108         }
1109
1110         /**
1111          * Implementation for stashFile() and tryStashFile().
1112          *
1113          * @param User|null $user
1114          * @return UploadStashFile Stashed file
1115          */
1116         protected function doStashFile( User $user = null ) {
1117                 $stash = RepoGroup::singleton()->getLocalRepo()->getUploadStash( $user );
1118                 $file = $stash->stashFile( $this->mTempPath, $this->getSourceType() );
1119                 $this->mStashFile = $file;
1120
1121                 return $file;
1122         }
1123
1124         /**
1125          * Stash a file in a temporary directory, returning a key which can be used
1126          * to find the file again. See stashFile().
1127          *
1128          * @deprecated since 1.28
1129          * @return string File key
1130          */
1131         public function stashFileGetKey() {
1132                 wfDeprecated( __METHOD__, '1.28' );
1133                 return $this->doStashFile()->getFileKey();
1134         }
1135
1136         /**
1137          * alias for stashFileGetKey, for backwards compatibility
1138          *
1139          * @deprecated since 1.28
1140          * @return string File key
1141          */
1142         public function stashSession() {
1143                 wfDeprecated( __METHOD__, '1.28' );
1144                 return $this->doStashFile()->getFileKey();
1145         }
1146
1147         /**
1148          * If we've modified the upload file we need to manually remove it
1149          * on exit to clean up.
1150          */
1151         public function cleanupTempFile() {
1152                 if ( $this->mRemoveTempFile && $this->tempFileObj ) {
1153                         // Delete when all relevant TempFSFile handles go out of scope
1154                         wfDebug( __METHOD__ . ": Marked temporary file '{$this->mTempPath}' for removal\n" );
1155                         $this->tempFileObj->autocollect();
1156                 }
1157         }
1158
1159         public function getTempPath() {
1160                 return $this->mTempPath;
1161         }
1162
1163         /**
1164          * Split a file into a base name and all dot-delimited 'extensions'
1165          * on the end. Some web server configurations will fall back to
1166          * earlier pseudo-'extensions' to determine type and execute
1167          * scripts, so the blacklist needs to check them all.
1168          *
1169          * @param string $filename
1170          * @return array [ string, string[] ]
1171          */
1172         public static function splitExtensions( $filename ) {
1173                 $bits = explode( '.', $filename );
1174                 $basename = array_shift( $bits );
1175
1176                 return [ $basename, $bits ];
1177         }
1178
1179         /**
1180          * Perform case-insensitive match against a list of file extensions.
1181          * Returns true if the extension is in the list.
1182          *
1183          * @param string $ext
1184          * @param array $list
1185          * @return bool
1186          */
1187         public static function checkFileExtension( $ext, $list ) {
1188                 return in_array( strtolower( $ext ), $list );
1189         }
1190
1191         /**
1192          * Perform case-insensitive match against a list of file extensions.
1193          * Returns an array of matching extensions.
1194          *
1195          * @param string[] $ext
1196          * @param string[] $list
1197          * @return bool
1198          */
1199         public static function checkFileExtensionList( $ext, $list ) {
1200                 return array_intersect( array_map( 'strtolower', $ext ), $list );
1201         }
1202
1203         /**
1204          * Checks if the MIME type of the uploaded file matches the file extension.
1205          *
1206          * @param string $mime The MIME type of the uploaded file
1207          * @param string $extension The filename extension that the file is to be served with
1208          * @return bool
1209          */
1210         public static function verifyExtension( $mime, $extension ) {
1211                 $magic = MediaWiki\MediaWikiServices::getInstance()->getMimeAnalyzer();
1212
1213                 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' ) {
1214                         if ( !$magic->isRecognizableExtension( $extension ) ) {
1215                                 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
1216                                         "unrecognized extension '$extension', can't verify\n" );
1217
1218                                 return true;
1219                         } else {
1220                                 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; " .
1221                                         "recognized extension '$extension', so probably invalid file\n" );
1222
1223                                 return false;
1224                         }
1225                 }
1226
1227                 $match = $magic->isMatchingExtension( $extension, $mime );
1228
1229                 if ( $match === null ) {
1230                         if ( $magic->getTypesForExtension( $extension ) !== null ) {
1231                                 wfDebug( __METHOD__ . ": No extension known for $mime, but we know a mime for $extension\n" );
1232
1233                                 return false;
1234                         } else {
1235                                 wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
1236
1237                                 return true;
1238                         }
1239                 } elseif ( $match === true ) {
1240                         wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
1241
1242                         /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1243                         return true;
1244                 } else {
1245                         wfDebug( __METHOD__
1246                                 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1247
1248                         return false;
1249                 }
1250         }
1251
1252         /**
1253          * Heuristic for detecting files that *could* contain JavaScript instructions or
1254          * things that may look like HTML to a browser and are thus
1255          * potentially harmful. The present implementation will produce false
1256          * positives in some situations.
1257          *
1258          * @param string $file Pathname to the temporary upload file
1259          * @param string $mime The MIME type of the file
1260          * @param string $extension The extension of the file
1261          * @return bool True if the file contains something looking like embedded scripts
1262          */
1263         public static function detectScript( $file, $mime, $extension ) {
1264                 global $wgAllowTitlesInSVG;
1265
1266                 # ugly hack: for text files, always look at the entire file.
1267                 # For binary field, just check the first K.
1268
1269                 if ( strpos( $mime, 'text/' ) === 0 ) {
1270                         $chunk = file_get_contents( $file );
1271                 } else {
1272                         $fp = fopen( $file, 'rb' );
1273                         $chunk = fread( $fp, 1024 );
1274                         fclose( $fp );
1275                 }
1276
1277                 $chunk = strtolower( $chunk );
1278
1279                 if ( !$chunk ) {
1280                         return false;
1281                 }
1282
1283                 # decode from UTF-16 if needed (could be used for obfuscation).
1284                 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1285                         $enc = 'UTF-16BE';
1286                 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1287                         $enc = 'UTF-16LE';
1288                 } else {
1289                         $enc = null;
1290                 }
1291
1292                 if ( $enc ) {
1293                         $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1294                 }
1295
1296                 $chunk = trim( $chunk );
1297
1298                 /** @todo FIXME: Convert from UTF-16 if necessary! */
1299                 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
1300
1301                 # check for HTML doctype
1302                 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1303                         return true;
1304                 }
1305
1306                 // Some browsers will interpret obscure xml encodings as UTF-8, while
1307                 // PHP/expat will interpret the given encoding in the xml declaration (T49304)
1308                 if ( $extension == 'svg' || strpos( $mime, 'image/svg' ) === 0 ) {
1309                         if ( self::checkXMLEncodingMissmatch( $file ) ) {
1310                                 return true;
1311                         }
1312                 }
1313
1314                 /**
1315                  * Internet Explorer for Windows performs some really stupid file type
1316                  * autodetection which can cause it to interpret valid image files as HTML
1317                  * and potentially execute JavaScript, creating a cross-site scripting
1318                  * attack vectors.
1319                  *
1320                  * Apple's Safari browser also performs some unsafe file type autodetection
1321                  * which can cause legitimate files to be interpreted as HTML if the
1322                  * web server is not correctly configured to send the right content-type
1323                  * (or if you're really uploading plain text and octet streams!)
1324                  *
1325                  * Returns true if IE is likely to mistake the given file for HTML.
1326                  * Also returns true if Safari would mistake the given file for HTML
1327                  * when served with a generic content-type.
1328                  */
1329                 $tags = [
1330                         '<a href',
1331                         '<body',
1332                         '<head',
1333                         '<html', # also in safari
1334                         '<img',
1335                         '<pre',
1336                         '<script', # also in safari
1337                         '<table'
1338                 ];
1339
1340                 if ( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
1341                         $tags[] = '<title';
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 mixed False of the file is verified (does not contain scripts), array otherwise.
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 (true if the filter identified something bad)
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 array|null $data
1527          * @return bool
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 mixed 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 .= " " . wfEscapeShellArg( $file );
1867                 } else {
1868                         # complex pattern: replace "%f" with file to scan
1869                         $command = str_replace( "%f", wfEscapeShellArg( $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 mixed True on success, array on failure
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                 $cache = MediaWikiServices::getInstance()->getMainObjectStash();
2210                 $key = $cache->makeKey( 'uploadstatus', $user->getId() ?: md5( $user->getName() ), $statusKey );
2211
2212                 return $cache->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          * @param User $user
2221          * @param string $statusKey
2222          * @param array|bool $value
2223          * @return void
2224          */
2225         public static function setSessionStatus( User $user, $statusKey, $value ) {
2226                 $cache = MediaWikiServices::getInstance()->getMainObjectStash();
2227                 $key = $cache->makeKey( 'uploadstatus', $user->getId() ?: md5( $user->getName() ), $statusKey );
2228
2229                 if ( $value === false ) {
2230                         $cache->delete( $key );
2231                 } else {
2232                         $cache->set( $key, $value, $cache::TTL_DAY );
2233                 }
2234         }
2235 }