includes/upload/UploadBase.php

   1 <?php
   2 /**
   3  * Base class for the backend of file upload.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Upload
  22  */
  23 use MediaWiki\MediaWikiServices;
  24
  25 /**
  26  * @defgroup Upload Upload related
  27  */
  28
  29 /**
  30  * @ingroup Upload
  31  *
  32  * UploadBase and subclasses are the backend of MediaWiki's file uploads.
  33  * The frontends are formed by ApiUpload and SpecialUpload.
  34  *
  35  * @author Brion Vibber
  36  * @author Bryan Tong Minh
  37  * @author Michael Dale
  38  */
  39 abstract class UploadBase {
  40         /** @var string Local file system path to the file to upload (or a local copy) */
  41         protected $mTempPath;
  42         /** @var TempFSFile|null Wrapper to handle deleting the temp file */
  43         protected $tempFileObj;
  44
  45         protected $mDesiredDestName, $mDestName, $mRemoveTempFile, $mSourceType;
  46         protected $mTitle = false, $mTitleError = 0;
  47         protected $mFilteredName, $mFinalExtension;
  48         protected $mLocalFile, $mStashFile, $mFileSize, $mFileProps;
  49         protected $mBlackListedExtensions;
  50         protected $mJavaDetected, $mSVGNSError;
  51
  52         protected static $safeXmlEncodings = [
  53                 'UTF-8',
  54                 'ISO-8859-1',
  55                 'ISO-8859-2',
  56                 'UTF-16',
  57                 'UTF-32',
  58                 'WINDOWS-1250',
  59                 'WINDOWS-1251',
  60                 'WINDOWS-1252',
  61                 'WINDOWS-1253',
  62                 'WINDOWS-1254',
  63                 'WINDOWS-1255',
  64                 'WINDOWS-1256',
  65                 'WINDOWS-1257',
  66                 'WINDOWS-1258',
  67         ];
  68
  69         const SUCCESS = 0;
  70         const OK = 0;
  71         const EMPTY_FILE = 3;
  72         const MIN_LENGTH_PARTNAME = 4;
  73         const ILLEGAL_FILENAME = 5;
  74         const OVERWRITE_EXISTING_FILE = 7; # Not used anymore; handled by verifyTitlePermissions()
  75         const FILETYPE_MISSING = 8;
  76         const FILETYPE_BADTYPE = 9;
  77         const VERIFICATION_ERROR = 10;
  78         const HOOK_ABORTED = 11;
  79         const FILE_TOO_LARGE = 12;
  80         const WINDOWS_NONASCII_FILENAME = 13;
  81         const FILENAME_TOO_LONG = 14;
  82
  83         /**
  84          * @param int $error
  85          * @return string
  86          */
  87         public function getVerificationErrorCode( $error ) {
  88                 $code_to_status = [
  89                         self::EMPTY_FILE => 'empty-file',
  90                         self::FILE_TOO_LARGE => 'file-too-large',
  91                         self::FILETYPE_MISSING => 'filetype-missing',
  92                         self::FILETYPE_BADTYPE => 'filetype-banned',
  93                         self::MIN_LENGTH_PARTNAME => 'filename-tooshort',
  94                         self::ILLEGAL_FILENAME => 'illegal-filename',
  95                         self::OVERWRITE_EXISTING_FILE => 'overwrite',
  96                         self::VERIFICATION_ERROR => 'verification-error',
  97                         self::HOOK_ABORTED => 'hookaborted',
  98                         self::WINDOWS_NONASCII_FILENAME => 'windows-nonascii-filename',
  99                         self::FILENAME_TOO_LONG => 'filename-toolong',
 100                 ];
 101                 if ( isset( $code_to_status[$error] ) ) {
 102                         return $code_to_status[$error];
 103                 }
 104
 105                 return 'unknown-error';
 106         }
 107
 108         /**
 109          * Returns true if uploads are enabled.
 110          * Can be override by subclasses.
 111          * @return bool
 112          */
 113         public static function isEnabled() {
 114                 global $wgEnableUploads;
 115
 116                 if ( !$wgEnableUploads ) {
 117                         return false;
 118                 }
 119
 120                 # Check php's file_uploads setting
 121                 return wfIsHHVM() || wfIniGetBool( 'file_uploads' );
 122         }
 123
 124         /**
 125          * Returns true if the user can use this upload module or else a string
 126          * identifying the missing permission.
 127          * Can be overridden by subclasses.
 128          *
 129          * @param User $user
 130          * @return bool|string
 131          */
 132         public static function isAllowed( $user ) {
 133                 foreach ( [ 'upload', 'edit' ] as $permission ) {
 134                         if ( !$user->isAllowed( $permission ) ) {
 135                                 return $permission;
 136                         }
 137                 }
 138
 139                 return true;
 140         }
 141
 142         /**
 143          * Returns true if the user has surpassed the upload rate limit, false otherwise.
 144          *
 145          * @param User $user
 146          * @return bool
 147          */
 148         public static function isThrottled( $user ) {
 149                 return $user->pingLimiter( 'upload' );
 150         }
 151
 152         // Upload handlers. Should probably just be a global.
 153         private static $uploadHandlers = [ 'Stash', 'File', 'Url' ];
 154
 155         /**
 156          * Create a form of UploadBase depending on wpSourceType and initializes it
 157          *
 158          * @param WebRequest $request
 159          * @param string|null $type
 160          * @return null|UploadBase
 161          */
 162         public static function createFromRequest( &$request, $type = null ) {
 163                 $type = $type ? $type : $request->getVal( 'wpSourceType', 'File' );
 164
 165                 if ( !$type ) {
 166                         return null;
 167                 }
 168
 169                 // Get the upload class
 170                 $type = ucfirst( $type );
 171
 172                 // Give hooks the chance to handle this request
 173                 $className = null;
 174                 Hooks::run( 'UploadCreateFromRequest', [ $type, &$className ] );
 175                 if ( is_null( $className ) ) {
 176                         $className = 'UploadFrom' . $type;
 177                         wfDebug( __METHOD__ . ": class name: $className\n" );
 178                         if ( !in_array( $type, self::$uploadHandlers ) ) {
 179                                 return null;
 180                         }
 181                 }
 182
 183                 // Check whether this upload class is enabled
 184                 if ( !call_user_func( [ $className, 'isEnabled' ] ) ) {
 185                         return null;
 186                 }
 187
 188                 // Check whether the request is valid
 189                 if ( !call_user_func( [ $className, 'isValidRequest' ], $request ) ) {
 190                         return null;
 191                 }
 192
 193                 /** @var UploadBase $handler */
 194                 $handler = new $className;
 195
 196                 $handler->initializeFromRequest( $request );
 197
 198                 return $handler;
 199         }
 200
 201         /**
 202          * Check whether a request if valid for this handler
 203          * @param WebRequest $request
 204          * @return bool
 205          */
 206         public static function isValidRequest( $request ) {
 207                 return false;
 208         }
 209
 210         public function __construct() {
 211         }
 212
 213         /**
 214          * Returns the upload type. Should be overridden by child classes
 215          *
 216          * @since 1.18
 217          * @return string
 218          */
 219         public function getSourceType() {
 220                 return null;
 221         }
 222
 223         /**
 224          * Initialize the path information
 225          * @param string $name The desired destination name
 226          * @param string $tempPath The temporary path
 227          * @param int $fileSize The file size
 228          * @param bool $removeTempFile (false) remove the temporary file?
 229          * @throws MWException
 230          */
 231         public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
 232                 $this->mDesiredDestName = $name;
 233                 if ( FileBackend::isStoragePath( $tempPath ) ) {
 234                         throw new MWException( __METHOD__ . " given storage path `$tempPath`." );
 235                 }
 236
 237                 $this->setTempFile( $tempPath, $fileSize );
 238                 $this->mRemoveTempFile = $removeTempFile;
 239         }
 240
 241         /**
 242          * Initialize from a WebRequest. Override this in a subclass.
 243          *
 244          * @param WebRequest $request
 245          */
 246         abstract public function initializeFromRequest( &$request );
 247
 248         /**
 249          * @param string $tempPath File system path to temporary file containing the upload
 250          * @param integer $fileSize
 251          */
 252         protected function setTempFile( $tempPath, $fileSize = null ) {
 253                 $this->mTempPath = $tempPath;
 254                 $this->mFileSize = $fileSize ?: null;
 255                 if ( strlen( $this->mTempPath ) && file_exists( $this->mTempPath ) ) {
 256                         $this->tempFileObj = new TempFSFile( $this->mTempPath );
 257                         if ( !$fileSize ) {
 258                                 $this->mFileSize = filesize( $this->mTempPath );
 259                         }
 260                 } else {
 261                         $this->tempFileObj = null;
 262                 }
 263         }
 264
 265         /**
 266          * Fetch the file. Usually a no-op
 267          * @return Status
 268          */
 269         public function fetchFile() {
 270                 return Status::newGood();
 271         }
 272
 273         /**
 274          * Return true if the file is empty
 275          * @return bool
 276          */
 277         public function isEmptyFile() {
 278                 return empty( $this->mFileSize );
 279         }
 280
 281         /**
 282          * Return the file size
 283          * @return int
 284          */
 285         public function getFileSize() {
 286                 return $this->mFileSize;
 287         }
 288
 289         /**
 290          * Get the base 36 SHA1 of the file
 291          * @return string
 292          */
 293         public function getTempFileSha1Base36() {
 294                 return FSFile::getSha1Base36FromPath( $this->mTempPath );
 295         }
 296
 297         /**
 298          * @param string $srcPath The source path
 299          * @return string|bool The real path if it was a virtual URL Returns false on failure
 300          */
 301         public function getRealPath( $srcPath ) {
 302                 $repo = RepoGroup::singleton()->getLocalRepo();
 303                 if ( $repo->isVirtualUrl( $srcPath ) ) {
 304                         /** @todo Just make uploads work with storage paths UploadFromStash
 305                          *  loads files via virtual URLs.
 306                          */
 307                         $tmpFile = $repo->getLocalCopy( $srcPath );
 308                         if ( $tmpFile ) {
 309                                 $tmpFile->bind( $this ); // keep alive with $this
 310                         }
 311                         $path = $tmpFile ? $tmpFile->getPath() : false;
 312                 } else {
 313                         $path = $srcPath;
 314                 }
 315
 316                 return $path;
 317         }
 318
 319         /**
 320          * Verify whether the upload is sane.
 321          * @return mixed Const self::OK or else an array with error information
 322          */
 323         public function verifyUpload() {
 324
 325                 /**
 326                  * If there was no filename or a zero size given, give up quick.
 327                  */
 328                 if ( $this->isEmptyFile() ) {
 329                         return [ 'status' => self::EMPTY_FILE ];
 330                 }
 331
 332                 /**
 333                  * Honor $wgMaxUploadSize
 334                  */
 335                 $maxSize = self::getMaxUploadSize( $this->getSourceType() );
 336                 if ( $this->mFileSize > $maxSize ) {
 337                         return [
 338                                 'status' => self::FILE_TOO_LARGE,
 339                                 'max' => $maxSize,
 340                         ];
 341                 }
 342
 343                 /**
 344                  * Look at the contents of the file; if we can recognize the
 345                  * type but it's corrupt or data of the wrong type, we should
 346                  * probably not accept it.
 347                  */
 348                 $verification = $this->verifyFile();
 349                 if ( $verification !== true ) {
 350                         return [
 351                                 'status' => self::VERIFICATION_ERROR,
 352                                 'details' => $verification
 353                         ];
 354                 }
 355
 356                 /**
 357                  * Make sure this file can be created
 358                  */
 359                 $result = $this->validateName();
 360                 if ( $result !== true ) {
 361                         return $result;
 362                 }
 363
 364                 $error = '';
 365                 if ( !Hooks::run( 'UploadVerification',
 366                         [ $this->mDestName, $this->mTempPath, &$error ], '1.28' )
 367                 ) {
 368                         return [ 'status' => self::HOOK_ABORTED, 'error' => $error ];
 369                 }
 370
 371                 return [ 'status' => self::OK ];
 372         }
 373
 374         /**
 375          * Verify that the name is valid and, if necessary, that we can overwrite
 376          *
 377          * @return mixed True if valid, otherwise and array with 'status'
 378          * and other keys
 379          */
 380         public function validateName() {
 381                 $nt = $this->getTitle();
 382                 if ( is_null( $nt ) ) {
 383                         $result = [ 'status' => $this->mTitleError ];
 384                         if ( $this->mTitleError == self::ILLEGAL_FILENAME ) {
 385                                 $result['filtered'] = $this->mFilteredName;
 386                         }
 387                         if ( $this->mTitleError == self::FILETYPE_BADTYPE ) {
 388                                 $result['finalExt'] = $this->mFinalExtension;
 389                                 if ( count( $this->mBlackListedExtensions ) ) {
 390                                         $result['blacklistedExt'] = $this->mBlackListedExtensions;
 391                                 }
 392                         }
 393
 394                         return $result;
 395                 }
 396                 $this->mDestName = $this->getLocalFile()->getName();
 397
 398                 return true;
 399         }
 400
 401         /**
 402          * Verify the MIME type.
 403          *
 404          * @note Only checks that it is not an evil MIME. The "does it have
 405          *  correct extension given its MIME type?" check is in verifyFile.
 406          *  in `verifyFile()` that MIME type and file extension correlate.
 407          * @param string $mime Representing the MIME
 408          * @return mixed True if the file is verified, an array otherwise
 409          */
 410         protected function verifyMimeType( $mime ) {
 411                 global $wgVerifyMimeType;
 412                 if ( $wgVerifyMimeType ) {
 413                         wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>\n" );
 414                         global $wgMimeTypeBlacklist;
 415                         if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
 416                                 return [ 'filetype-badmime', $mime ];
 417                         }
 418
 419                         # Check what Internet Explorer would detect
 420                         $fp = fopen( $this->mTempPath, 'rb' );
 421                         $chunk = fread( $fp, 256 );
 422                         fclose( $fp );
 423
 424                         $magic = MimeMagic::singleton();
 425                         $extMime = $magic->guessTypesForExtension( $this->mFinalExtension );
 426                         $ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime );
 427                         foreach ( $ieTypes as $ieType ) {
 428                                 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
 429                                         return [ 'filetype-bad-ie-mime', $ieType ];
 430                                 }
 431                         }
 432                 }
 433
 434                 return true;
 435         }
 436
 437         /**
 438          * Verifies that it's ok to include the uploaded file
 439          *
 440          * @return mixed True of the file is verified, array otherwise.
 441          */
 442         protected function verifyFile() {
 443                 global $wgVerifyMimeType, $wgDisableUploadScriptChecks;
 444
 445                 $status = $this->verifyPartialFile();
 446                 if ( $status !== true ) {
 447                         return $status;
 448                 }
 449
 450                 $mwProps = new MWFileProps( MimeMagic::singleton() );
 451                 $this->mFileProps = $mwProps->getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 452                 $mime = $this->mFileProps['mime'];
 453
 454                 if ( $wgVerifyMimeType ) {
 455                         # XXX: Missing extension will be caught by validateName() via getTitle()
 456                         if ( $this->mFinalExtension != '' && !$this->verifyExtension( $mime, $this->mFinalExtension ) ) {
 457                                 return [ 'filetype-mime-mismatch', $this->mFinalExtension, $mime ];
 458                         }
 459                 }
 460
 461                 # check for htmlish code and javascript
 462                 if ( !$wgDisableUploadScriptChecks ) {
 463                         if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 464                                 $svgStatus = $this->detectScriptInSvg( $this->mTempPath, false );
 465                                 if ( $svgStatus !== false ) {
 466                                         return $svgStatus;
 467                                 }
 468                         }
 469                 }
 470
 471                 $handler = MediaHandler::getHandler( $mime );
 472                 if ( $handler ) {
 473                         $handlerStatus = $handler->verifyUpload( $this->mTempPath );
 474                         if ( !$handlerStatus->isOK() ) {
 475                                 $errors = $handlerStatus->getErrorsArray();
 476
 477                                 return reset( $errors );
 478                         }
 479                 }
 480
 481                 $error = true;
 482                 Hooks::run( 'UploadVerifyFile', [ $this, $mime, &$error ] );
 483                 if ( $error !== true ) {
 484                         if ( !is_array( $error ) ) {
 485                                 $error = [ $error ];
 486                         }
 487                         return $error;
 488                 }
 489
 490                 wfDebug( __METHOD__ . ": all clear; passing.\n" );
 491
 492                 return true;
 493         }
 494
 495         /**
 496          * A verification routine suitable for partial files
 497          *
 498          * Runs the blacklist checks, but not any checks that may
 499          * assume the entire file is present.
 500          *
 501          * @return mixed True for valid or array with error message key.
 502          */
 503         protected function verifyPartialFile() {
 504                 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
 505
 506                 # getTitle() sets some internal parameters like $this->mFinalExtension
 507                 $this->getTitle();
 508
 509                 $mwProps = new MWFileProps( MimeMagic::singleton() );
 510                 $this->mFileProps = $mwProps->getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 511
 512                 # check MIME type, if desired
 513                 $mime = $this->mFileProps['file-mime'];
 514                 $status = $this->verifyMimeType( $mime );
 515                 if ( $status !== true ) {
 516                         return $status;
 517                 }
 518
 519                 # check for htmlish code and javascript
 520                 if ( !$wgDisableUploadScriptChecks ) {
 521                         if ( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
 522                                 return [ 'uploadscripted' ];
 523                         }
 524                         if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 525                                 $svgStatus = $this->detectScriptInSvg( $this->mTempPath, true );
 526                                 if ( $svgStatus !== false ) {
 527                                         return $svgStatus;
 528                                 }
 529                         }
 530                 }
 531
 532                 # Check for Java applets, which if uploaded can bypass cross-site
 533                 # restrictions.
 534                 if ( !$wgAllowJavaUploads ) {
 535                         $this->mJavaDetected = false;
 536                         $zipStatus = ZipDirectoryReader::read( $this->mTempPath,
 537                                 [ $this, 'zipEntryCallback' ] );
 538                         if ( !$zipStatus->isOK() ) {
 539                                 $errors = $zipStatus->getErrorsArray();
 540                                 $error = reset( $errors );
 541                                 if ( $error[0] !== 'zip-wrong-format' ) {
 542                                         return $error;
 543                                 }
 544                         }
 545                         if ( $this->mJavaDetected ) {
 546                                 return [ 'uploadjava' ];
 547                         }
 548                 }
 549
 550                 # Scan the uploaded file for viruses
 551                 $virus = $this->detectVirus( $this->mTempPath );
 552                 if ( $virus ) {
 553                         return [ 'uploadvirus', $virus ];
 554                 }
 555
 556                 return true;
 557         }
 558
 559         /**
 560          * Callback for ZipDirectoryReader to detect Java class files.
 561          *
 562          * @param array $entry
 563          */
 564         public function zipEntryCallback( $entry ) {
 565                 $names = [ $entry['name'] ];
 566
 567                 // If there is a null character, cut off the name at it, because JDK's
 568                 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
 569                 // were constructed which had ".class\0" followed by a string chosen to
 570                 // make the hash collide with the truncated name, that file could be
 571                 // returned in response to a request for the .class file.
 572                 $nullPos = strpos( $entry['name'], "\000" );
 573                 if ( $nullPos !== false ) {
 574                         $names[] = substr( $entry['name'], 0, $nullPos );
 575                 }
 576
 577                 // If there is a trailing slash in the file name, we have to strip it,
 578                 // because that's what ZIP_GetEntry() does.
 579                 if ( preg_grep( '!\.class/?$!', $names ) ) {
 580                         $this->mJavaDetected = true;
 581                 }
 582         }
 583
 584         /**
 585          * Alias for verifyTitlePermissions. The function was originally
 586          * 'verifyPermissions', but that suggests it's checking the user, when it's
 587          * really checking the title + user combination.
 588          *
 589          * @param User $user User object to verify the permissions against
 590          * @return mixed An array as returned by getUserPermissionsErrors or true
 591          *   in case the user has proper permissions.
 592          */
 593         public function verifyPermissions( $user ) {
 594                 return $this->verifyTitlePermissions( $user );
 595         }
 596
 597         /**
 598          * Check whether the user can edit, upload and create the image. This
 599          * checks only against the current title; if it returns errors, it may
 600          * very well be that another title will not give errors. Therefore
 601          * isAllowed() should be called as well for generic is-user-blocked or
 602          * can-user-upload checking.
 603          *
 604          * @param User $user User object to verify the permissions against
 605          * @return mixed An array as returned by getUserPermissionsErrors or true
 606          *   in case the user has proper permissions.
 607          */
 608         public function verifyTitlePermissions( $user ) {
 609                 /**
 610                  * If the image is protected, non-sysop users won't be able
 611                  * to modify it by uploading a new revision.
 612                  */
 613                 $nt = $this->getTitle();
 614                 if ( is_null( $nt ) ) {
 615                         return true;
 616                 }
 617                 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
 618                 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
 619                 if ( !$nt->exists() ) {
 620                         $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
 621                 } else {
 622                         $permErrorsCreate = [];
 623                 }
 624                 if ( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
 625                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
 626                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
 627
 628                         return $permErrors;
 629                 }
 630
 631                 $overwriteError = $this->checkOverwrite( $user );
 632                 if ( $overwriteError !== true ) {
 633                         return [ $overwriteError ];
 634                 }
 635
 636                 return true;
 637         }
 638
 639         /**
 640          * Check for non fatal problems with the file.
 641          *
 642          * This should not assume that mTempPath is set.
 643          *
 644          * @return mixed[] Array of warnings
 645          */
 646         public function checkWarnings() {
 647                 $warnings = [];
 648
 649                 $localFile = $this->getLocalFile();
 650                 $localFile->load( File::READ_LATEST );
 651                 $filename = $localFile->getName();
 652                 $hash = $this->getTempFileSha1Base36();
 653
 654                 $badFileName = $this->checkBadFileName( $filename, $this->mDesiredDestName );
 655                 if ( $badFileName !== null ) {
 656                         $warnings['badfilename'] = $badFileName;
 657                 }
 658
 659                 $unwantedFileExtensionDetails = $this->checkUnwantedFileExtensions( $this->mFinalExtension );
 660                 if ( $unwantedFileExtensionDetails !== null ) {
 661                         $warnings['filetype-unwanted-type'] = $unwantedFileExtensionDetails;
 662                 }
 663
 664                 $fileSizeWarnings = $this->checkFileSize( $this->mFileSize );
 665                 if ( $fileSizeWarnings ) {
 666                         $warnings = array_merge( $warnings, $fileSizeWarnings );
 667                 }
 668
 669                 $localFileExistsWarnings = $this->checkLocalFileExists( $localFile, $hash );
 670                 if ( $localFileExistsWarnings ) {
 671                         $warnings = array_merge( $warnings, $localFileExistsWarnings );
 672                 }
 673
 674                 if ( $this->checkLocalFileWasDeleted( $localFile ) ) {
 675                         $warnings['was-deleted'] = $filename;
 676                 }
 677
 678                 $dupes = $this->checkAgainstExistingDupes( $hash );
 679                 if ( $dupes ) {
 680                         $warnings['duplicate'] = $dupes;
 681                 }
 682
 683                 $archivedDupes = $this->checkAgainstArchiveDupes( $hash );
 684                 if ( $archivedDupes !== null ) {
 685                         $warnings['duplicate-archive'] = $archivedDupes;
 686                 }
 687
 688                 return $warnings;
 689         }
 690
 691         /**
 692          * Check whether the resulting filename is different from the desired one,
 693          * but ignore things like ucfirst() and spaces/underscore things
 694          *
 695          * @param string $filename
 696          * @param string $desiredFileName
 697          *
 698          * @return string|null String that was determined to be bad or null if the filename is okay
 699          */
 700         private function checkBadFileName( $filename, $desiredFileName ) {
 701                 $comparableName = str_replace( ' ', '_', $desiredFileName );
 702                 $comparableName = Title::capitalize( $comparableName, NS_FILE );
 703
 704                 if ( $desiredFileName != $filename && $comparableName != $filename ) {
 705                         return $filename;
 706                 }
 707
 708                 return null;
 709         }
 710
 711         /**
 712          * @param string $fileExtension The file extension to check
 713          *
 714          * @return array|null array with the following keys:
 715          *                    0 => string The final extension being used
 716          *                    1 => string[] The extensions that are allowed
 717          *                    2 => int The number of extensions that are allowed.
 718          */
 719         private function checkUnwantedFileExtensions( $fileExtension ) {
 720                 global $wgCheckFileExtensions, $wgFileExtensions, $wgLang;
 721
 722                 if ( $wgCheckFileExtensions ) {
 723                         $extensions = array_unique( $wgFileExtensions );
 724                         if ( !$this->checkFileExtension( $fileExtension, $extensions ) ) {
 725                                 return [
 726                                         $fileExtension,
 727                                         $wgLang->commaList( $extensions ),
 728                                         count( $extensions )
 729                                 ];
 730                         }
 731                 }
 732
 733                 return null;
 734         }
 735
 736         /**
 737          * @param int $fileSize
 738          *
 739          * @return array warnings
 740          */
 741         private function checkFileSize( $fileSize ) {
 742                 global $wgUploadSizeWarning;
 743
 744                 $warnings = [];
 745
 746                 if ( $wgUploadSizeWarning && ( $fileSize > $wgUploadSizeWarning ) ) {
 747                         $warnings['large-file'] = [ $wgUploadSizeWarning, $fileSize ];
 748                 }
 749
 750                 if ( $fileSize == 0 ) {
 751                         $warnings['empty-file'] = true;
 752                 }
 753
 754                 return $warnings;
 755         }
 756
 757         /**
 758          * @param LocalFile $localFile
 759          * @param string $hash sha1 hash of the file to check
 760          *
 761          * @return array warnings
 762          */
 763         private function checkLocalFileExists( LocalFile $localFile, $hash ) {
 764                 $warnings = [];
 765
 766                 $exists = self::getExistsWarning( $localFile );
 767                 if ( $exists !== false ) {
 768                         $warnings['exists'] = $exists;
 769
 770                         // check if file is an exact duplicate of current file version
 771                         if ( $hash === $localFile->getSha1() ) {
 772                                 $warnings['no-change'] = $localFile;
 773                         }
 774
 775                         // check if file is an exact duplicate of older versions of this file
 776                         $history = $localFile->getHistory();
 777                         foreach ( $history as $oldFile ) {
 778                                 if ( $hash === $oldFile->getSha1() ) {
 779                                         $warnings['duplicate-version'][] = $oldFile;
 780                                 }
 781                         }
 782                 }
 783
 784                 return $warnings;
 785         }
 786
 787         private function checkLocalFileWasDeleted( LocalFile $localFile ) {
 788                 return $localFile->wasDeleted() && !$localFile->exists();
 789         }
 790
 791         /**
 792          * @param string $hash sha1 hash of the file to check
 793          *
 794          * @return File[] Duplicate files, if found.
 795          */
 796         private function checkAgainstExistingDupes( $hash ) {
 797                 $dupes = RepoGroup::singleton()->findBySha1( $hash );
 798                 $title = $this->getTitle();
 799                 // Remove all matches against self
 800                 foreach ( $dupes as $key => $dupe ) {
 801                         if ( $title->equals( $dupe->getTitle() ) ) {
 802                                 unset( $dupes[$key] );
 803                         }
 804                 }
 805
 806                 return $dupes;
 807         }
 808
 809         /**
 810          * @param string $hash sha1 hash of the file to check
 811          *
 812          * @return string|null Name of the dupe or empty string if discovered (depending on visibility)
 813          *                     null if the check discovered no dupes.
 814          */
 815         private function checkAgainstArchiveDupes( $hash ) {
 816                 $archivedFile = new ArchivedFile( null, 0, '', $hash );
 817                 if ( $archivedFile->getID() > 0 ) {
 818                         if ( $archivedFile->userCan( File::DELETED_FILE ) ) {
 819                                 return $archivedFile->getName();
 820                         } else {
 821                                 return '';
 822                         }
 823                 }
 824
 825                 return null;
 826         }
 827
 828         /**
 829          * Really perform the upload. Stores the file in the local repo, watches
 830          * if necessary and runs the UploadComplete hook.
 831          *
 832          * @param string $comment
 833          * @param string $pageText
 834          * @param bool $watch Whether the file page should be added to user's watchlist.
 835          *   (This doesn't check $user's permissions.)
 836          * @param User $user
 837          * @param string[] $tags Change tags to add to the log entry and page revision.
 838          *   (This doesn't check $user's permissions.)
 839          * @return Status Indicating the whether the upload succeeded.
 840          */
 841         public function performUpload( $comment, $pageText, $watch, $user, $tags = [] ) {
 842                 $this->getLocalFile()->load( File::READ_LATEST );
 843                 $props = $this->mFileProps;
 844
 845                 $error = null;
 846                 Hooks::run( 'UploadVerifyUpload', [ $this, $user, $props, $comment, $pageText, &$error ] );
 847                 if ( $error ) {
 848                         if ( !is_array( $error ) ) {
 849                                 $error = [ $error ];
 850                         }
 851                         return call_user_func_array( 'Status::newFatal', $error );
 852                 }
 853
 854                 $status = $this->getLocalFile()->upload(
 855                         $this->mTempPath,
 856                         $comment,
 857                         $pageText,
 858                         File::DELETE_SOURCE,
 859                         $props,
 860                         false,
 861                         $user,
 862                         $tags
 863                 );
 864
 865                 if ( $status->isGood() ) {
 866                         if ( $watch ) {
 867                                 WatchAction::doWatch(
 868                                         $this->getLocalFile()->getTitle(),
 869                                         $user,
 870                                         User::IGNORE_USER_RIGHTS
 871                                 );
 872                         }
 873                         // Avoid PHP 7.1 warning of passing $this by reference
 874                         $uploadBase = $this;
 875                         Hooks::run( 'UploadComplete', [ &$uploadBase ] );
 876
 877                         $this->postProcessUpload();
 878                 }
 879
 880                 return $status;
 881         }
 882
 883         /**
 884          * Perform extra steps after a successful upload.
 885          *
 886          * @since  1.25
 887          */
 888         public function postProcessUpload() {
 889         }
 890
 891         /**
 892          * Returns the title of the file to be uploaded. Sets mTitleError in case
 893          * the name was illegal.
 894          *
 895          * @return Title|null The title of the file or null in case the name was illegal
 896          */
 897         public function getTitle() {
 898                 if ( $this->mTitle !== false ) {
 899                         return $this->mTitle;
 900                 }
 901                 if ( !is_string( $this->mDesiredDestName ) ) {
 902                         $this->mTitleError = self::ILLEGAL_FILENAME;
 903                         $this->mTitle = null;
 904
 905                         return $this->mTitle;
 906                 }
 907                 /* Assume that if a user specified File:Something.jpg, this is an error
 908                  * and that the namespace prefix needs to be stripped of.
 909                  */
 910                 $title = Title::newFromText( $this->mDesiredDestName );
 911                 if ( $title && $title->getNamespace() == NS_FILE ) {
 912                         $this->mFilteredName = $title->getDBkey();
 913                 } else {
 914                         $this->mFilteredName = $this->mDesiredDestName;
 915                 }
 916
 917                 # oi_archive_name is max 255 bytes, which include a timestamp and an
 918                 # exclamation mark, so restrict file name to 240 bytes.
 919                 if ( strlen( $this->mFilteredName ) > 240 ) {
 920                         $this->mTitleError = self::FILENAME_TOO_LONG;
 921                         $this->mTitle = null;
 922
 923                         return $this->mTitle;
 924                 }
 925
 926                 /**
 927                  * Chop off any directories in the given filename. Then
 928                  * filter out illegal characters, and try to make a legible name
 929                  * out of it. We'll strip some silently that Title would die on.
 930                  */
 931                 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mFilteredName );
 932                 /* Normalize to title form before we do any further processing */
 933                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 934                 if ( is_null( $nt ) ) {
 935                         $this->mTitleError = self::ILLEGAL_FILENAME;
 936                         $this->mTitle = null;
 937
 938                         return $this->mTitle;
 939                 }
 940                 $this->mFilteredName = $nt->getDBkey();
 941
 942                 /**
 943                  * We'll want to blacklist against *any* 'extension', and use
 944                  * only the final one for the whitelist.
 945                  */
 946                 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
 947
 948                 if ( count( $ext ) ) {
 949                         $this->mFinalExtension = trim( $ext[count( $ext ) - 1] );
 950                 } else {
 951                         $this->mFinalExtension = '';
 952
 953                         # No extension, try guessing one
 954                         $magic = MimeMagic::singleton();
 955                         $mime = $magic->guessMimeType( $this->mTempPath );
 956                         if ( $mime !== 'unknown/unknown' ) {
 957                                 # Get a space separated list of extensions
 958                                 $extList = $magic->getExtensionsForType( $mime );
 959                                 if ( $extList ) {
 960                                         # Set the extension to the canonical extension
 961                                         $this->mFinalExtension = strtok( $extList, ' ' );
 962
 963                                         # Fix up the other variables
 964                                         $this->mFilteredName .= ".{$this->mFinalExtension}";
 965                                         $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 966                                         $ext = [ $this->mFinalExtension ];
 967                                 }
 968                         }
 969                 }
 970
 971                 /* Don't allow users to override the blacklist (check file extension) */
 972                 global $wgCheckFileExtensions, $wgStrictFileExtensions;
 973                 global $wgFileExtensions, $wgFileBlacklist;
 974
 975                 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
 976
 977                 if ( $this->mFinalExtension == '' ) {
 978                         $this->mTitleError = self::FILETYPE_MISSING;
 979                         $this->mTitle = null;
 980
 981                         return $this->mTitle;
 982                 } elseif ( $blackListedExtensions ||
 983                         ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
 984                                 !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) )
 985                 ) {
 986                         $this->mBlackListedExtensions = $blackListedExtensions;
 987                         $this->mTitleError = self::FILETYPE_BADTYPE;
 988                         $this->mTitle = null;
 989
 990                         return $this->mTitle;
 991                 }
 992
 993                 // Windows may be broken with special characters, see T3780
 994                 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
 995                         && !RepoGroup::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
 996                 ) {
 997                         $this->mTitleError = self::WINDOWS_NONASCII_FILENAME;
 998                         $this->mTitle = null;
 999
1000                         return $this->mTitle;
1001                 }
1002
1003                 # If there was more than one "extension", reassemble the base
1004                 # filename to prevent bogus complaints about length
1005                 if ( count( $ext ) > 1 ) {
1006                         $iterations = count( $ext ) - 1;
1007                         for ( $i = 0; $i < $iterations; $i++ ) {
1008                                 $partname .= '.' . $ext[$i];
1009                         }
1010                 }
1011
1012                 if ( strlen( $partname ) < 1 ) {
1013                         $this->mTitleError = self::MIN_LENGTH_PARTNAME;
1014                         $this->mTitle = null;
1015
1016                         return $this->mTitle;
1017                 }
1018
1019                 $this->mTitle = $nt;
1020
1021                 return $this->mTitle;
1022         }
1023
1024         /**
1025          * Return the local file and initializes if necessary.
1026          *
1027          * @return LocalFile|null
1028          */
1029         public function getLocalFile() {
1030                 if ( is_null( $this->mLocalFile ) ) {
1031                         $nt = $this->getTitle();
1032                         $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
1033                 }
1034
1035                 return $this->mLocalFile;
1036         }
1037
1038         /**
1039          * @return UploadStashFile|null
1040          */
1041         public function getStashFile() {
1042                 return $this->mStashFile;
1043         }
1044
1045         /**
1046          * Like stashFile(), but respects extensions' wishes to prevent the stashing. verifyUpload() must
1047          * be called before calling this method (unless $isPartial is true).
1048          *
1049          * Upload stash exceptions are also caught and converted to an error status.
1050          *
1051          * @since 1.28
1052          * @param User $user
1053          * @param bool $isPartial Pass `true` if this is a part of a chunked upload (not a complete file).
1054          * @return Status If successful, value is an UploadStashFile instance
1055          */
1056         public function tryStashFile( User $user, $isPartial = false ) {
1057                 if ( !$isPartial ) {
1058                         $error = $this->runUploadStashFileHook( $user );
1059                         if ( $error ) {
1060                                 return call_user_func_array( 'Status::newFatal', $error );
1061                         }
1062                 }
1063                 try {
1064                         $file = $this->doStashFile( $user );
1065                         return Status::newGood( $file );
1066                 } catch ( UploadStashException $e ) {
1067                         return Status::newFatal( 'uploadstash-exception', get_class( $e ), $e->getMessage() );
1068                 }
1069         }
1070
1071         /**
1072          * @param User $user
1073          * @return array|null Error message and parameters, null if there's no error
1074          */
1075         protected function runUploadStashFileHook( User $user ) {
1076                 $props = $this->mFileProps;
1077                 $error = null;
1078                 Hooks::run( 'UploadStashFile', [ $this, $user, $props, &$error ] );
1079                 if ( $error ) {
1080                         if ( !is_array( $error ) ) {
1081                                 $error = [ $error ];
1082                         }
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 $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 $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
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 array $ext
1196          * @param array $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                 $magic = MimeMagic::singleton();
1212
1213                 if ( !$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 ( false !== strpos( $chunk, $tag ) ) {
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, -1, $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                         MediaWiki\suppressWarnings();
1423                         $str = iconv( $encoding, 'UTF-8', $contents );
1424                         MediaWiki\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          */
1501         public static function checkSvgExternalDTD( $type, $publicId, $systemId ) {
1502                 // This doesn't include the XHTML+MathML+SVG doctype since we don't
1503                 // allow XHTML anyways.
1504                 $allowedDTDs = [
1505                         'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd',
1506                         'http://www.w3.org/TR/2001/REC-SVG-20010904/DTD/svg10.dtd',
1507                         'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-basic.dtd',
1508                         'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-tiny.dtd',
1509                         // https://phabricator.wikimedia.org/T168856
1510                         'http://www.w3.org/TR/2001/PR-SVG-20010719/DTD/svg10.dtd',
1511                 ];
1512                 if ( $type !== 'PUBLIC'
1513                         || !in_array( $systemId, $allowedDTDs )
1514                         || strpos( $publicId, "-//W3C//" ) !== 0
1515                 ) {
1516                         return [ 'upload-scripted-dtd' ];
1517                 }
1518                 return false;
1519         }
1520
1521         /**
1522          * @todo Replace this with a whitelist filter!
1523          * @param string $element
1524          * @param array $attribs
1525          * @return bool
1526          */
1527         public function checkSvgScriptCallback( $element, $attribs, $data = null ) {
1528
1529                 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1530
1531                 // We specifically don't include:
1532                 // http://www.w3.org/1999/xhtml (T62771)
1533                 static $validNamespaces = [
1534                         '',
1535                         'adobe:ns:meta/',
1536                         'http://creativecommons.org/ns#',
1537                         'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1538                         'http://ns.adobe.com/adobeillustrator/10.0/',
1539                         'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1540                         'http://ns.adobe.com/extensibility/1.0/',
1541                         'http://ns.adobe.com/flows/1.0/',
1542                         'http://ns.adobe.com/illustrator/1.0/',
1543                         'http://ns.adobe.com/imagereplacement/1.0/',
1544                         'http://ns.adobe.com/pdf/1.3/',
1545                         'http://ns.adobe.com/photoshop/1.0/',
1546                         'http://ns.adobe.com/saveforweb/1.0/',
1547                         'http://ns.adobe.com/variables/1.0/',
1548                         'http://ns.adobe.com/xap/1.0/',
1549                         'http://ns.adobe.com/xap/1.0/g/',
1550                         'http://ns.adobe.com/xap/1.0/g/img/',
1551                         'http://ns.adobe.com/xap/1.0/mm/',
1552                         'http://ns.adobe.com/xap/1.0/rights/',
1553                         'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1554                         'http://ns.adobe.com/xap/1.0/stype/font#',
1555                         'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1556                         'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1557                         'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1558                         'http://ns.adobe.com/xap/1.0/t/pg/',
1559                         'http://purl.org/dc/elements/1.1/',
1560                         'http://purl.org/dc/elements/1.1',
1561                         'http://schemas.microsoft.com/visio/2003/svgextensions/',
1562                         'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1563                         'http://taptrix.com/inkpad/svg_extensions',
1564                         'http://web.resource.org/cc/',
1565                         'http://www.freesoftware.fsf.org/bkchem/cdml',
1566                         'http://www.inkscape.org/namespaces/inkscape',
1567                         'http://www.opengis.net/gml',
1568                         'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1569                         'http://www.w3.org/2000/svg',
1570                         'http://www.w3.org/tr/rec-rdf-syntax/',
1571                         'http://www.w3.org/2000/01/rdf-schema#',
1572                 ];
1573
1574                 // Inkscape mangles namespace definitions created by Adobe Illustrator.
1575                 // This is nasty but harmless. (T144827)
1576                 $isBuggyInkscape = preg_match( '/^&(#38;)*ns_[a-z_]+;$/', $namespace );
1577
1578                 if ( !( $isBuggyInkscape || in_array( $namespace, $validNamespaces ) ) ) {
1579                         wfDebug( __METHOD__ . ": Non-svg namespace '$namespace' in uploaded file.\n" );
1580                         /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1581                         $this->mSVGNSError = $namespace;
1582
1583                         return true;
1584                 }
1585
1586                 /*
1587                  * check for elements that can contain javascript
1588                  */
1589                 if ( $strippedElement == 'script' ) {
1590                         wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
1591
1592                         return [ 'uploaded-script-svg', $strippedElement ];
1593                 }
1594
1595                 # e.g., <svg xmlns="http://www.w3.org/2000/svg">
1596                 #  <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1597                 if ( $strippedElement == 'handler' ) {
1598                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1599
1600                         return [ 'uploaded-script-svg', $strippedElement ];
1601                 }
1602
1603                 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1604                 if ( $strippedElement == 'stylesheet' ) {
1605                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1606
1607                         return [ 'uploaded-script-svg', $strippedElement ];
1608                 }
1609
1610                 # Block iframes, in case they pass the namespace check
1611                 if ( $strippedElement == 'iframe' ) {
1612                         wfDebug( __METHOD__ . ": iframe in uploaded file.\n" );
1613
1614                         return [ 'uploaded-script-svg', $strippedElement ];
1615                 }
1616
1617                 # Check <style> css
1618                 if ( $strippedElement == 'style'
1619                         && self::checkCssFragment( Sanitizer::normalizeCss( $data ) )
1620                 ) {
1621                         wfDebug( __METHOD__ . ": hostile css in style element.\n" );
1622                         return [ 'uploaded-hostile-svg' ];
1623                 }
1624
1625                 foreach ( $attribs as $attrib => $value ) {
1626                         $stripped = $this->stripXmlNamespace( $attrib );
1627                         $value = strtolower( $value );
1628
1629                         if ( substr( $stripped, 0, 2 ) == 'on' ) {
1630                                 wfDebug( __METHOD__
1631                                         . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1632
1633                                 return [ 'uploaded-event-handler-on-svg', $attrib, $value ];
1634                         }
1635
1636                         # Do not allow relative links, or unsafe url schemas.
1637                         # For <a> tags, only data:, http: and https: and same-document
1638                         # fragment links are allowed. For all other tags, only data:
1639                         # and fragment are allowed.
1640                         if ( $stripped == 'href'
1641                                 && $value !== ''
1642                                 && strpos( $value, 'data:' ) !== 0
1643                                 && strpos( $value, '#' ) !== 0
1644                         ) {
1645                                 if ( !( $strippedElement === 'a'
1646                                         && preg_match( '!^https?://!i', $value ) )
1647                                 ) {
1648                                         wfDebug( __METHOD__ . ": Found href attribute <$strippedElement "
1649                                                 . "'$attrib'='$value' in uploaded file.\n" );
1650
1651                                         return [ 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value ];
1652                                 }
1653                         }
1654
1655                         # only allow data: targets that should be safe. This prevents vectors like,
1656                         # image/svg, text/xml, application/xml, and text/html, which can contain scripts
1657                         if ( $stripped == 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) {
1658                                 // rfc2397 parameters. This is only slightly slower than (;[\w;]+)*.
1659                                 // @codingStandardsIgnoreStart Generic.Files.LineLength
1660                                 $parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?';
1661                                 // @codingStandardsIgnoreEnd
1662
1663                                 if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|png)$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                         if ( $strippedElement == 'image'
1742                                 && $stripped == 'filter'
1743                                 && preg_match( '!url\s*\(!sim', $value )
1744                         ) {
1745                                 wfDebug( __METHOD__ . ": Found image filter with url: "
1746                                         . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1747
1748                                 return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ];
1749                         }
1750                 }
1751
1752                 return false; // No scripts detected
1753         }
1754
1755         /**
1756          * Check a block of CSS or CSS fragment for anything that looks like
1757          * it is bringing in remote code.
1758          * @param string $value a string of CSS
1759          * @param bool $propOnly only check css properties (start regex with :)
1760          * @return bool true if the CSS contains an illegal string, false if otherwise
1761          */
1762         private static function checkCssFragment( $value ) {
1763
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 = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
1863                         $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
1864
1865                 if ( strpos( $command, "%f" ) === false ) {
1866                         # simple pattern: append file to scan
1867                         $command .= " " . wfEscapeShellArg( $file );
1868                 } else {
1869                         # complex pattern: replace "%f" with file to scan
1870                         $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
1871                 }
1872
1873                 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
1874
1875                 # execute virus scanner
1876                 $exitCode = false;
1877
1878                 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1879                 #      that does not seem to be worth the pain.
1880                 #      Ask me (Duesentrieb) about it if it's ever needed.
1881                 $output = wfShellExecWithStderr( $command, $exitCode );
1882
1883                 # map exit code to AV_xxx constants.
1884                 $mappedCode = $exitCode;
1885                 if ( $exitCodeMap ) {
1886                         if ( isset( $exitCodeMap[$exitCode] ) ) {
1887                                 $mappedCode = $exitCodeMap[$exitCode];
1888                         } elseif ( isset( $exitCodeMap["*"] ) ) {
1889                                 $mappedCode = $exitCodeMap["*"];
1890                         }
1891                 }
1892
1893                 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1894                  * so we need the strict equalities === and thus can't use a switch here
1895                  */
1896                 if ( $mappedCode === AV_SCAN_FAILED ) {
1897                         # scan failed (code was mapped to false by $exitCodeMap)
1898                         wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
1899
1900                         $output = $wgAntivirusRequired
1901                                 ? wfMessage( 'virus-scanfailed', [ $exitCode ] )->text()
1902                                 : null;
1903                 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
1904                         # scan failed because filetype is unknown (probably imune)
1905                         wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
1906                         $output = null;
1907                 } elseif ( $mappedCode === AV_NO_VIRUS ) {
1908                         # no virus found
1909                         wfDebug( __METHOD__ . ": file passed virus scan.\n" );
1910                         $output = false;
1911                 } else {
1912                         $output = trim( $output );
1913
1914                         if ( !$output ) {
1915                                 $output = true; # if there's no output, return true
1916                         } elseif ( $msgPattern ) {
1917                                 $groups = [];
1918                                 if ( preg_match( $msgPattern, $output, $groups ) ) {
1919                                         if ( $groups[1] ) {
1920                                                 $output = $groups[1];
1921                                         }
1922                                 }
1923                         }
1924
1925                         wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
1926                 }
1927
1928                 return $output;
1929         }
1930
1931         /**
1932          * Check if there's an overwrite conflict and, if so, if restrictions
1933          * forbid this user from performing the upload.
1934          *
1935          * @param User $user
1936          *
1937          * @return mixed True on success, array on failure
1938          */
1939         private function checkOverwrite( $user ) {
1940                 // First check whether the local file can be overwritten
1941                 $file = $this->getLocalFile();
1942                 $file->load( File::READ_LATEST );
1943                 if ( $file->exists() ) {
1944                         if ( !self::userCanReUpload( $user, $file ) ) {
1945                                 return [ 'fileexists-forbidden', $file->getName() ];
1946                         } else {
1947                                 return true;
1948                         }
1949                 }
1950
1951                 /* Check shared conflicts: if the local file does not exist, but
1952                  * wfFindFile finds a file, it exists in a shared repository.
1953                  */
1954                 $file = wfFindFile( $this->getTitle(), [ 'latest' => true ] );
1955                 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1956                         return [ 'fileexists-shared-forbidden', $file->getName() ];
1957                 }
1958
1959                 return true;
1960         }
1961
1962         /**
1963          * Check if a user is the last uploader
1964          *
1965          * @param User $user
1966          * @param File $img
1967          * @return bool
1968          */
1969         public static function userCanReUpload( User $user, File $img ) {
1970                 if ( $user->isAllowed( 'reupload' ) ) {
1971                         return true; // non-conditional
1972                 } elseif ( !$user->isAllowed( 'reupload-own' ) ) {
1973                         return false;
1974                 }
1975
1976                 if ( !( $img instanceof LocalFile ) ) {
1977                         return false;
1978                 }
1979
1980                 $img->load();
1981
1982                 return $user->getId() == $img->getUser( 'id' );
1983         }
1984
1985         /**
1986          * Helper function that does various existence checks for a file.
1987          * The following checks are performed:
1988          * - The file exists
1989          * - Article with the same name as the file exists
1990          * - File exists with normalized extension
1991          * - The file looks like a thumbnail and the original exists
1992          *
1993          * @param File $file The File object to check
1994          * @return mixed False if the file does not exists, else an array
1995          */
1996         public static function getExistsWarning( $file ) {
1997                 if ( $file->exists() ) {
1998                         return [ 'warning' => 'exists', 'file' => $file ];
1999                 }
2000
2001                 if ( $file->getTitle()->getArticleID() ) {
2002                         return [ 'warning' => 'page-exists', 'file' => $file ];
2003                 }
2004
2005                 if ( strpos( $file->getName(), '.' ) == false ) {
2006                         $partname = $file->getName();
2007                         $extension = '';
2008                 } else {
2009                         $n = strrpos( $file->getName(), '.' );
2010                         $extension = substr( $file->getName(), $n + 1 );
2011                         $partname = substr( $file->getName(), 0, $n );
2012                 }
2013                 $normalizedExtension = File::normalizeExtension( $extension );
2014
2015                 if ( $normalizedExtension != $extension ) {
2016                         // We're not using the normalized form of the extension.
2017                         // Normal form is lowercase, using most common of alternate
2018                         // extensions (eg 'jpg' rather than 'JPEG').
2019
2020                         // Check for another file using the normalized form...
2021                         $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
2022                         $file_lc = wfLocalFile( $nt_lc );
2023
2024                         if ( $file_lc->exists() ) {
2025                                 return [
2026                                         'warning' => 'exists-normalized',
2027                                         'file' => $file,
2028                                         'normalizedFile' => $file_lc
2029                                 ];
2030                         }
2031                 }
2032
2033                 // Check for files with the same name but a different extension
2034                 $similarFiles = RepoGroup::singleton()->getLocalRepo()->findFilesByPrefix(
2035                         "{$partname}.", 1 );
2036                 if ( count( $similarFiles ) ) {
2037                         return [
2038                                 'warning' => 'exists-normalized',
2039                                 'file' => $file,
2040                                 'normalizedFile' => $similarFiles[0],
2041                         ];
2042                 }
2043
2044                 if ( self::isThumbName( $file->getName() ) ) {
2045                         # Check for filenames like 50px- or 180px-, these are mostly thumbnails
2046                         $nt_thb = Title::newFromText(
2047                                 substr( $partname, strpos( $partname, '-' ) + 1 ) . '.' . $extension,
2048                                 NS_FILE
2049                         );
2050                         $file_thb = wfLocalFile( $nt_thb );
2051                         if ( $file_thb->exists() ) {
2052                                 return [
2053                                         'warning' => 'thumb',
2054                                         'file' => $file,
2055                                         'thumbFile' => $file_thb
2056                                 ];
2057                         } else {
2058                                 // File does not exist, but we just don't like the name
2059                                 return [
2060                                         'warning' => 'thumb-name',
2061                                         'file' => $file,
2062                                         'thumbFile' => $file_thb
2063                                 ];
2064                         }
2065                 }
2066
2067                 foreach ( self::getFilenamePrefixBlacklist() as $prefix ) {
2068                         if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
2069                                 return [
2070                                         'warning' => 'bad-prefix',
2071                                         'file' => $file,
2072                                         'prefix' => $prefix
2073                                 ];
2074                         }
2075                 }
2076
2077                 return false;
2078         }
2079
2080         /**
2081          * Helper function that checks whether the filename looks like a thumbnail
2082          * @param string $filename
2083          * @return bool
2084          */
2085         public static function isThumbName( $filename ) {
2086                 $n = strrpos( $filename, '.' );
2087                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
2088
2089                 return (
2090                         substr( $partname, 3, 3 ) == 'px-' ||
2091                         substr( $partname, 2, 3 ) == 'px-'
2092                 ) &&
2093                 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
2094         }
2095
2096         /**
2097          * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
2098          *
2099          * @return array List of prefixes
2100          */
2101         public static function getFilenamePrefixBlacklist() {
2102                 $blacklist = [];
2103                 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
2104                 if ( !$message->isDisabled() ) {
2105                         $lines = explode( "\n", $message->plain() );
2106                         foreach ( $lines as $line ) {
2107                                 // Remove comment lines
2108                                 $comment = substr( trim( $line ), 0, 1 );
2109                                 if ( $comment == '#' || $comment == '' ) {
2110                                         continue;
2111                                 }
2112                                 // Remove additional comments after a prefix
2113                                 $comment = strpos( $line, '#' );
2114                                 if ( $comment > 0 ) {
2115                                         $line = substr( $line, 0, $comment - 1 );
2116                                 }
2117                                 $blacklist[] = trim( $line );
2118                         }
2119                 }
2120
2121                 return $blacklist;
2122         }
2123
2124         /**
2125          * Gets image info about the file just uploaded.
2126          *
2127          * Also has the effect of setting metadata to be an 'indexed tag name' in
2128          * returned API result if 'metadata' was requested. Oddly, we have to pass
2129          * the "result" object down just so it can do that with the appropriate
2130          * format, presumably.
2131          *
2132          * @param ApiResult $result
2133          * @return array Image info
2134          */
2135         public function getImageInfo( $result ) {
2136                 $localFile = $this->getLocalFile();
2137                 $stashFile = $this->getStashFile();
2138                 // Calling a different API module depending on whether the file was stashed is less than optimal.
2139                 // In fact, calling API modules here at all is less than optimal. Maybe it should be refactored.
2140                 if ( $stashFile ) {
2141                         $imParam = ApiQueryStashImageInfo::getPropertyNames();
2142                         $info = ApiQueryStashImageInfo::getInfo( $stashFile, array_flip( $imParam ), $result );
2143                 } else {
2144                         $imParam = ApiQueryImageInfo::getPropertyNames();
2145                         $info = ApiQueryImageInfo::getInfo( $localFile, array_flip( $imParam ), $result );
2146                 }
2147
2148                 return $info;
2149         }
2150
2151         /**
2152          * @param array $error
2153          * @return Status
2154          */
2155         public function convertVerifyErrorToStatus( $error ) {
2156                 $code = $error['status'];
2157                 unset( $code['status'] );
2158
2159                 return Status::newFatal( $this->getVerificationErrorCode( $code ), $error );
2160         }
2161
2162         /**
2163          * Get the MediaWiki maximum uploaded file size for given type of upload, based on
2164          * $wgMaxUploadSize.
2165          *
2166          * @param null|string $forType
2167          * @return int
2168          */
2169         public static function getMaxUploadSize( $forType = null ) {
2170                 global $wgMaxUploadSize;
2171
2172                 if ( is_array( $wgMaxUploadSize ) ) {
2173                         if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
2174                                 return $wgMaxUploadSize[$forType];
2175                         } else {
2176                                 return $wgMaxUploadSize['*'];
2177                         }
2178                 } else {
2179                         return intval( $wgMaxUploadSize );
2180                 }
2181         }
2182
2183         /**
2184          * Get the PHP maximum uploaded file size, based on ini settings. If there is no limit or the
2185          * limit can't be guessed, returns a very large number (PHP_INT_MAX).
2186          *
2187          * @since 1.27
2188          * @return int
2189          */
2190         public static function getMaxPhpUploadSize() {
2191                 $phpMaxFileSize = wfShorthandToInteger(
2192                         ini_get( 'upload_max_filesize' ) ?: ini_get( 'hhvm.server.upload.upload_max_file_size' ),
2193                         PHP_INT_MAX
2194                 );
2195                 $phpMaxPostSize = wfShorthandToInteger(
2196                         ini_get( 'post_max_size' ) ?: ini_get( 'hhvm.server.max_post_size' ),
2197                         PHP_INT_MAX
2198                 ) ?: PHP_INT_MAX;
2199                 return min( $phpMaxFileSize, $phpMaxPostSize );
2200         }
2201
2202         /**
2203          * Get the current status of a chunked upload (used for polling)
2204          *
2205          * The value will be read from cache.
2206          *
2207          * @param User $user
2208          * @param string $statusKey
2209          * @return Status[]|bool
2210          */
2211         public static function getSessionStatus( User $user, $statusKey ) {
2212                 $cache = MediaWikiServices::getInstance()->getMainObjectStash();
2213                 $key = $cache->makeKey( 'uploadstatus', $user->getId() ?: md5( $user->getName() ), $statusKey );
2214
2215                 return $cache->get( $key );
2216         }
2217
2218         /**
2219          * Set the current status of a chunked upload (used for polling)
2220          *
2221          * The value will be set in cache for 1 day
2222          *
2223          * @param User $user
2224          * @param string $statusKey
2225          * @param array|bool $value
2226          * @return void
2227          */
2228         public static function setSessionStatus( User $user, $statusKey, $value ) {
2229                 $cache = MediaWikiServices::getInstance()->getMainObjectStash();
2230                 $key = $cache->makeKey( 'uploadstatus', $user->getId() ?: md5( $user->getName() ), $statusKey );
2231
2232                 if ( $value === false ) {
2233                         $cache->delete( $key );
2234                 } else {
2235                         $cache->set( $key, $value, $cache::TTL_DAY );
2236                 }
2237         }
2238 }