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 array Array of warnings
 645          */
 646         public function checkWarnings() {
 647                 global $wgLang;
 648
 649                 $warnings = [];
 650
 651                 $localFile = $this->getLocalFile();
 652                 $localFile->load( File::READ_LATEST );
 653                 $filename = $localFile->getName();
 654
 655                 /**
 656                  * Check whether the resulting filename is different from the desired one,
 657                  * but ignore things like ucfirst() and spaces/underscore things
 658                  */
 659                 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName );
 660                 $comparableName = Title::capitalize( $comparableName, NS_FILE );
 661
 662                 if ( $this->mDesiredDestName != $filename && $comparableName != $filename ) {
 663                         $warnings['badfilename'] = $filename;
 664                 }
 665
 666                 // Check whether the file extension is on the unwanted list
 667                 global $wgCheckFileExtensions, $wgFileExtensions;
 668                 if ( $wgCheckFileExtensions ) {
 669                         $extensions = array_unique( $wgFileExtensions );
 670                         if ( !$this->checkFileExtension( $this->mFinalExtension, $extensions ) ) {
 671                                 $warnings['filetype-unwanted-type'] = [ $this->mFinalExtension,
 672                                         $wgLang->commaList( $extensions ), count( $extensions ) ];
 673                         }
 674                 }
 675
 676                 global $wgUploadSizeWarning;
 677                 if ( $wgUploadSizeWarning && ( $this->mFileSize > $wgUploadSizeWarning ) ) {
 678                         $warnings['large-file'] = [ $wgUploadSizeWarning, $this->mFileSize ];
 679                 }
 680
 681                 if ( $this->mFileSize == 0 ) {
 682                         $warnings['empty-file'] = true;
 683                 }
 684
 685                 $hash = $this->getTempFileSha1Base36();
 686                 $exists = self::getExistsWarning( $localFile );
 687                 if ( $exists !== false ) {
 688                         $warnings['exists'] = $exists;
 689
 690                         // check if file is an exact duplicate of current file version
 691                         if ( $hash === $localFile->getSha1() ) {
 692                                 $warnings['no-change'] = $localFile;
 693                         }
 694
 695                         // check if file is an exact duplicate of older versions of this file
 696                         $history = $localFile->getHistory();
 697                         foreach ( $history as $oldFile ) {
 698                                 if ( $hash === $oldFile->getSha1() ) {
 699                                         $warnings['duplicate-version'][] = $oldFile;
 700                                 }
 701                         }
 702                 }
 703
 704                 if ( $localFile->wasDeleted() && !$localFile->exists() ) {
 705                         $warnings['was-deleted'] = $filename;
 706                 }
 707
 708                 // Check dupes against existing files
 709                 $dupes = RepoGroup::singleton()->findBySha1( $hash );
 710                 $title = $this->getTitle();
 711                 // Remove all matches against self
 712                 foreach ( $dupes as $key => $dupe ) {
 713                         if ( $title->equals( $dupe->getTitle() ) ) {
 714                                 unset( $dupes[$key] );
 715                         }
 716                 }
 717                 if ( $dupes ) {
 718                         $warnings['duplicate'] = $dupes;
 719                 }
 720
 721                 // Check dupes against archives
 722                 $archivedFile = new ArchivedFile( null, 0, '', $hash );
 723                 if ( $archivedFile->getID() > 0 ) {
 724                         if ( $archivedFile->userCan( File::DELETED_FILE ) ) {
 725                                 $warnings['duplicate-archive'] = $archivedFile->getName();
 726                         } else {
 727                                 $warnings['duplicate-archive'] = '';
 728                         }
 729                 }
 730
 731                 return $warnings;
 732         }
 733
 734         /**
 735          * Really perform the upload. Stores the file in the local repo, watches
 736          * if necessary and runs the UploadComplete hook.
 737          *
 738          * @param string $comment
 739          * @param string $pageText
 740          * @param bool $watch Whether the file page should be added to user's watchlist.
 741          *   (This doesn't check $user's permissions.)
 742          * @param User $user
 743          * @param string[] $tags Change tags to add to the log entry and page revision.
 744          *   (This doesn't check $user's permissions.)
 745          * @return Status Indicating the whether the upload succeeded.
 746          */
 747         public function performUpload( $comment, $pageText, $watch, $user, $tags = [] ) {
 748                 $this->getLocalFile()->load( File::READ_LATEST );
 749                 $props = $this->mFileProps;
 750
 751                 $error = null;
 752                 Hooks::run( 'UploadVerifyUpload', [ $this, $user, $props, $comment, $pageText, &$error ] );
 753                 if ( $error ) {
 754                         if ( !is_array( $error ) ) {
 755                                 $error = [ $error ];
 756                         }
 757                         return call_user_func_array( 'Status::newFatal', $error );
 758                 }
 759
 760                 $status = $this->getLocalFile()->upload(
 761                         $this->mTempPath,
 762                         $comment,
 763                         $pageText,
 764                         File::DELETE_SOURCE,
 765                         $props,
 766                         false,
 767                         $user,
 768                         $tags
 769                 );
 770
 771                 if ( $status->isGood() ) {
 772                         if ( $watch ) {
 773                                 WatchAction::doWatch(
 774                                         $this->getLocalFile()->getTitle(),
 775                                         $user,
 776                                         User::IGNORE_USER_RIGHTS
 777                                 );
 778                         }
 779                         // Avoid PHP 7.1 warning of passing $this by reference
 780                         $uploadBase = $this;
 781                         Hooks::run( 'UploadComplete', [ &$uploadBase ] );
 782
 783                         $this->postProcessUpload();
 784                 }
 785
 786                 return $status;
 787         }
 788
 789         /**
 790          * Perform extra steps after a successful upload.
 791          *
 792          * @since  1.25
 793          */
 794         public function postProcessUpload() {
 795         }
 796
 797         /**
 798          * Returns the title of the file to be uploaded. Sets mTitleError in case
 799          * the name was illegal.
 800          *
 801          * @return Title|null The title of the file or null in case the name was illegal
 802          */
 803         public function getTitle() {
 804                 if ( $this->mTitle !== false ) {
 805                         return $this->mTitle;
 806                 }
 807                 if ( !is_string( $this->mDesiredDestName ) ) {
 808                         $this->mTitleError = self::ILLEGAL_FILENAME;
 809                         $this->mTitle = null;
 810
 811                         return $this->mTitle;
 812                 }
 813                 /* Assume that if a user specified File:Something.jpg, this is an error
 814                  * and that the namespace prefix needs to be stripped of.
 815                  */
 816                 $title = Title::newFromText( $this->mDesiredDestName );
 817                 if ( $title && $title->getNamespace() == NS_FILE ) {
 818                         $this->mFilteredName = $title->getDBkey();
 819                 } else {
 820                         $this->mFilteredName = $this->mDesiredDestName;
 821                 }
 822
 823                 # oi_archive_name is max 255 bytes, which include a timestamp and an
 824                 # exclamation mark, so restrict file name to 240 bytes.
 825                 if ( strlen( $this->mFilteredName ) > 240 ) {
 826                         $this->mTitleError = self::FILENAME_TOO_LONG;
 827                         $this->mTitle = null;
 828
 829                         return $this->mTitle;
 830                 }
 831
 832                 /**
 833                  * Chop off any directories in the given filename. Then
 834                  * filter out illegal characters, and try to make a legible name
 835                  * out of it. We'll strip some silently that Title would die on.
 836                  */
 837                 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mFilteredName );
 838                 /* Normalize to title form before we do any further processing */
 839                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 840                 if ( is_null( $nt ) ) {
 841                         $this->mTitleError = self::ILLEGAL_FILENAME;
 842                         $this->mTitle = null;
 843
 844                         return $this->mTitle;
 845                 }
 846                 $this->mFilteredName = $nt->getDBkey();
 847
 848                 /**
 849                  * We'll want to blacklist against *any* 'extension', and use
 850                  * only the final one for the whitelist.
 851                  */
 852                 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
 853
 854                 if ( count( $ext ) ) {
 855                         $this->mFinalExtension = trim( $ext[count( $ext ) - 1] );
 856                 } else {
 857                         $this->mFinalExtension = '';
 858
 859                         # No extension, try guessing one
 860                         $magic = MimeMagic::singleton();
 861                         $mime = $magic->guessMimeType( $this->mTempPath );
 862                         if ( $mime !== 'unknown/unknown' ) {
 863                                 # Get a space separated list of extensions
 864                                 $extList = $magic->getExtensionsForType( $mime );
 865                                 if ( $extList ) {
 866                                         # Set the extension to the canonical extension
 867                                         $this->mFinalExtension = strtok( $extList, ' ' );
 868
 869                                         # Fix up the other variables
 870                                         $this->mFilteredName .= ".{$this->mFinalExtension}";
 871                                         $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 872                                         $ext = [ $this->mFinalExtension ];
 873                                 }
 874                         }
 875                 }
 876
 877                 /* Don't allow users to override the blacklist (check file extension) */
 878                 global $wgCheckFileExtensions, $wgStrictFileExtensions;
 879                 global $wgFileExtensions, $wgFileBlacklist;
 880
 881                 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
 882
 883                 if ( $this->mFinalExtension == '' ) {
 884                         $this->mTitleError = self::FILETYPE_MISSING;
 885                         $this->mTitle = null;
 886
 887                         return $this->mTitle;
 888                 } elseif ( $blackListedExtensions ||
 889                         ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
 890                                 !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) )
 891                 ) {
 892                         $this->mBlackListedExtensions = $blackListedExtensions;
 893                         $this->mTitleError = self::FILETYPE_BADTYPE;
 894                         $this->mTitle = null;
 895
 896                         return $this->mTitle;
 897                 }
 898
 899                 // Windows may be broken with special characters, see T3780
 900                 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
 901                         && !RepoGroup::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
 902                 ) {
 903                         $this->mTitleError = self::WINDOWS_NONASCII_FILENAME;
 904                         $this->mTitle = null;
 905
 906                         return $this->mTitle;
 907                 }
 908
 909                 # If there was more than one "extension", reassemble the base
 910                 # filename to prevent bogus complaints about length
 911                 if ( count( $ext ) > 1 ) {
 912                         $iterations = count( $ext ) - 1;
 913                         for ( $i = 0; $i < $iterations; $i++ ) {
 914                                 $partname .= '.' . $ext[$i];
 915                         }
 916                 }
 917
 918                 if ( strlen( $partname ) < 1 ) {
 919                         $this->mTitleError = self::MIN_LENGTH_PARTNAME;
 920                         $this->mTitle = null;
 921
 922                         return $this->mTitle;
 923                 }
 924
 925                 $this->mTitle = $nt;
 926
 927                 return $this->mTitle;
 928         }
 929
 930         /**
 931          * Return the local file and initializes if necessary.
 932          *
 933          * @return LocalFile|null
 934          */
 935         public function getLocalFile() {
 936                 if ( is_null( $this->mLocalFile ) ) {
 937                         $nt = $this->getTitle();
 938                         $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
 939                 }
 940
 941                 return $this->mLocalFile;
 942         }
 943
 944         /**
 945          * @return UploadStashFile|null
 946          */
 947         public function getStashFile() {
 948                 return $this->mStashFile;
 949         }
 950
 951         /**
 952          * Like stashFile(), but respects extensions' wishes to prevent the stashing. verifyUpload() must
 953          * be called before calling this method (unless $isPartial is true).
 954          *
 955          * Upload stash exceptions are also caught and converted to an error status.
 956          *
 957          * @since 1.28
 958          * @param User $user
 959          * @param bool $isPartial Pass `true` if this is a part of a chunked upload (not a complete file).
 960          * @return Status If successful, value is an UploadStashFile instance
 961          */
 962         public function tryStashFile( User $user, $isPartial = false ) {
 963                 if ( !$isPartial ) {
 964                         $error = $this->runUploadStashFileHook( $user );
 965                         if ( $error ) {
 966                                 return call_user_func_array( 'Status::newFatal', $error );
 967                         }
 968                 }
 969                 try {
 970                         $file = $this->doStashFile( $user );
 971                         return Status::newGood( $file );
 972                 } catch ( UploadStashException $e ) {
 973                         return Status::newFatal( 'uploadstash-exception', get_class( $e ), $e->getMessage() );
 974                 }
 975         }
 976
 977         /**
 978          * @param User $user
 979          * @return array|null Error message and parameters, null if there's no error
 980          */
 981         protected function runUploadStashFileHook( User $user ) {
 982                 $props = $this->mFileProps;
 983                 $error = null;
 984                 Hooks::run( 'UploadStashFile', [ $this, $user, $props, &$error ] );
 985                 if ( $error ) {
 986                         if ( !is_array( $error ) ) {
 987                                 $error = [ $error ];
 988                         }
 989                 }
 990                 return $error;
 991         }
 992
 993         /**
 994          * If the user does not supply all necessary information in the first upload
 995          * form submission (either by accident or by design) then we may want to
 996          * stash the file temporarily, get more information, and publish the file
 997          * later.
 998          *
 999          * This method will stash a file in a temporary directory for later
1000          * processing, and save the necessary descriptive info into the database.
1001          * This method returns the file object, which also has a 'fileKey' property
1002          * which can be passed through a form or API request to find this stashed
1003          * file again.
1004          *
1005          * @deprecated since 1.28 Use tryStashFile() instead
1006          * @param User $user
1007          * @return UploadStashFile Stashed file
1008          * @throws UploadStashBadPathException
1009          * @throws UploadStashFileException
1010          * @throws UploadStashNotLoggedInException
1011          */
1012         public function stashFile( User $user = null ) {
1013                 return $this->doStashFile( $user );
1014         }
1015
1016         /**
1017          * Implementation for stashFile() and tryStashFile().
1018          *
1019          * @param User $user
1020          * @return UploadStashFile Stashed file
1021          */
1022         protected function doStashFile( User $user = null ) {
1023                 $stash = RepoGroup::singleton()->getLocalRepo()->getUploadStash( $user );
1024                 $file = $stash->stashFile( $this->mTempPath, $this->getSourceType() );
1025                 $this->mStashFile = $file;
1026
1027                 return $file;
1028         }
1029
1030         /**
1031          * Stash a file in a temporary directory, returning a key which can be used
1032          * to find the file again. See stashFile().
1033          *
1034          * @deprecated since 1.28
1035          * @return string File key
1036          */
1037         public function stashFileGetKey() {
1038                 wfDeprecated( __METHOD__, '1.28' );
1039                 return $this->doStashFile()->getFileKey();
1040         }
1041
1042         /**
1043          * alias for stashFileGetKey, for backwards compatibility
1044          *
1045          * @deprecated since 1.28
1046          * @return string File key
1047          */
1048         public function stashSession() {
1049                 wfDeprecated( __METHOD__, '1.28' );
1050                 return $this->doStashFile()->getFileKey();
1051         }
1052
1053         /**
1054          * If we've modified the upload file we need to manually remove it
1055          * on exit to clean up.
1056          */
1057         public function cleanupTempFile() {
1058                 if ( $this->mRemoveTempFile && $this->tempFileObj ) {
1059                         // Delete when all relevant TempFSFile handles go out of scope
1060                         wfDebug( __METHOD__ . ": Marked temporary file '{$this->mTempPath}' for removal\n" );
1061                         $this->tempFileObj->autocollect();
1062                 }
1063         }
1064
1065         public function getTempPath() {
1066                 return $this->mTempPath;
1067         }
1068
1069         /**
1070          * Split a file into a base name and all dot-delimited 'extensions'
1071          * on the end. Some web server configurations will fall back to
1072          * earlier pseudo-'extensions' to determine type and execute
1073          * scripts, so the blacklist needs to check them all.
1074          *
1075          * @param string $filename
1076          * @return array
1077          */
1078         public static function splitExtensions( $filename ) {
1079                 $bits = explode( '.', $filename );
1080                 $basename = array_shift( $bits );
1081
1082                 return [ $basename, $bits ];
1083         }
1084
1085         /**
1086          * Perform case-insensitive match against a list of file extensions.
1087          * Returns true if the extension is in the list.
1088          *
1089          * @param string $ext
1090          * @param array $list
1091          * @return bool
1092          */
1093         public static function checkFileExtension( $ext, $list ) {
1094                 return in_array( strtolower( $ext ), $list );
1095         }
1096
1097         /**
1098          * Perform case-insensitive match against a list of file extensions.
1099          * Returns an array of matching extensions.
1100          *
1101          * @param array $ext
1102          * @param array $list
1103          * @return bool
1104          */
1105         public static function checkFileExtensionList( $ext, $list ) {
1106                 return array_intersect( array_map( 'strtolower', $ext ), $list );
1107         }
1108
1109         /**
1110          * Checks if the MIME type of the uploaded file matches the file extension.
1111          *
1112          * @param string $mime The MIME type of the uploaded file
1113          * @param string $extension The filename extension that the file is to be served with
1114          * @return bool
1115          */
1116         public static function verifyExtension( $mime, $extension ) {
1117                 $magic = MimeMagic::singleton();
1118
1119                 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' ) {
1120                         if ( !$magic->isRecognizableExtension( $extension ) ) {
1121                                 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
1122                                         "unrecognized extension '$extension', can't verify\n" );
1123
1124                                 return true;
1125                         } else {
1126                                 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; " .
1127                                         "recognized extension '$extension', so probably invalid file\n" );
1128
1129                                 return false;
1130                         }
1131                 }
1132
1133                 $match = $magic->isMatchingExtension( $extension, $mime );
1134
1135                 if ( $match === null ) {
1136                         if ( $magic->getTypesForExtension( $extension ) !== null ) {
1137                                 wfDebug( __METHOD__ . ": No extension known for $mime, but we know a mime for $extension\n" );
1138
1139                                 return false;
1140                         } else {
1141                                 wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
1142
1143                                 return true;
1144                         }
1145                 } elseif ( $match === true ) {
1146                         wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
1147
1148                         /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1149                         return true;
1150                 } else {
1151                         wfDebug( __METHOD__
1152                                 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1153
1154                         return false;
1155                 }
1156         }
1157
1158         /**
1159          * Heuristic for detecting files that *could* contain JavaScript instructions or
1160          * things that may look like HTML to a browser and are thus
1161          * potentially harmful. The present implementation will produce false
1162          * positives in some situations.
1163          *
1164          * @param string $file Pathname to the temporary upload file
1165          * @param string $mime The MIME type of the file
1166          * @param string $extension The extension of the file
1167          * @return bool True if the file contains something looking like embedded scripts
1168          */
1169         public static function detectScript( $file, $mime, $extension ) {
1170                 global $wgAllowTitlesInSVG;
1171
1172                 # ugly hack: for text files, always look at the entire file.
1173                 # For binary field, just check the first K.
1174
1175                 if ( strpos( $mime, 'text/' ) === 0 ) {
1176                         $chunk = file_get_contents( $file );
1177                 } else {
1178                         $fp = fopen( $file, 'rb' );
1179                         $chunk = fread( $fp, 1024 );
1180                         fclose( $fp );
1181                 }
1182
1183                 $chunk = strtolower( $chunk );
1184
1185                 if ( !$chunk ) {
1186                         return false;
1187                 }
1188
1189                 # decode from UTF-16 if needed (could be used for obfuscation).
1190                 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1191                         $enc = 'UTF-16BE';
1192                 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1193                         $enc = 'UTF-16LE';
1194                 } else {
1195                         $enc = null;
1196                 }
1197
1198                 if ( $enc ) {
1199                         $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1200                 }
1201
1202                 $chunk = trim( $chunk );
1203
1204                 /** @todo FIXME: Convert from UTF-16 if necessary! */
1205                 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
1206
1207                 # check for HTML doctype
1208                 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1209                         return true;
1210                 }
1211
1212                 // Some browsers will interpret obscure xml encodings as UTF-8, while
1213                 // PHP/expat will interpret the given encoding in the xml declaration (T49304)
1214                 if ( $extension == 'svg' || strpos( $mime, 'image/svg' ) === 0 ) {
1215                         if ( self::checkXMLEncodingMissmatch( $file ) ) {
1216                                 return true;
1217                         }
1218                 }
1219
1220                 /**
1221                  * Internet Explorer for Windows performs some really stupid file type
1222                  * autodetection which can cause it to interpret valid image files as HTML
1223                  * and potentially execute JavaScript, creating a cross-site scripting
1224                  * attack vectors.
1225                  *
1226                  * Apple's Safari browser also performs some unsafe file type autodetection
1227                  * which can cause legitimate files to be interpreted as HTML if the
1228                  * web server is not correctly configured to send the right content-type
1229                  * (or if you're really uploading plain text and octet streams!)
1230                  *
1231                  * Returns true if IE is likely to mistake the given file for HTML.
1232                  * Also returns true if Safari would mistake the given file for HTML
1233                  * when served with a generic content-type.
1234                  */
1235                 $tags = [
1236                         '<a href',
1237                         '<body',
1238                         '<head',
1239                         '<html', # also in safari
1240                         '<img',
1241                         '<pre',
1242                         '<script', # also in safari
1243                         '<table'
1244                 ];
1245
1246                 if ( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
1247                         $tags[] = '<title';
1248                 }
1249
1250                 foreach ( $tags as $tag ) {
1251                         if ( false !== strpos( $chunk, $tag ) ) {
1252                                 wfDebug( __METHOD__ . ": found something that may make it be mistaken for html: $tag\n" );
1253
1254                                 return true;
1255                         }
1256                 }
1257
1258                 /*
1259                  * look for JavaScript
1260                  */
1261
1262                 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1263                 $chunk = Sanitizer::decodeCharReferences( $chunk );
1264
1265                 # look for script-types
1266                 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1267                         wfDebug( __METHOD__ . ": found script types\n" );
1268
1269                         return true;
1270                 }
1271
1272                 # look for html-style script-urls
1273                 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1274                         wfDebug( __METHOD__ . ": found html-style script urls\n" );
1275
1276                         return true;
1277                 }
1278
1279                 # look for css-style script-urls
1280                 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1281                         wfDebug( __METHOD__ . ": found css-style script urls\n" );
1282
1283                         return true;
1284                 }
1285
1286                 wfDebug( __METHOD__ . ": no scripts found\n" );
1287
1288                 return false;
1289         }
1290
1291         /**
1292          * Check a whitelist of xml encodings that are known not to be interpreted differently
1293          * by the server's xml parser (expat) and some common browsers.
1294          *
1295          * @param string $file Pathname to the temporary upload file
1296          * @return bool True if the file contains an encoding that could be misinterpreted
1297          */
1298         public static function checkXMLEncodingMissmatch( $file ) {
1299                 global $wgSVGMetadataCutoff;
1300                 $contents = file_get_contents( $file, false, null, -1, $wgSVGMetadataCutoff );
1301                 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1302
1303                 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1304                         if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1305                                 && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1306                         ) {
1307                                 wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1308
1309                                 return true;
1310                         }
1311                 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1312                         // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1313                         // bytes. There shouldn't be a legitimate reason for this to happen.
1314                         wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1315
1316                         return true;
1317                 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1318                         // EBCDIC encoded XML
1319                         wfDebug( __METHOD__ . ": EBCDIC Encoded XML\n" );
1320
1321                         return true;
1322                 }
1323
1324                 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1325                 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1326                 $attemptEncodings = [ 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' ];
1327                 foreach ( $attemptEncodings as $encoding ) {
1328                         MediaWiki\suppressWarnings();
1329                         $str = iconv( $encoding, 'UTF-8', $contents );
1330                         MediaWiki\restoreWarnings();
1331                         if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1332                                 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1333                                         && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1334                                 ) {
1335                                         wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1336
1337                                         return true;
1338                                 }
1339                         } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1340                                 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1341                                 // bytes. There shouldn't be a legitimate reason for this to happen.
1342                                 wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1343
1344                                 return true;
1345                         }
1346                 }
1347
1348                 return false;
1349         }
1350
1351         /**
1352          * @param string $filename
1353          * @param bool $partial
1354          * @return mixed False of the file is verified (does not contain scripts), array otherwise.
1355          */
1356         protected function detectScriptInSvg( $filename, $partial ) {
1357                 $this->mSVGNSError = false;
1358                 $check = new XmlTypeCheck(
1359                         $filename,
1360                         [ $this, 'checkSvgScriptCallback' ],
1361                         true,
1362                         [
1363                                 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback',
1364                                 'external_dtd_handler' => 'UploadBase::checkSvgExternalDTD',
1365                         ]
1366                 );
1367                 if ( $check->wellFormed !== true ) {
1368                         // Invalid xml (T60553)
1369                         // But only when non-partial (T67724)
1370                         return $partial ? false : [ 'uploadinvalidxml' ];
1371                 } elseif ( $check->filterMatch ) {
1372                         if ( $this->mSVGNSError ) {
1373                                 return [ 'uploadscriptednamespace', $this->mSVGNSError ];
1374                         }
1375
1376                         return $check->filterMatchType;
1377                 }
1378
1379                 return false;
1380         }
1381
1382         /**
1383          * Callback to filter SVG Processing Instructions.
1384          * @param string $target Processing instruction name
1385          * @param string $data Processing instruction attribute and value
1386          * @return bool (true if the filter identified something bad)
1387          */
1388         public static function checkSvgPICallback( $target, $data ) {
1389                 // Don't allow external stylesheets (T59550)
1390                 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1391                         return [ 'upload-scripted-pi-callback' ];
1392                 }
1393
1394                 return false;
1395         }
1396
1397         /**
1398          * Verify that DTD urls referenced are only the standard dtds
1399          *
1400          * Browsers seem to ignore external dtds. However just to be on the
1401          * safe side, only allow dtds from the svg standard.
1402          *
1403          * @param string $type PUBLIC or SYSTEM
1404          * @param string $publicId The well-known public identifier for the dtd
1405          * @param string $systemId The url for the external dtd
1406          */
1407         public static function checkSvgExternalDTD( $type, $publicId, $systemId ) {
1408                 // This doesn't include the XHTML+MathML+SVG doctype since we don't
1409                 // allow XHTML anyways.
1410                 $allowedDTDs = [
1411                         'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd',
1412                         'http://www.w3.org/TR/2001/REC-SVG-20010904/DTD/svg10.dtd',
1413                         'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-basic.dtd',
1414                         'http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-tiny.dtd',
1415                         // https://phabricator.wikimedia.org/T168856
1416                         'http://www.w3.org/TR/2001/PR-SVG-20010719/DTD/svg10.dtd',
1417                 ];
1418                 if ( $type !== 'PUBLIC'
1419                         || !in_array( $systemId, $allowedDTDs )
1420                         || strpos( $publicId, "-//W3C//" ) !== 0
1421                 ) {
1422                         return [ 'upload-scripted-dtd' ];
1423                 }
1424                 return false;
1425         }
1426
1427         /**
1428          * @todo Replace this with a whitelist filter!
1429          * @param string $element
1430          * @param array $attribs
1431          * @return bool
1432          */
1433         public function checkSvgScriptCallback( $element, $attribs, $data = null ) {
1434
1435                 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1436
1437                 // We specifically don't include:
1438                 // http://www.w3.org/1999/xhtml (T62771)
1439                 static $validNamespaces = [
1440                         '',
1441                         'adobe:ns:meta/',
1442                         'http://creativecommons.org/ns#',
1443                         'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1444                         'http://ns.adobe.com/adobeillustrator/10.0/',
1445                         'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1446                         'http://ns.adobe.com/extensibility/1.0/',
1447                         'http://ns.adobe.com/flows/1.0/',
1448                         'http://ns.adobe.com/illustrator/1.0/',
1449                         'http://ns.adobe.com/imagereplacement/1.0/',
1450                         'http://ns.adobe.com/pdf/1.3/',
1451                         'http://ns.adobe.com/photoshop/1.0/',
1452                         'http://ns.adobe.com/saveforweb/1.0/',
1453                         'http://ns.adobe.com/variables/1.0/',
1454                         'http://ns.adobe.com/xap/1.0/',
1455                         'http://ns.adobe.com/xap/1.0/g/',
1456                         'http://ns.adobe.com/xap/1.0/g/img/',
1457                         'http://ns.adobe.com/xap/1.0/mm/',
1458                         'http://ns.adobe.com/xap/1.0/rights/',
1459                         'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1460                         'http://ns.adobe.com/xap/1.0/stype/font#',
1461                         'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1462                         'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1463                         'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1464                         'http://ns.adobe.com/xap/1.0/t/pg/',
1465                         'http://purl.org/dc/elements/1.1/',
1466                         'http://purl.org/dc/elements/1.1',
1467                         'http://schemas.microsoft.com/visio/2003/svgextensions/',
1468                         'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1469                         'http://taptrix.com/inkpad/svg_extensions',
1470                         'http://web.resource.org/cc/',
1471                         'http://www.freesoftware.fsf.org/bkchem/cdml',
1472                         'http://www.inkscape.org/namespaces/inkscape',
1473                         'http://www.opengis.net/gml',
1474                         'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1475                         'http://www.w3.org/2000/svg',
1476                         'http://www.w3.org/tr/rec-rdf-syntax/',
1477                         'http://www.w3.org/2000/01/rdf-schema#',
1478                 ];
1479
1480                 // Inkscape mangles namespace definitions created by Adobe Illustrator.
1481                 // This is nasty but harmless. (T144827)
1482                 $isBuggyInkscape = preg_match( '/^&(#38;)*ns_[a-z_]+;$/', $namespace );
1483
1484                 if ( !( $isBuggyInkscape || in_array( $namespace, $validNamespaces ) ) ) {
1485                         wfDebug( __METHOD__ . ": Non-svg namespace '$namespace' in uploaded file.\n" );
1486                         /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1487                         $this->mSVGNSError = $namespace;
1488
1489                         return true;
1490                 }
1491
1492                 /*
1493                  * check for elements that can contain javascript
1494                  */
1495                 if ( $strippedElement == 'script' ) {
1496                         wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
1497
1498                         return [ 'uploaded-script-svg', $strippedElement ];
1499                 }
1500
1501                 # e.g., <svg xmlns="http://www.w3.org/2000/svg">
1502                 #  <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1503                 if ( $strippedElement == 'handler' ) {
1504                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1505
1506                         return [ 'uploaded-script-svg', $strippedElement ];
1507                 }
1508
1509                 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1510                 if ( $strippedElement == 'stylesheet' ) {
1511                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1512
1513                         return [ 'uploaded-script-svg', $strippedElement ];
1514                 }
1515
1516                 # Block iframes, in case they pass the namespace check
1517                 if ( $strippedElement == 'iframe' ) {
1518                         wfDebug( __METHOD__ . ": iframe in uploaded file.\n" );
1519
1520                         return [ 'uploaded-script-svg', $strippedElement ];
1521                 }
1522
1523                 # Check <style> css
1524                 if ( $strippedElement == 'style'
1525                         && self::checkCssFragment( Sanitizer::normalizeCss( $data ) )
1526                 ) {
1527                         wfDebug( __METHOD__ . ": hostile css in style element.\n" );
1528                         return [ 'uploaded-hostile-svg' ];
1529                 }
1530
1531                 foreach ( $attribs as $attrib => $value ) {
1532                         $stripped = $this->stripXmlNamespace( $attrib );
1533                         $value = strtolower( $value );
1534
1535                         if ( substr( $stripped, 0, 2 ) == 'on' ) {
1536                                 wfDebug( __METHOD__
1537                                         . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1538
1539                                 return [ 'uploaded-event-handler-on-svg', $attrib, $value ];
1540                         }
1541
1542                         # Do not allow relative links, or unsafe url schemas.
1543                         # For <a> tags, only data:, http: and https: and same-document
1544                         # fragment links are allowed. For all other tags, only data:
1545                         # and fragment are allowed.
1546                         if ( $stripped == 'href'
1547                                 && $value !== ''
1548                                 && strpos( $value, 'data:' ) !== 0
1549                                 && strpos( $value, '#' ) !== 0
1550                         ) {
1551                                 if ( !( $strippedElement === 'a'
1552                                         && preg_match( '!^https?://!i', $value ) )
1553                                 ) {
1554                                         wfDebug( __METHOD__ . ": Found href attribute <$strippedElement "
1555                                                 . "'$attrib'='$value' in uploaded file.\n" );
1556
1557                                         return [ 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value ];
1558                                 }
1559                         }
1560
1561                         # only allow data: targets that should be safe. This prevents vectors like,
1562                         # image/svg, text/xml, application/xml, and text/html, which can contain scripts
1563                         if ( $stripped == 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) {
1564                                 // rfc2397 parameters. This is only slightly slower than (;[\w;]+)*.
1565                                 // @codingStandardsIgnoreStart Generic.Files.LineLength
1566                                 $parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?';
1567                                 // @codingStandardsIgnoreEnd
1568
1569                                 if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|png)$parameters,!i", $value ) ) {
1570                                         wfDebug( __METHOD__ . ": Found href to unwhitelisted data: uri "
1571                                                 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1572                                         return [ 'uploaded-href-unsafe-target-svg', $strippedElement, $attrib, $value ];
1573                                 }
1574                         }
1575
1576                         # Change href with animate from (http://html5sec.org/#137).
1577                         if ( $stripped === 'attributename'
1578                                 && $strippedElement === 'animate'
1579                                 && $this->stripXmlNamespace( $value ) == 'href'
1580                         ) {
1581                                 wfDebug( __METHOD__ . ": Found animate that might be changing href using from "
1582                                         . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1583
1584                                 return [ 'uploaded-animate-svg', $strippedElement, $attrib, $value ];
1585                         }
1586
1587                         # use set/animate to add event-handler attribute to parent
1588                         if ( ( $strippedElement == 'set' || $strippedElement == 'animate' )
1589                                 && $stripped == 'attributename'
1590                                 && substr( $value, 0, 2 ) == 'on'
1591                         ) {
1592                                 wfDebug( __METHOD__ . ": Found svg setting event-handler attribute with "
1593                                         . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1594
1595                                 return [ 'uploaded-setting-event-handler-svg', $strippedElement, $stripped, $value ];
1596                         }
1597
1598                         # use set to add href attribute to parent element
1599                         if ( $strippedElement == 'set'
1600                                 && $stripped == 'attributename'
1601                                 && strpos( $value, 'href' ) !== false
1602                         ) {
1603                                 wfDebug( __METHOD__ . ": Found svg setting href attribute '$value' in uploaded file.\n" );
1604
1605                                 return [ 'uploaded-setting-href-svg' ];
1606                         }
1607
1608                         # use set to add a remote / data / script target to an element
1609                         if ( $strippedElement == 'set'
1610                                 && $stripped == 'to'
1611                                 && preg_match( '!(http|https|data|script):!sim', $value )
1612                         ) {
1613                                 wfDebug( __METHOD__ . ": Found svg setting attribute to '$value' in uploaded file.\n" );
1614
1615                                 return [ 'uploaded-wrong-setting-svg', $value ];
1616                         }
1617
1618                         # use handler attribute with remote / data / script
1619                         if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1620                                 wfDebug( __METHOD__ . ": Found svg setting handler with remote/data/script "
1621                                         . "'$attrib'='$value' in uploaded file.\n" );
1622
1623                                 return [ 'uploaded-setting-handler-svg', $attrib, $value ];
1624                         }
1625
1626                         # use CSS styles to bring in remote code
1627                         if ( $stripped == 'style'
1628                                 && self::checkCssFragment( Sanitizer::normalizeCss( $value ) )
1629                         ) {
1630                                 wfDebug( __METHOD__ . ": Found svg setting a style with "
1631                                         . "remote url '$attrib'='$value' in uploaded file.\n" );
1632                                 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1633                         }
1634
1635                         # Several attributes can include css, css character escaping isn't allowed
1636                         $cssAttrs = [ 'font', 'clip-path', 'fill', 'filter', 'marker',
1637                                 'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke' ];
1638                         if ( in_array( $stripped, $cssAttrs )
1639                                 && self::checkCssFragment( $value )
1640                         ) {
1641                                 wfDebug( __METHOD__ . ": Found svg setting a style with "
1642                                         . "remote url '$attrib'='$value' in uploaded file.\n" );
1643                                 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1644                         }
1645
1646                         # image filters can pull in url, which could be svg that executes scripts
1647                         if ( $strippedElement == 'image'
1648                                 && $stripped == 'filter'
1649                                 && preg_match( '!url\s*\(!sim', $value )
1650                         ) {
1651                                 wfDebug( __METHOD__ . ": Found image filter with url: "
1652                                         . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1653
1654                                 return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ];
1655                         }
1656                 }
1657
1658                 return false; // No scripts detected
1659         }
1660
1661         /**
1662          * Check a block of CSS or CSS fragment for anything that looks like
1663          * it is bringing in remote code.
1664          * @param string $value a string of CSS
1665          * @param bool $propOnly only check css properties (start regex with :)
1666          * @return bool true if the CSS contains an illegal string, false if otherwise
1667          */
1668         private static function checkCssFragment( $value ) {
1669
1670                 # Forbid external stylesheets, for both reliability and to protect viewer's privacy
1671                 if ( stripos( $value, '@import' ) !== false ) {
1672                         return true;
1673                 }
1674
1675                 # We allow @font-face to embed fonts with data: urls, so we snip the string
1676                 # 'url' out so this case won't match when we check for urls below
1677                 $pattern = '!(@font-face\s*{[^}]*src:)url(\("data:;base64,)!im';
1678                 $value = preg_replace( $pattern, '$1$2', $value );
1679
1680                 # Check for remote and executable CSS. Unlike in Sanitizer::checkCss, the CSS
1681                 # properties filter and accelerator don't seem to be useful for xss in SVG files.
1682                 # Expression and -o-link don't seem to work either, but filtering them here in case.
1683                 # Additionally, we catch remote urls like url("http:..., url('http:..., url(http:...,
1684                 # but not local ones such as url("#..., url('#..., url(#....
1685                 if ( preg_match( '!expression
1686                                 | -o-link\s*:
1687                                 | -o-link-source\s*:
1688                                 | -o-replace\s*:!imx', $value ) ) {
1689                         return true;
1690                 }
1691
1692                 if ( preg_match_all(
1693                                 "!(\s*(url|image|image-set)\s*\(\s*[\"']?\s*[^#]+.*?\))!sim",
1694                                 $value,
1695                                 $matches
1696                         ) !== 0
1697                 ) {
1698                         # TODO: redo this in one regex. Until then, url("#whatever") matches the first
1699                         foreach ( $matches[1] as $match ) {
1700                                 if ( !preg_match( "!\s*(url|image|image-set)\s*\(\s*(#|'#|\"#)!im", $match ) ) {
1701                                         return true;
1702                                 }
1703                         }
1704                 }
1705
1706                 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
1707                         return true;
1708                 }
1709
1710                 return false;
1711         }
1712
1713         /**
1714          * Divide the element name passed by the xml parser to the callback into URI and prifix.
1715          * @param string $element
1716          * @return array Containing the namespace URI and prefix
1717          */
1718         private static function splitXmlNamespace( $element ) {
1719                 // 'http://www.w3.org/2000/svg:script' -> [ 'http://www.w3.org/2000/svg', 'script' ]
1720                 $parts = explode( ':', strtolower( $element ) );
1721                 $name = array_pop( $parts );
1722                 $ns = implode( ':', $parts );
1723
1724                 return [ $ns, $name ];
1725         }
1726
1727         /**
1728          * @param string $name
1729          * @return string
1730          */
1731         private function stripXmlNamespace( $name ) {
1732                 // 'http://www.w3.org/2000/svg:script' -> 'script'
1733                 $parts = explode( ':', strtolower( $name ) );
1734
1735                 return array_pop( $parts );
1736         }
1737
1738         /**
1739          * Generic wrapper function for a virus scanner program.
1740          * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1741          * $wgAntivirusRequired may be used to deny upload if the scan fails.
1742          *
1743          * @param string $file Pathname to the temporary upload file
1744          * @return mixed False if not virus is found, null if the scan fails or is disabled,
1745          *   or a string containing feedback from the virus scanner if a virus was found.
1746          *   If textual feedback is missing but a virus was found, this function returns true.
1747          */
1748         public static function detectVirus( $file ) {
1749                 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1750
1751                 if ( !$wgAntivirus ) {
1752                         wfDebug( __METHOD__ . ": virus scanner disabled\n" );
1753
1754                         return null;
1755                 }
1756
1757                 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1758                         wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
1759                         $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1760                                 [ 'virus-badscanner', $wgAntivirus ] );
1761
1762                         return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1763                 }
1764
1765                 # look up scanner configuration
1766                 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1767                 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1768                 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
1769                         $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
1770
1771                 if ( strpos( $command, "%f" ) === false ) {
1772                         # simple pattern: append file to scan
1773                         $command .= " " . wfEscapeShellArg( $file );
1774                 } else {
1775                         # complex pattern: replace "%f" with file to scan
1776                         $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
1777                 }
1778
1779                 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
1780
1781                 # execute virus scanner
1782                 $exitCode = false;
1783
1784                 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1785                 #      that does not seem to be worth the pain.
1786                 #      Ask me (Duesentrieb) about it if it's ever needed.
1787                 $output = wfShellExecWithStderr( $command, $exitCode );
1788
1789                 # map exit code to AV_xxx constants.
1790                 $mappedCode = $exitCode;
1791                 if ( $exitCodeMap ) {
1792                         if ( isset( $exitCodeMap[$exitCode] ) ) {
1793                                 $mappedCode = $exitCodeMap[$exitCode];
1794                         } elseif ( isset( $exitCodeMap["*"] ) ) {
1795                                 $mappedCode = $exitCodeMap["*"];
1796                         }
1797                 }
1798
1799                 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1800                  * so we need the strict equalities === and thus can't use a switch here
1801                  */
1802                 if ( $mappedCode === AV_SCAN_FAILED ) {
1803                         # scan failed (code was mapped to false by $exitCodeMap)
1804                         wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
1805
1806                         $output = $wgAntivirusRequired
1807                                 ? wfMessage( 'virus-scanfailed', [ $exitCode ] )->text()
1808                                 : null;
1809                 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
1810                         # scan failed because filetype is unknown (probably imune)
1811                         wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
1812                         $output = null;
1813                 } elseif ( $mappedCode === AV_NO_VIRUS ) {
1814                         # no virus found
1815                         wfDebug( __METHOD__ . ": file passed virus scan.\n" );
1816                         $output = false;
1817                 } else {
1818                         $output = trim( $output );
1819
1820                         if ( !$output ) {
1821                                 $output = true; # if there's no output, return true
1822                         } elseif ( $msgPattern ) {
1823                                 $groups = [];
1824                                 if ( preg_match( $msgPattern, $output, $groups ) ) {
1825                                         if ( $groups[1] ) {
1826                                                 $output = $groups[1];
1827                                         }
1828                                 }
1829                         }
1830
1831                         wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
1832                 }
1833
1834                 return $output;
1835         }
1836
1837         /**
1838          * Check if there's an overwrite conflict and, if so, if restrictions
1839          * forbid this user from performing the upload.
1840          *
1841          * @param User $user
1842          *
1843          * @return mixed True on success, array on failure
1844          */
1845         private function checkOverwrite( $user ) {
1846                 // First check whether the local file can be overwritten
1847                 $file = $this->getLocalFile();
1848                 $file->load( File::READ_LATEST );
1849                 if ( $file->exists() ) {
1850                         if ( !self::userCanReUpload( $user, $file ) ) {
1851                                 return [ 'fileexists-forbidden', $file->getName() ];
1852                         } else {
1853                                 return true;
1854                         }
1855                 }
1856
1857                 /* Check shared conflicts: if the local file does not exist, but
1858                  * wfFindFile finds a file, it exists in a shared repository.
1859                  */
1860                 $file = wfFindFile( $this->getTitle(), [ 'latest' => true ] );
1861                 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1862                         return [ 'fileexists-shared-forbidden', $file->getName() ];
1863                 }
1864
1865                 return true;
1866         }
1867
1868         /**
1869          * Check if a user is the last uploader
1870          *
1871          * @param User $user
1872          * @param File $img
1873          * @return bool
1874          */
1875         public static function userCanReUpload( User $user, File $img ) {
1876                 if ( $user->isAllowed( 'reupload' ) ) {
1877                         return true; // non-conditional
1878                 } elseif ( !$user->isAllowed( 'reupload-own' ) ) {
1879                         return false;
1880                 }
1881
1882                 if ( !( $img instanceof LocalFile ) ) {
1883                         return false;
1884                 }
1885
1886                 $img->load();
1887
1888                 return $user->getId() == $img->getUser( 'id' );
1889         }
1890
1891         /**
1892          * Helper function that does various existence checks for a file.
1893          * The following checks are performed:
1894          * - The file exists
1895          * - Article with the same name as the file exists
1896          * - File exists with normalized extension
1897          * - The file looks like a thumbnail and the original exists
1898          *
1899          * @param File $file The File object to check
1900          * @return mixed False if the file does not exists, else an array
1901          */
1902         public static function getExistsWarning( $file ) {
1903                 if ( $file->exists() ) {
1904                         return [ 'warning' => 'exists', 'file' => $file ];
1905                 }
1906
1907                 if ( $file->getTitle()->getArticleID() ) {
1908                         return [ 'warning' => 'page-exists', 'file' => $file ];
1909                 }
1910
1911                 if ( strpos( $file->getName(), '.' ) == false ) {
1912                         $partname = $file->getName();
1913                         $extension = '';
1914                 } else {
1915                         $n = strrpos( $file->getName(), '.' );
1916                         $extension = substr( $file->getName(), $n + 1 );
1917                         $partname = substr( $file->getName(), 0, $n );
1918                 }
1919                 $normalizedExtension = File::normalizeExtension( $extension );
1920
1921                 if ( $normalizedExtension != $extension ) {
1922                         // We're not using the normalized form of the extension.
1923                         // Normal form is lowercase, using most common of alternate
1924                         // extensions (eg 'jpg' rather than 'JPEG').
1925
1926                         // Check for another file using the normalized form...
1927                         $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
1928                         $file_lc = wfLocalFile( $nt_lc );
1929
1930                         if ( $file_lc->exists() ) {
1931                                 return [
1932                                         'warning' => 'exists-normalized',
1933                                         'file' => $file,
1934                                         'normalizedFile' => $file_lc
1935                                 ];
1936                         }
1937                 }
1938
1939                 // Check for files with the same name but a different extension
1940                 $similarFiles = RepoGroup::singleton()->getLocalRepo()->findFilesByPrefix(
1941                         "{$partname}.", 1 );
1942                 if ( count( $similarFiles ) ) {
1943                         return [
1944                                 'warning' => 'exists-normalized',
1945                                 'file' => $file,
1946                                 'normalizedFile' => $similarFiles[0],
1947                         ];
1948                 }
1949
1950                 if ( self::isThumbName( $file->getName() ) ) {
1951                         # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1952                         $nt_thb = Title::newFromText(
1953                                 substr( $partname, strpos( $partname, '-' ) + 1 ) . '.' . $extension,
1954                                 NS_FILE
1955                         );
1956                         $file_thb = wfLocalFile( $nt_thb );
1957                         if ( $file_thb->exists() ) {
1958                                 return [
1959                                         'warning' => 'thumb',
1960                                         'file' => $file,
1961                                         'thumbFile' => $file_thb
1962                                 ];
1963                         } else {
1964                                 // File does not exist, but we just don't like the name
1965                                 return [
1966                                         'warning' => 'thumb-name',
1967                                         'file' => $file,
1968                                         'thumbFile' => $file_thb
1969                                 ];
1970                         }
1971                 }
1972
1973                 foreach ( self::getFilenamePrefixBlacklist() as $prefix ) {
1974                         if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1975                                 return [
1976                                         'warning' => 'bad-prefix',
1977                                         'file' => $file,
1978                                         'prefix' => $prefix
1979                                 ];
1980                         }
1981                 }
1982
1983                 return false;
1984         }
1985
1986         /**
1987          * Helper function that checks whether the filename looks like a thumbnail
1988          * @param string $filename
1989          * @return bool
1990          */
1991         public static function isThumbName( $filename ) {
1992                 $n = strrpos( $filename, '.' );
1993                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
1994
1995                 return (
1996                         substr( $partname, 3, 3 ) == 'px-' ||
1997                         substr( $partname, 2, 3 ) == 'px-'
1998                 ) &&
1999                 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
2000         }
2001
2002         /**
2003          * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
2004          *
2005          * @return array List of prefixes
2006          */
2007         public static function getFilenamePrefixBlacklist() {
2008                 $blacklist = [];
2009                 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
2010                 if ( !$message->isDisabled() ) {
2011                         $lines = explode( "\n", $message->plain() );
2012                         foreach ( $lines as $line ) {
2013                                 // Remove comment lines
2014                                 $comment = substr( trim( $line ), 0, 1 );
2015                                 if ( $comment == '#' || $comment == '' ) {
2016                                         continue;
2017                                 }
2018                                 // Remove additional comments after a prefix
2019                                 $comment = strpos( $line, '#' );
2020                                 if ( $comment > 0 ) {
2021                                         $line = substr( $line, 0, $comment - 1 );
2022                                 }
2023                                 $blacklist[] = trim( $line );
2024                         }
2025                 }
2026
2027                 return $blacklist;
2028         }
2029
2030         /**
2031          * Gets image info about the file just uploaded.
2032          *
2033          * Also has the effect of setting metadata to be an 'indexed tag name' in
2034          * returned API result if 'metadata' was requested. Oddly, we have to pass
2035          * the "result" object down just so it can do that with the appropriate
2036          * format, presumably.
2037          *
2038          * @param ApiResult $result
2039          * @return array Image info
2040          */
2041         public function getImageInfo( $result ) {
2042                 $localFile = $this->getLocalFile();
2043                 $stashFile = $this->getStashFile();
2044                 // Calling a different API module depending on whether the file was stashed is less than optimal.
2045                 // In fact, calling API modules here at all is less than optimal. Maybe it should be refactored.
2046                 if ( $stashFile ) {
2047                         $imParam = ApiQueryStashImageInfo::getPropertyNames();
2048                         $info = ApiQueryStashImageInfo::getInfo( $stashFile, array_flip( $imParam ), $result );
2049                 } else {
2050                         $imParam = ApiQueryImageInfo::getPropertyNames();
2051                         $info = ApiQueryImageInfo::getInfo( $localFile, array_flip( $imParam ), $result );
2052                 }
2053
2054                 return $info;
2055         }
2056
2057         /**
2058          * @param array $error
2059          * @return Status
2060          */
2061         public function convertVerifyErrorToStatus( $error ) {
2062                 $code = $error['status'];
2063                 unset( $code['status'] );
2064
2065                 return Status::newFatal( $this->getVerificationErrorCode( $code ), $error );
2066         }
2067
2068         /**
2069          * Get the MediaWiki maximum uploaded file size for given type of upload, based on
2070          * $wgMaxUploadSize.
2071          *
2072          * @param null|string $forType
2073          * @return int
2074          */
2075         public static function getMaxUploadSize( $forType = null ) {
2076                 global $wgMaxUploadSize;
2077
2078                 if ( is_array( $wgMaxUploadSize ) ) {
2079                         if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
2080                                 return $wgMaxUploadSize[$forType];
2081                         } else {
2082                                 return $wgMaxUploadSize['*'];
2083                         }
2084                 } else {
2085                         return intval( $wgMaxUploadSize );
2086                 }
2087         }
2088
2089         /**
2090          * Get the PHP maximum uploaded file size, based on ini settings. If there is no limit or the
2091          * limit can't be guessed, returns a very large number (PHP_INT_MAX).
2092          *
2093          * @since 1.27
2094          * @return int
2095          */
2096         public static function getMaxPhpUploadSize() {
2097                 $phpMaxFileSize = wfShorthandToInteger(
2098                         ini_get( 'upload_max_filesize' ) ?: ini_get( 'hhvm.server.upload.upload_max_file_size' ),
2099                         PHP_INT_MAX
2100                 );
2101                 $phpMaxPostSize = wfShorthandToInteger(
2102                         ini_get( 'post_max_size' ) ?: ini_get( 'hhvm.server.max_post_size' ),
2103                         PHP_INT_MAX
2104                 ) ?: PHP_INT_MAX;
2105                 return min( $phpMaxFileSize, $phpMaxPostSize );
2106         }
2107
2108         /**
2109          * Get the current status of a chunked upload (used for polling)
2110          *
2111          * The value will be read from cache.
2112          *
2113          * @param User $user
2114          * @param string $statusKey
2115          * @return Status[]|bool
2116          */
2117         public static function getSessionStatus( User $user, $statusKey ) {
2118                 $cache = MediaWikiServices::getInstance()->getMainObjectStash();
2119                 $key = $cache->makeKey( 'uploadstatus', $user->getId() ?: md5( $user->getName() ), $statusKey );
2120
2121                 return $cache->get( $key );
2122         }
2123
2124         /**
2125          * Set the current status of a chunked upload (used for polling)
2126          *
2127          * The value will be set in cache for 1 day
2128          *
2129          * @param User $user
2130          * @param string $statusKey
2131          * @param array|bool $value
2132          * @return void
2133          */
2134         public static function setSessionStatus( User $user, $statusKey, $value ) {
2135                 $cache = MediaWikiServices::getInstance()->getMainObjectStash();
2136                 $key = $cache->makeKey( 'uploadstatus', $user->getId() ?: md5( $user->getName() ), $statusKey );
2137
2138                 if ( $value === false ) {
2139                         $cache->delete( $key );
2140                 } else {
2141                         $cache->set( $key, $value, $cache::TTL_DAY );
2142                 }
2143         }
2144 }