includes/upload/UploadBase.php

   1 <?php
   2 /**
   3  * Base class for the backend of file upload.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Upload
  22  */
  23
  24 /**
  25  * @defgroup Upload Upload related
  26  */
  27
  28 /**
  29  * @ingroup Upload
  30  *
  31  * UploadBase and subclasses are the backend of MediaWiki's file uploads.
  32  * The frontends are formed by ApiUpload and SpecialUpload.
  33  *
  34  * @author Brion Vibber
  35  * @author Bryan Tong Minh
  36  * @author Michael Dale
  37  */
  38 abstract class UploadBase {
  39         /** @var string Local file system path to the file to upload (or a local copy) */
  40         protected $mTempPath;
  41         /** @var TempFSFile|null Wrapper to handle deleting the temp file */
  42         protected $tempFileObj;
  43
  44         protected $mDesiredDestName, $mDestName, $mRemoveTempFile, $mSourceType;
  45         protected $mTitle = false, $mTitleError = 0;
  46         protected $mFilteredName, $mFinalExtension;
  47         protected $mLocalFile, $mStashFile, $mFileSize, $mFileProps;
  48         protected $mBlackListedExtensions;
  49         protected $mJavaDetected, $mSVGNSError;
  50
  51         protected static $safeXmlEncodings = [
  52                 'UTF-8',
  53                 'ISO-8859-1',
  54                 'ISO-8859-2',
  55                 'UTF-16',
  56                 'UTF-32',
  57                 'WINDOWS-1250',
  58                 'WINDOWS-1251',
  59                 'WINDOWS-1252',
  60                 'WINDOWS-1253',
  61                 'WINDOWS-1254',
  62                 'WINDOWS-1255',
  63                 'WINDOWS-1256',
  64                 'WINDOWS-1257',
  65                 'WINDOWS-1258',
  66         ];
  67
  68         const SUCCESS = 0;
  69         const OK = 0;
  70         const EMPTY_FILE = 3;
  71         const MIN_LENGTH_PARTNAME = 4;
  72         const ILLEGAL_FILENAME = 5;
  73         const OVERWRITE_EXISTING_FILE = 7; # Not used anymore; handled by verifyTitlePermissions()
  74         const FILETYPE_MISSING = 8;
  75         const FILETYPE_BADTYPE = 9;
  76         const VERIFICATION_ERROR = 10;
  77         const HOOK_ABORTED = 11;
  78         const FILE_TOO_LARGE = 12;
  79         const WINDOWS_NONASCII_FILENAME = 13;
  80         const FILENAME_TOO_LONG = 14;
  81
  82         /**
  83          * @param int $error
  84          * @return string
  85          */
  86         public function getVerificationErrorCode( $error ) {
  87                 $code_to_status = [
  88                         self::EMPTY_FILE => 'empty-file',
  89                         self::FILE_TOO_LARGE => 'file-too-large',
  90                         self::FILETYPE_MISSING => 'filetype-missing',
  91                         self::FILETYPE_BADTYPE => 'filetype-banned',
  92                         self::MIN_LENGTH_PARTNAME => 'filename-tooshort',
  93                         self::ILLEGAL_FILENAME => 'illegal-filename',
  94                         self::OVERWRITE_EXISTING_FILE => 'overwrite',
  95                         self::VERIFICATION_ERROR => 'verification-error',
  96                         self::HOOK_ABORTED => 'hookaborted',
  97                         self::WINDOWS_NONASCII_FILENAME => 'windows-nonascii-filename',
  98                         self::FILENAME_TOO_LONG => 'filename-toolong',
  99                 ];
 100                 if ( isset( $code_to_status[$error] ) ) {
 101                         return $code_to_status[$error];
 102                 }
 103
 104                 return 'unknown-error';
 105         }
 106
 107         /**
 108          * Returns true if uploads are enabled.
 109          * Can be override by subclasses.
 110          * @return bool
 111          */
 112         public static function isEnabled() {
 113                 global $wgEnableUploads;
 114
 115                 if ( !$wgEnableUploads ) {
 116                         return false;
 117                 }
 118
 119                 # Check php's file_uploads setting
 120                 return wfIsHHVM() || wfIniGetBool( 'file_uploads' );
 121         }
 122
 123         /**
 124          * Returns true if the user can use this upload module or else a string
 125          * identifying the missing permission.
 126          * Can be overridden by subclasses.
 127          *
 128          * @param User $user
 129          * @return bool|string
 130          */
 131         public static function isAllowed( $user ) {
 132                 foreach ( [ 'upload', 'edit' ] as $permission ) {
 133                         if ( !$user->isAllowed( $permission ) ) {
 134                                 return $permission;
 135                         }
 136                 }
 137
 138                 return true;
 139         }
 140
 141         /**
 142          * Returns true if the user has surpassed the upload rate limit, false otherwise.
 143          *
 144          * @param User $user
 145          * @return bool
 146          */
 147         public static function isThrottled( $user ) {
 148                 return $user->pingLimiter( 'upload' );
 149         }
 150
 151         // Upload handlers. Should probably just be a global.
 152         private static $uploadHandlers = [ 'Stash', 'File', 'Url' ];
 153
 154         /**
 155          * Create a form of UploadBase depending on wpSourceType and initializes it
 156          *
 157          * @param WebRequest $request
 158          * @param string|null $type
 159          * @return null|UploadBase
 160          */
 161         public static function createFromRequest( &$request, $type = null ) {
 162                 $type = $type ? $type : $request->getVal( 'wpSourceType', 'File' );
 163
 164                 if ( !$type ) {
 165                         return null;
 166                 }
 167
 168                 // Get the upload class
 169                 $type = ucfirst( $type );
 170
 171                 // Give hooks the chance to handle this request
 172                 $className = null;
 173                 Hooks::run( 'UploadCreateFromRequest', [ $type, &$className ] );
 174                 if ( is_null( $className ) ) {
 175                         $className = 'UploadFrom' . $type;
 176                         wfDebug( __METHOD__ . ": class name: $className\n" );
 177                         if ( !in_array( $type, self::$uploadHandlers ) ) {
 178                                 return null;
 179                         }
 180                 }
 181
 182                 // Check whether this upload class is enabled
 183                 if ( !call_user_func( [ $className, 'isEnabled' ] ) ) {
 184                         return null;
 185                 }
 186
 187                 // Check whether the request is valid
 188                 if ( !call_user_func( [ $className, 'isValidRequest' ], $request ) ) {
 189                         return null;
 190                 }
 191
 192                 /** @var UploadBase $handler */
 193                 $handler = new $className;
 194
 195                 $handler->initializeFromRequest( $request );
 196
 197                 return $handler;
 198         }
 199
 200         /**
 201          * Check whether a request if valid for this handler
 202          * @param WebRequest $request
 203          * @return bool
 204          */
 205         public static function isValidRequest( $request ) {
 206                 return false;
 207         }
 208
 209         public function __construct() {
 210         }
 211
 212         /**
 213          * Returns the upload type. Should be overridden by child classes
 214          *
 215          * @since 1.18
 216          * @return string
 217          */
 218         public function getSourceType() {
 219                 return null;
 220         }
 221
 222         /**
 223          * Initialize the path information
 224          * @param string $name The desired destination name
 225          * @param string $tempPath The temporary path
 226          * @param int $fileSize The file size
 227          * @param bool $removeTempFile (false) remove the temporary file?
 228          * @throws MWException
 229          */
 230         public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
 231                 $this->mDesiredDestName = $name;
 232                 if ( FileBackend::isStoragePath( $tempPath ) ) {
 233                         throw new MWException( __METHOD__ . " given storage path `$tempPath`." );
 234                 }
 235
 236                 $this->setTempFile( $tempPath, $fileSize );
 237                 $this->mRemoveTempFile = $removeTempFile;
 238         }
 239
 240         /**
 241          * Initialize from a WebRequest. Override this in a subclass.
 242          *
 243          * @param WebRequest $request
 244          */
 245         abstract public function initializeFromRequest( &$request );
 246
 247         /**
 248          * @param string $tempPath File system path to temporary file containing the upload
 249          * @param integer $fileSize
 250          */
 251         protected function setTempFile( $tempPath, $fileSize = null ) {
 252                 $this->mTempPath = $tempPath;
 253                 $this->mFileSize = $fileSize ?: null;
 254                 if ( strlen( $this->mTempPath ) && file_exists( $this->mTempPath ) ) {
 255                         $this->tempFileObj = new TempFSFile( $this->mTempPath );
 256                         if ( !$fileSize ) {
 257                                 $this->mFileSize = filesize( $this->mTempPath );
 258                         }
 259                 } else {
 260                         $this->tempFileObj = null;
 261                 }
 262         }
 263
 264         /**
 265          * Fetch the file. Usually a no-op
 266          * @return Status
 267          */
 268         public function fetchFile() {
 269                 return Status::newGood();
 270         }
 271
 272         /**
 273          * Return true if the file is empty
 274          * @return bool
 275          */
 276         public function isEmptyFile() {
 277                 return empty( $this->mFileSize );
 278         }
 279
 280         /**
 281          * Return the file size
 282          * @return int
 283          */
 284         public function getFileSize() {
 285                 return $this->mFileSize;
 286         }
 287
 288         /**
 289          * Get the base 36 SHA1 of the file
 290          * @return string
 291          */
 292         public function getTempFileSha1Base36() {
 293                 return FSFile::getSha1Base36FromPath( $this->mTempPath );
 294         }
 295
 296         /**
 297          * @param string $srcPath The source path
 298          * @return string|bool The real path if it was a virtual URL Returns false on failure
 299          */
 300         function getRealPath( $srcPath ) {
 301                 $repo = RepoGroup::singleton()->getLocalRepo();
 302                 if ( $repo->isVirtualUrl( $srcPath ) ) {
 303                         /** @todo Just make uploads work with storage paths UploadFromStash
 304                          *  loads files via virtual URLs.
 305                          */
 306                         $tmpFile = $repo->getLocalCopy( $srcPath );
 307                         if ( $tmpFile ) {
 308                                 $tmpFile->bind( $this ); // keep alive with $this
 309                         }
 310                         $path = $tmpFile ? $tmpFile->getPath() : false;
 311                 } else {
 312                         $path = $srcPath;
 313                 }
 314
 315                 return $path;
 316         }
 317
 318         /**
 319          * Verify whether the upload is sane.
 320          * @return mixed Const self::OK or else an array with error information
 321          */
 322         public function verifyUpload() {
 323
 324                 /**
 325                  * If there was no filename or a zero size given, give up quick.
 326                  */
 327                 if ( $this->isEmptyFile() ) {
 328                         return [ 'status' => self::EMPTY_FILE ];
 329                 }
 330
 331                 /**
 332                  * Honor $wgMaxUploadSize
 333                  */
 334                 $maxSize = self::getMaxUploadSize( $this->getSourceType() );
 335                 if ( $this->mFileSize > $maxSize ) {
 336                         return [
 337                                 'status' => self::FILE_TOO_LARGE,
 338                                 'max' => $maxSize,
 339                         ];
 340                 }
 341
 342                 /**
 343                  * Look at the contents of the file; if we can recognize the
 344                  * type but it's corrupt or data of the wrong type, we should
 345                  * probably not accept it.
 346                  */
 347                 $verification = $this->verifyFile();
 348                 if ( $verification !== true ) {
 349                         return [
 350                                 'status' => self::VERIFICATION_ERROR,
 351                                 'details' => $verification
 352                         ];
 353                 }
 354
 355                 /**
 356                  * Make sure this file can be created
 357                  */
 358                 $result = $this->validateName();
 359                 if ( $result !== true ) {
 360                         return $result;
 361                 }
 362
 363                 $error = '';
 364                 if ( !Hooks::run( 'UploadVerification',
 365                         [ $this->mDestName, $this->mTempPath, &$error ], '1.28' )
 366                 ) {
 367                         return [ 'status' => self::HOOK_ABORTED, 'error' => $error ];
 368                 }
 369
 370                 return [ 'status' => self::OK ];
 371         }
 372
 373         /**
 374          * Verify that the name is valid and, if necessary, that we can overwrite
 375          *
 376          * @return mixed True if valid, otherwise and array with 'status'
 377          * and other keys
 378          */
 379         public function validateName() {
 380                 $nt = $this->getTitle();
 381                 if ( is_null( $nt ) ) {
 382                         $result = [ 'status' => $this->mTitleError ];
 383                         if ( $this->mTitleError == self::ILLEGAL_FILENAME ) {
 384                                 $result['filtered'] = $this->mFilteredName;
 385                         }
 386                         if ( $this->mTitleError == self::FILETYPE_BADTYPE ) {
 387                                 $result['finalExt'] = $this->mFinalExtension;
 388                                 if ( count( $this->mBlackListedExtensions ) ) {
 389                                         $result['blacklistedExt'] = $this->mBlackListedExtensions;
 390                                 }
 391                         }
 392
 393                         return $result;
 394                 }
 395                 $this->mDestName = $this->getLocalFile()->getName();
 396
 397                 return true;
 398         }
 399
 400         /**
 401          * Verify the MIME type.
 402          *
 403          * @note Only checks that it is not an evil MIME. The "does it have
 404          *  correct extension given its MIME type?" check is in verifyFile.
 405          *  in `verifyFile()` that MIME type and file extension correlate.
 406          * @param string $mime Representing the MIME
 407          * @return mixed True if the file is verified, an array otherwise
 408          */
 409         protected function verifyMimeType( $mime ) {
 410                 global $wgVerifyMimeType;
 411                 if ( $wgVerifyMimeType ) {
 412                         wfDebug( "mime: <$mime> extension: <{$this->mFinalExtension}>\n" );
 413                         global $wgMimeTypeBlacklist;
 414                         if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
 415                                 return [ 'filetype-badmime', $mime ];
 416                         }
 417
 418                         # Check what Internet Explorer would detect
 419                         $fp = fopen( $this->mTempPath, 'rb' );
 420                         $chunk = fread( $fp, 256 );
 421                         fclose( $fp );
 422
 423                         $magic = MimeMagic::singleton();
 424                         $extMime = $magic->guessTypesForExtension( $this->mFinalExtension );
 425                         $ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime );
 426                         foreach ( $ieTypes as $ieType ) {
 427                                 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
 428                                         return [ 'filetype-bad-ie-mime', $ieType ];
 429                                 }
 430                         }
 431                 }
 432
 433                 return true;
 434         }
 435
 436         /**
 437          * Verifies that it's ok to include the uploaded file
 438          *
 439          * @return mixed True of the file is verified, array otherwise.
 440          */
 441         protected function verifyFile() {
 442                 global $wgVerifyMimeType, $wgDisableUploadScriptChecks;
 443
 444                 $status = $this->verifyPartialFile();
 445                 if ( $status !== true ) {
 446                         return $status;
 447                 }
 448
 449                 $mwProps = new MWFileProps( MimeMagic::singleton() );
 450                 $this->mFileProps = $mwProps->getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 451                 $mime = $this->mFileProps['mime'];
 452
 453                 if ( $wgVerifyMimeType ) {
 454                         # XXX: Missing extension will be caught by validateName() via getTitle()
 455                         if ( $this->mFinalExtension != '' && !$this->verifyExtension( $mime, $this->mFinalExtension ) ) {
 456                                 return [ 'filetype-mime-mismatch', $this->mFinalExtension, $mime ];
 457                         }
 458                 }
 459
 460                 # check for htmlish code and javascript
 461                 if ( !$wgDisableUploadScriptChecks ) {
 462                         if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 463                                 $svgStatus = $this->detectScriptInSvg( $this->mTempPath, false );
 464                                 if ( $svgStatus !== false ) {
 465                                         return $svgStatus;
 466                                 }
 467                         }
 468                 }
 469
 470                 $handler = MediaHandler::getHandler( $mime );
 471                 if ( $handler ) {
 472                         $handlerStatus = $handler->verifyUpload( $this->mTempPath );
 473                         if ( !$handlerStatus->isOK() ) {
 474                                 $errors = $handlerStatus->getErrorsArray();
 475
 476                                 return reset( $errors );
 477                         }
 478                 }
 479
 480                 $error = true;
 481                 Hooks::run( 'UploadVerifyFile', [ $this, $mime, &$error ] );
 482                 if ( $error !== true ) {
 483                         if ( !is_array( $error ) ) {
 484                                 $error = [ $error ];
 485                         }
 486                         return $error;
 487                 }
 488
 489                 wfDebug( __METHOD__ . ": all clear; passing.\n" );
 490
 491                 return true;
 492         }
 493
 494         /**
 495          * A verification routine suitable for partial files
 496          *
 497          * Runs the blacklist checks, but not any checks that may
 498          * assume the entire file is present.
 499          *
 500          * @return mixed True for valid or array with error message key.
 501          */
 502         protected function verifyPartialFile() {
 503                 global $wgAllowJavaUploads, $wgDisableUploadScriptChecks;
 504
 505                 # getTitle() sets some internal parameters like $this->mFinalExtension
 506                 $this->getTitle();
 507
 508                 $mwProps = new MWFileProps( MimeMagic::singleton() );
 509                 $this->mFileProps = $mwProps->getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 510
 511                 # check MIME type, if desired
 512                 $mime = $this->mFileProps['file-mime'];
 513                 $status = $this->verifyMimeType( $mime );
 514                 if ( $status !== true ) {
 515                         return $status;
 516                 }
 517
 518                 # check for htmlish code and javascript
 519                 if ( !$wgDisableUploadScriptChecks ) {
 520                         if ( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
 521                                 return [ 'uploadscripted' ];
 522                         }
 523                         if ( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 524                                 $svgStatus = $this->detectScriptInSvg( $this->mTempPath, true );
 525                                 if ( $svgStatus !== false ) {
 526                                         return $svgStatus;
 527                                 }
 528                         }
 529                 }
 530
 531                 # Check for Java applets, which if uploaded can bypass cross-site
 532                 # restrictions.
 533                 if ( !$wgAllowJavaUploads ) {
 534                         $this->mJavaDetected = false;
 535                         $zipStatus = ZipDirectoryReader::read( $this->mTempPath,
 536                                 [ $this, 'zipEntryCallback' ] );
 537                         if ( !$zipStatus->isOK() ) {
 538                                 $errors = $zipStatus->getErrorsArray();
 539                                 $error = reset( $errors );
 540                                 if ( $error[0] !== 'zip-wrong-format' ) {
 541                                         return $error;
 542                                 }
 543                         }
 544                         if ( $this->mJavaDetected ) {
 545                                 return [ 'uploadjava' ];
 546                         }
 547                 }
 548
 549                 # Scan the uploaded file for viruses
 550                 $virus = $this->detectVirus( $this->mTempPath );
 551                 if ( $virus ) {
 552                         return [ 'uploadvirus', $virus ];
 553                 }
 554
 555                 return true;
 556         }
 557
 558         /**
 559          * Callback for ZipDirectoryReader to detect Java class files.
 560          *
 561          * @param array $entry
 562          */
 563         function zipEntryCallback( $entry ) {
 564                 $names = [ $entry['name'] ];
 565
 566                 // If there is a null character, cut off the name at it, because JDK's
 567                 // ZIP_GetEntry() uses strcmp() if the name hashes match. If a file name
 568                 // were constructed which had ".class\0" followed by a string chosen to
 569                 // make the hash collide with the truncated name, that file could be
 570                 // returned in response to a request for the .class file.
 571                 $nullPos = strpos( $entry['name'], "\000" );
 572                 if ( $nullPos !== false ) {
 573                         $names[] = substr( $entry['name'], 0, $nullPos );
 574                 }
 575
 576                 // If there is a trailing slash in the file name, we have to strip it,
 577                 // because that's what ZIP_GetEntry() does.
 578                 if ( preg_grep( '!\.class/?$!', $names ) ) {
 579                         $this->mJavaDetected = true;
 580                 }
 581         }
 582
 583         /**
 584          * Alias for verifyTitlePermissions. The function was originally
 585          * 'verifyPermissions', but that suggests it's checking the user, when it's
 586          * really checking the title + user combination.
 587          *
 588          * @param User $user User object to verify the permissions against
 589          * @return mixed An array as returned by getUserPermissionsErrors or true
 590          *   in case the user has proper permissions.
 591          */
 592         public function verifyPermissions( $user ) {
 593                 return $this->verifyTitlePermissions( $user );
 594         }
 595
 596         /**
 597          * Check whether the user can edit, upload and create the image. This
 598          * checks only against the current title; if it returns errors, it may
 599          * very well be that another title will not give errors. Therefore
 600          * isAllowed() should be called as well for generic is-user-blocked or
 601          * can-user-upload checking.
 602          *
 603          * @param User $user User object to verify the permissions against
 604          * @return mixed An array as returned by getUserPermissionsErrors or true
 605          *   in case the user has proper permissions.
 606          */
 607         public function verifyTitlePermissions( $user ) {
 608                 /**
 609                  * If the image is protected, non-sysop users won't be able
 610                  * to modify it by uploading a new revision.
 611                  */
 612                 $nt = $this->getTitle();
 613                 if ( is_null( $nt ) ) {
 614                         return true;
 615                 }
 616                 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
 617                 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
 618                 if ( !$nt->exists() ) {
 619                         $permErrorsCreate = $nt->getUserPermissionsErrors( 'create', $user );
 620                 } else {
 621                         $permErrorsCreate = [];
 622                 }
 623                 if ( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
 624                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
 625                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
 626
 627                         return $permErrors;
 628                 }
 629
 630                 $overwriteError = $this->checkOverwrite( $user );
 631                 if ( $overwriteError !== true ) {
 632                         return [ $overwriteError ];
 633                 }
 634
 635                 return true;
 636         }
 637
 638         /**
 639          * Check for non fatal problems with the file.
 640          *
 641          * This should not assume that mTempPath is set.
 642          *
 643          * @return array Array of warnings
 644          */
 645         public function checkWarnings() {
 646                 global $wgLang;
 647
 648                 $warnings = [];
 649
 650                 $localFile = $this->getLocalFile();
 651                 $localFile->load( File::READ_LATEST );
 652                 $filename = $localFile->getName();
 653
 654                 /**
 655                  * Check whether the resulting filename is different from the desired one,
 656                  * but ignore things like ucfirst() and spaces/underscore things
 657                  */
 658                 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName );
 659                 $comparableName = Title::capitalize( $comparableName, NS_FILE );
 660
 661                 if ( $this->mDesiredDestName != $filename && $comparableName != $filename ) {
 662                         $warnings['badfilename'] = $filename;
 663                 }
 664
 665                 // Check whether the file extension is on the unwanted list
 666                 global $wgCheckFileExtensions, $wgFileExtensions;
 667                 if ( $wgCheckFileExtensions ) {
 668                         $extensions = array_unique( $wgFileExtensions );
 669                         if ( !$this->checkFileExtension( $this->mFinalExtension, $extensions ) ) {
 670                                 $warnings['filetype-unwanted-type'] = [ $this->mFinalExtension,
 671                                         $wgLang->commaList( $extensions ), count( $extensions ) ];
 672                         }
 673                 }
 674
 675                 global $wgUploadSizeWarning;
 676                 if ( $wgUploadSizeWarning && ( $this->mFileSize > $wgUploadSizeWarning ) ) {
 677                         $warnings['large-file'] = [ $wgUploadSizeWarning, $this->mFileSize ];
 678                 }
 679
 680                 if ( $this->mFileSize == 0 ) {
 681                         $warnings['empty-file'] = true;
 682                 }
 683
 684                 $hash = $this->getTempFileSha1Base36();
 685                 $exists = self::getExistsWarning( $localFile );
 686                 if ( $exists !== false ) {
 687                         $warnings['exists'] = $exists;
 688
 689                         // check if file is an exact duplicate of current file version
 690                         if ( $hash === $localFile->getSha1() ) {
 691                                 $warnings['no-change'] = $localFile;
 692                         }
 693
 694                         // check if file is an exact duplicate of older versions of this file
 695                         $history = $localFile->getHistory();
 696                         foreach ( $history as $oldFile ) {
 697                                 if ( $hash === $oldFile->getSha1() ) {
 698                                         $warnings['duplicate-version'][] = $oldFile;
 699                                 }
 700                         }
 701                 }
 702
 703                 if ( $localFile->wasDeleted() && !$localFile->exists() ) {
 704                         $warnings['was-deleted'] = $filename;
 705                 }
 706
 707                 // Check dupes against existing files
 708                 $dupes = RepoGroup::singleton()->findBySha1( $hash );
 709                 $title = $this->getTitle();
 710                 // Remove all matches against self
 711                 foreach ( $dupes as $key => $dupe ) {
 712                         if ( $title->equals( $dupe->getTitle() ) ) {
 713                                 unset( $dupes[$key] );
 714                         }
 715                 }
 716                 if ( $dupes ) {
 717                         $warnings['duplicate'] = $dupes;
 718                 }
 719
 720                 // Check dupes against archives
 721                 $archivedFile = new ArchivedFile( null, 0, '', $hash );
 722                 if ( $archivedFile->getID() > 0 ) {
 723                         if ( $archivedFile->userCan( File::DELETED_FILE ) ) {
 724                                 $warnings['duplicate-archive'] = $archivedFile->getName();
 725                         } else {
 726                                 $warnings['duplicate-archive'] = '';
 727                         }
 728                 }
 729
 730                 return $warnings;
 731         }
 732
 733         /**
 734          * Really perform the upload. Stores the file in the local repo, watches
 735          * if necessary and runs the UploadComplete hook.
 736          *
 737          * @param string $comment
 738          * @param string $pageText
 739          * @param bool $watch Whether the file page should be added to user's watchlist.
 740          *   (This doesn't check $user's permissions.)
 741          * @param User $user
 742          * @param string[] $tags Change tags to add to the log entry and page revision.
 743          *   (This doesn't check $user's permissions.)
 744          * @return Status Indicating the whether the upload succeeded.
 745          */
 746         public function performUpload( $comment, $pageText, $watch, $user, $tags = [] ) {
 747                 $this->getLocalFile()->load( File::READ_LATEST );
 748                 $props = $this->mFileProps;
 749
 750                 $error = null;
 751                 Hooks::run( 'UploadVerifyUpload', [ $this, $user, $props, $comment, $pageText, &$error ] );
 752                 if ( $error ) {
 753                         if ( !is_array( $error ) ) {
 754                                 $error = [ $error ];
 755                         }
 756                         return call_user_func_array( 'Status::newFatal', $error );
 757                 }
 758
 759                 $status = $this->getLocalFile()->upload(
 760                         $this->mTempPath,
 761                         $comment,
 762                         $pageText,
 763                         File::DELETE_SOURCE,
 764                         $props,
 765                         false,
 766                         $user,
 767                         $tags
 768                 );
 769
 770                 if ( $status->isGood() ) {
 771                         if ( $watch ) {
 772                                 WatchAction::doWatch(
 773                                         $this->getLocalFile()->getTitle(),
 774                                         $user,
 775                                         User::IGNORE_USER_RIGHTS
 776                                 );
 777                         }
 778                         Hooks::run( 'UploadComplete', [ &$this ] );
 779
 780                         $this->postProcessUpload();
 781                 }
 782
 783                 return $status;
 784         }
 785
 786         /**
 787          * Perform extra steps after a successful upload.
 788          *
 789          * @since  1.25
 790          */
 791         public function postProcessUpload() {
 792         }
 793
 794         /**
 795          * Returns the title of the file to be uploaded. Sets mTitleError in case
 796          * the name was illegal.
 797          *
 798          * @return Title The title of the file or null in case the name was illegal
 799          */
 800         public function getTitle() {
 801                 if ( $this->mTitle !== false ) {
 802                         return $this->mTitle;
 803                 }
 804                 if ( !is_string( $this->mDesiredDestName ) ) {
 805                         $this->mTitleError = self::ILLEGAL_FILENAME;
 806                         $this->mTitle = null;
 807
 808                         return $this->mTitle;
 809                 }
 810                 /* Assume that if a user specified File:Something.jpg, this is an error
 811                  * and that the namespace prefix needs to be stripped of.
 812                  */
 813                 $title = Title::newFromText( $this->mDesiredDestName );
 814                 if ( $title && $title->getNamespace() == NS_FILE ) {
 815                         $this->mFilteredName = $title->getDBkey();
 816                 } else {
 817                         $this->mFilteredName = $this->mDesiredDestName;
 818                 }
 819
 820                 # oi_archive_name is max 255 bytes, which include a timestamp and an
 821                 # exclamation mark, so restrict file name to 240 bytes.
 822                 if ( strlen( $this->mFilteredName ) > 240 ) {
 823                         $this->mTitleError = self::FILENAME_TOO_LONG;
 824                         $this->mTitle = null;
 825
 826                         return $this->mTitle;
 827                 }
 828
 829                 /**
 830                  * Chop off any directories in the given filename. Then
 831                  * filter out illegal characters, and try to make a legible name
 832                  * out of it. We'll strip some silently that Title would die on.
 833                  */
 834                 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mFilteredName );
 835                 /* Normalize to title form before we do any further processing */
 836                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 837                 if ( is_null( $nt ) ) {
 838                         $this->mTitleError = self::ILLEGAL_FILENAME;
 839                         $this->mTitle = null;
 840
 841                         return $this->mTitle;
 842                 }
 843                 $this->mFilteredName = $nt->getDBkey();
 844
 845                 /**
 846                  * We'll want to blacklist against *any* 'extension', and use
 847                  * only the final one for the whitelist.
 848                  */
 849                 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
 850
 851                 if ( count( $ext ) ) {
 852                         $this->mFinalExtension = trim( $ext[count( $ext ) - 1] );
 853                 } else {
 854                         $this->mFinalExtension = '';
 855
 856                         # No extension, try guessing one
 857                         $magic = MimeMagic::singleton();
 858                         $mime = $magic->guessMimeType( $this->mTempPath );
 859                         if ( $mime !== 'unknown/unknown' ) {
 860                                 # Get a space separated list of extensions
 861                                 $extList = $magic->getExtensionsForType( $mime );
 862                                 if ( $extList ) {
 863                                         # Set the extension to the canonical extension
 864                                         $this->mFinalExtension = strtok( $extList, ' ' );
 865
 866                                         # Fix up the other variables
 867                                         $this->mFilteredName .= ".{$this->mFinalExtension}";
 868                                         $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 869                                         $ext = [ $this->mFinalExtension ];
 870                                 }
 871                         }
 872                 }
 873
 874                 /* Don't allow users to override the blacklist (check file extension) */
 875                 global $wgCheckFileExtensions, $wgStrictFileExtensions;
 876                 global $wgFileExtensions, $wgFileBlacklist;
 877
 878                 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
 879
 880                 if ( $this->mFinalExtension == '' ) {
 881                         $this->mTitleError = self::FILETYPE_MISSING;
 882                         $this->mTitle = null;
 883
 884                         return $this->mTitle;
 885                 } elseif ( $blackListedExtensions ||
 886                         ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
 887                                 !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) )
 888                 ) {
 889                         $this->mBlackListedExtensions = $blackListedExtensions;
 890                         $this->mTitleError = self::FILETYPE_BADTYPE;
 891                         $this->mTitle = null;
 892
 893                         return $this->mTitle;
 894                 }
 895
 896                 // Windows may be broken with special characters, see bug 1780
 897                 if ( !preg_match( '/^[\x0-\x7f]*$/', $nt->getText() )
 898                         && !RepoGroup::singleton()->getLocalRepo()->backendSupportsUnicodePaths()
 899                 ) {
 900                         $this->mTitleError = self::WINDOWS_NONASCII_FILENAME;
 901                         $this->mTitle = null;
 902
 903                         return $this->mTitle;
 904                 }
 905
 906                 # If there was more than one "extension", reassemble the base
 907                 # filename to prevent bogus complaints about length
 908                 if ( count( $ext ) > 1 ) {
 909                         $iterations = count( $ext ) - 1;
 910                         for ( $i = 0; $i < $iterations; $i++ ) {
 911                                 $partname .= '.' . $ext[$i];
 912                         }
 913                 }
 914
 915                 if ( strlen( $partname ) < 1 ) {
 916                         $this->mTitleError = self::MIN_LENGTH_PARTNAME;
 917                         $this->mTitle = null;
 918
 919                         return $this->mTitle;
 920                 }
 921
 922                 $this->mTitle = $nt;
 923
 924                 return $this->mTitle;
 925         }
 926
 927         /**
 928          * Return the local file and initializes if necessary.
 929          *
 930          * @return LocalFile|null
 931          */
 932         public function getLocalFile() {
 933                 if ( is_null( $this->mLocalFile ) ) {
 934                         $nt = $this->getTitle();
 935                         $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
 936                 }
 937
 938                 return $this->mLocalFile;
 939         }
 940
 941         /**
 942          * @return UploadStashFile|null
 943          */
 944         public function getStashFile() {
 945                 return $this->mStashFile;
 946         }
 947
 948         /**
 949          * Like stashFile(), but respects extensions' wishes to prevent the stashing. verifyUpload() must
 950          * be called before calling this method (unless $isPartial is true).
 951          *
 952          * Upload stash exceptions are also caught and converted to an error status.
 953          *
 954          * @since 1.28
 955          * @param User $user
 956          * @param bool $isPartial Pass `true` if this is a part of a chunked upload (not a complete file).
 957          * @return Status If successful, value is an UploadStashFile instance
 958          */
 959         public function tryStashFile( User $user, $isPartial = false ) {
 960                 if ( !$isPartial ) {
 961                         $error = $this->runUploadStashFileHook( $user );
 962                         if ( $error ) {
 963                                 return call_user_func_array( 'Status::newFatal', $error );
 964                         }
 965                 }
 966                 try {
 967                         $file = $this->doStashFile( $user );
 968                         return Status::newGood( $file );
 969                 } catch ( UploadStashException $e ) {
 970                         return Status::newFatal( 'uploadstash-exception', get_class( $e ), $e->getMessage() );
 971                 }
 972         }
 973
 974         /**
 975          * @param User $user
 976          * @return array|null Error message and parameters, null if there's no error
 977          */
 978         protected function runUploadStashFileHook( User $user ) {
 979                 $props = $this->mFileProps;
 980                 $error = null;
 981                 Hooks::run( 'UploadStashFile', [ $this, $user, $props, &$error ] );
 982                 if ( $error ) {
 983                         if ( !is_array( $error ) ) {
 984                                 $error = [ $error ];
 985                         }
 986                 }
 987                 return $error;
 988         }
 989
 990         /**
 991          * If the user does not supply all necessary information in the first upload
 992          * form submission (either by accident or by design) then we may want to
 993          * stash the file temporarily, get more information, and publish the file
 994          * later.
 995          *
 996          * This method will stash a file in a temporary directory for later
 997          * processing, and save the necessary descriptive info into the database.
 998          * This method returns the file object, which also has a 'fileKey' property
 999          * which can be passed through a form or API request to find this stashed
1000          * file again.
1001          *
1002          * @deprecated since 1.28 Use tryStashFile() instead
1003          * @param User $user
1004          * @return UploadStashFile Stashed file
1005          * @throws UploadStashBadPathException
1006          * @throws UploadStashFileException
1007          * @throws UploadStashNotLoggedInException
1008          */
1009         public function stashFile( User $user = null ) {
1010                 return $this->doStashFile( $user );
1011         }
1012
1013         /**
1014          * Implementation for stashFile() and tryStashFile().
1015          *
1016          * @param User $user
1017          * @return UploadStashFile Stashed file
1018          */
1019         protected function doStashFile( User $user = null ) {
1020                 $stash = RepoGroup::singleton()->getLocalRepo()->getUploadStash( $user );
1021                 $file = $stash->stashFile( $this->mTempPath, $this->getSourceType() );
1022                 $this->mStashFile = $file;
1023
1024                 return $file;
1025         }
1026
1027         /**
1028          * Stash a file in a temporary directory, returning a key which can be used
1029          * to find the file again. See stashFile().
1030          *
1031          * @deprecated since 1.28
1032          * @return string File key
1033          */
1034         public function stashFileGetKey() {
1035                 wfDeprecated( __METHOD__, '1.28' );
1036                 return $this->doStashFile()->getFileKey();
1037         }
1038
1039         /**
1040          * alias for stashFileGetKey, for backwards compatibility
1041          *
1042          * @deprecated since 1.28
1043          * @return string File key
1044          */
1045         public function stashSession() {
1046                 wfDeprecated( __METHOD__, '1.28' );
1047                 return $this->doStashFile()->getFileKey();
1048         }
1049
1050         /**
1051          * If we've modified the upload file we need to manually remove it
1052          * on exit to clean up.
1053          */
1054         public function cleanupTempFile() {
1055                 if ( $this->mRemoveTempFile && $this->tempFileObj ) {
1056                         // Delete when all relevant TempFSFile handles go out of scope
1057                         wfDebug( __METHOD__ . ": Marked temporary file '{$this->mTempPath}' for removal\n" );
1058                         $this->tempFileObj->autocollect();
1059                 }
1060         }
1061
1062         public function getTempPath() {
1063                 return $this->mTempPath;
1064         }
1065
1066         /**
1067          * Split a file into a base name and all dot-delimited 'extensions'
1068          * on the end. Some web server configurations will fall back to
1069          * earlier pseudo-'extensions' to determine type and execute
1070          * scripts, so the blacklist needs to check them all.
1071          *
1072          * @param string $filename
1073          * @return array
1074          */
1075         public static function splitExtensions( $filename ) {
1076                 $bits = explode( '.', $filename );
1077                 $basename = array_shift( $bits );
1078
1079                 return [ $basename, $bits ];
1080         }
1081
1082         /**
1083          * Perform case-insensitive match against a list of file extensions.
1084          * Returns true if the extension is in the list.
1085          *
1086          * @param string $ext
1087          * @param array $list
1088          * @return bool
1089          */
1090         public static function checkFileExtension( $ext, $list ) {
1091                 return in_array( strtolower( $ext ), $list );
1092         }
1093
1094         /**
1095          * Perform case-insensitive match against a list of file extensions.
1096          * Returns an array of matching extensions.
1097          *
1098          * @param array $ext
1099          * @param array $list
1100          * @return bool
1101          */
1102         public static function checkFileExtensionList( $ext, $list ) {
1103                 return array_intersect( array_map( 'strtolower', $ext ), $list );
1104         }
1105
1106         /**
1107          * Checks if the MIME type of the uploaded file matches the file extension.
1108          *
1109          * @param string $mime The MIME type of the uploaded file
1110          * @param string $extension The filename extension that the file is to be served with
1111          * @return bool
1112          */
1113         public static function verifyExtension( $mime, $extension ) {
1114                 $magic = MimeMagic::singleton();
1115
1116                 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' ) {
1117                         if ( !$magic->isRecognizableExtension( $extension ) ) {
1118                                 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
1119                                         "unrecognized extension '$extension', can't verify\n" );
1120
1121                                 return true;
1122                         } else {
1123                                 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; " .
1124                                         "recognized extension '$extension', so probably invalid file\n" );
1125
1126                                 return false;
1127                         }
1128                 }
1129
1130                 $match = $magic->isMatchingExtension( $extension, $mime );
1131
1132                 if ( $match === null ) {
1133                         if ( $magic->getTypesForExtension( $extension ) !== null ) {
1134                                 wfDebug( __METHOD__ . ": No extension known for $mime, but we know a mime for $extension\n" );
1135
1136                                 return false;
1137                         } else {
1138                                 wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
1139
1140                                 return true;
1141                         }
1142                 } elseif ( $match === true ) {
1143                         wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
1144
1145                         /** @todo If it's a bitmap, make sure PHP or ImageMagick resp. can handle it! */
1146                         return true;
1147                 } else {
1148                         wfDebug( __METHOD__
1149                                 . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
1150
1151                         return false;
1152                 }
1153         }
1154
1155         /**
1156          * Heuristic for detecting files that *could* contain JavaScript instructions or
1157          * things that may look like HTML to a browser and are thus
1158          * potentially harmful. The present implementation will produce false
1159          * positives in some situations.
1160          *
1161          * @param string $file Pathname to the temporary upload file
1162          * @param string $mime The MIME type of the file
1163          * @param string $extension The extension of the file
1164          * @return bool True if the file contains something looking like embedded scripts
1165          */
1166         public static function detectScript( $file, $mime, $extension ) {
1167                 global $wgAllowTitlesInSVG;
1168
1169                 # ugly hack: for text files, always look at the entire file.
1170                 # For binary field, just check the first K.
1171
1172                 if ( strpos( $mime, 'text/' ) === 0 ) {
1173                         $chunk = file_get_contents( $file );
1174                 } else {
1175                         $fp = fopen( $file, 'rb' );
1176                         $chunk = fread( $fp, 1024 );
1177                         fclose( $fp );
1178                 }
1179
1180                 $chunk = strtolower( $chunk );
1181
1182                 if ( !$chunk ) {
1183                         return false;
1184                 }
1185
1186                 # decode from UTF-16 if needed (could be used for obfuscation).
1187                 if ( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
1188                         $enc = 'UTF-16BE';
1189                 } elseif ( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
1190                         $enc = 'UTF-16LE';
1191                 } else {
1192                         $enc = null;
1193                 }
1194
1195                 if ( $enc ) {
1196                         $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
1197                 }
1198
1199                 $chunk = trim( $chunk );
1200
1201                 /** @todo FIXME: Convert from UTF-16 if necessary! */
1202                 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
1203
1204                 # check for HTML doctype
1205                 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
1206                         return true;
1207                 }
1208
1209                 // Some browsers will interpret obscure xml encodings as UTF-8, while
1210                 // PHP/expat will interpret the given encoding in the xml declaration (bug 47304)
1211                 if ( $extension == 'svg' || strpos( $mime, 'image/svg' ) === 0 ) {
1212                         if ( self::checkXMLEncodingMissmatch( $file ) ) {
1213                                 return true;
1214                         }
1215                 }
1216
1217                 /**
1218                  * Internet Explorer for Windows performs some really stupid file type
1219                  * autodetection which can cause it to interpret valid image files as HTML
1220                  * and potentially execute JavaScript, creating a cross-site scripting
1221                  * attack vectors.
1222                  *
1223                  * Apple's Safari browser also performs some unsafe file type autodetection
1224                  * which can cause legitimate files to be interpreted as HTML if the
1225                  * web server is not correctly configured to send the right content-type
1226                  * (or if you're really uploading plain text and octet streams!)
1227                  *
1228                  * Returns true if IE is likely to mistake the given file for HTML.
1229                  * Also returns true if Safari would mistake the given file for HTML
1230                  * when served with a generic content-type.
1231                  */
1232                 $tags = [
1233                         '<a href',
1234                         '<body',
1235                         '<head',
1236                         '<html', # also in safari
1237                         '<img',
1238                         '<pre',
1239                         '<script', # also in safari
1240                         '<table'
1241                 ];
1242
1243                 if ( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
1244                         $tags[] = '<title';
1245                 }
1246
1247                 foreach ( $tags as $tag ) {
1248                         if ( false !== strpos( $chunk, $tag ) ) {
1249                                 wfDebug( __METHOD__ . ": found something that may make it be mistaken for html: $tag\n" );
1250
1251                                 return true;
1252                         }
1253                 }
1254
1255                 /*
1256                  * look for JavaScript
1257                  */
1258
1259                 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
1260                 $chunk = Sanitizer::decodeCharReferences( $chunk );
1261
1262                 # look for script-types
1263                 if ( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
1264                         wfDebug( __METHOD__ . ": found script types\n" );
1265
1266                         return true;
1267                 }
1268
1269                 # look for html-style script-urls
1270                 if ( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1271                         wfDebug( __METHOD__ . ": found html-style script urls\n" );
1272
1273                         return true;
1274                 }
1275
1276                 # look for css-style script-urls
1277                 if ( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
1278                         wfDebug( __METHOD__ . ": found css-style script urls\n" );
1279
1280                         return true;
1281                 }
1282
1283                 wfDebug( __METHOD__ . ": no scripts found\n" );
1284
1285                 return false;
1286         }
1287
1288         /**
1289          * Check a whitelist of xml encodings that are known not to be interpreted differently
1290          * by the server's xml parser (expat) and some common browsers.
1291          *
1292          * @param string $file Pathname to the temporary upload file
1293          * @return bool True if the file contains an encoding that could be misinterpreted
1294          */
1295         public static function checkXMLEncodingMissmatch( $file ) {
1296                 global $wgSVGMetadataCutoff;
1297                 $contents = file_get_contents( $file, false, null, -1, $wgSVGMetadataCutoff );
1298                 $encodingRegex = '!encoding[ \t\n\r]*=[ \t\n\r]*[\'"](.*?)[\'"]!si';
1299
1300                 if ( preg_match( "!<\?xml\b(.*?)\?>!si", $contents, $matches ) ) {
1301                         if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1302                                 && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1303                         ) {
1304                                 wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1305
1306                                 return true;
1307                         }
1308                 } elseif ( preg_match( "!<\?xml\b!si", $contents ) ) {
1309                         // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1310                         // bytes. There shouldn't be a legitimate reason for this to happen.
1311                         wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1312
1313                         return true;
1314                 } elseif ( substr( $contents, 0, 4 ) == "\x4C\x6F\xA7\x94" ) {
1315                         // EBCDIC encoded XML
1316                         wfDebug( __METHOD__ . ": EBCDIC Encoded XML\n" );
1317
1318                         return true;
1319                 }
1320
1321                 // It's possible the file is encoded with multi-byte encoding, so re-encode attempt to
1322                 // detect the encoding in case is specifies an encoding not whitelisted in self::$safeXmlEncodings
1323                 $attemptEncodings = [ 'UTF-16', 'UTF-16BE', 'UTF-32', 'UTF-32BE' ];
1324                 foreach ( $attemptEncodings as $encoding ) {
1325                         MediaWiki\suppressWarnings();
1326                         $str = iconv( $encoding, 'UTF-8', $contents );
1327                         MediaWiki\restoreWarnings();
1328                         if ( $str != '' && preg_match( "!<\?xml\b(.*?)\?>!si", $str, $matches ) ) {
1329                                 if ( preg_match( $encodingRegex, $matches[1], $encMatch )
1330                                         && !in_array( strtoupper( $encMatch[1] ), self::$safeXmlEncodings )
1331                                 ) {
1332                                         wfDebug( __METHOD__ . ": Found unsafe XML encoding '{$encMatch[1]}'\n" );
1333
1334                                         return true;
1335                                 }
1336                         } elseif ( $str != '' && preg_match( "!<\?xml\b!si", $str ) ) {
1337                                 // Start of XML declaration without an end in the first $wgSVGMetadataCutoff
1338                                 // bytes. There shouldn't be a legitimate reason for this to happen.
1339                                 wfDebug( __METHOD__ . ": Unmatched XML declaration start\n" );
1340
1341                                 return true;
1342                         }
1343                 }
1344
1345                 return false;
1346         }
1347
1348         /**
1349          * @param string $filename
1350          * @param bool $partial
1351          * @return mixed False of the file is verified (does not contain scripts), array otherwise.
1352          */
1353         protected function detectScriptInSvg( $filename, $partial ) {
1354                 $this->mSVGNSError = false;
1355                 $check = new XmlTypeCheck(
1356                         $filename,
1357                         [ $this, 'checkSvgScriptCallback' ],
1358                         true,
1359                         [ 'processing_instruction_handler' => 'UploadBase::checkSvgPICallback' ]
1360                 );
1361                 if ( $check->wellFormed !== true ) {
1362                         // Invalid xml (bug 58553)
1363                         // But only when non-partial (bug 65724)
1364                         return $partial ? false : [ 'uploadinvalidxml' ];
1365                 } elseif ( $check->filterMatch ) {
1366                         if ( $this->mSVGNSError ) {
1367                                 return [ 'uploadscriptednamespace', $this->mSVGNSError ];
1368                         }
1369
1370                         return $check->filterMatchType;
1371                 }
1372
1373                 return false;
1374         }
1375
1376         /**
1377          * Callback to filter SVG Processing Instructions.
1378          * @param string $target Processing instruction name
1379          * @param string $data Processing instruction attribute and value
1380          * @return bool (true if the filter identified something bad)
1381          */
1382         public static function checkSvgPICallback( $target, $data ) {
1383                 // Don't allow external stylesheets (bug 57550)
1384                 if ( preg_match( '/xml-stylesheet/i', $target ) ) {
1385                         return [ 'upload-scripted-pi-callback' ];
1386                 }
1387
1388                 return false;
1389         }
1390
1391         /**
1392          * @todo Replace this with a whitelist filter!
1393          * @param string $element
1394          * @param array $attribs
1395          * @return bool
1396          */
1397         public function checkSvgScriptCallback( $element, $attribs, $data = null ) {
1398
1399                 list( $namespace, $strippedElement ) = $this->splitXmlNamespace( $element );
1400
1401                 // We specifically don't include:
1402                 // http://www.w3.org/1999/xhtml (bug 60771)
1403                 static $validNamespaces = [
1404                         '',
1405                         'adobe:ns:meta/',
1406                         'http://creativecommons.org/ns#',
1407                         'http://inkscape.sourceforge.net/dtd/sodipodi-0.dtd',
1408                         'http://ns.adobe.com/adobeillustrator/10.0/',
1409                         'http://ns.adobe.com/adobesvgviewerextensions/3.0/',
1410                         'http://ns.adobe.com/extensibility/1.0/',
1411                         'http://ns.adobe.com/flows/1.0/',
1412                         'http://ns.adobe.com/illustrator/1.0/',
1413                         'http://ns.adobe.com/imagereplacement/1.0/',
1414                         'http://ns.adobe.com/pdf/1.3/',
1415                         'http://ns.adobe.com/photoshop/1.0/',
1416                         'http://ns.adobe.com/saveforweb/1.0/',
1417                         'http://ns.adobe.com/variables/1.0/',
1418                         'http://ns.adobe.com/xap/1.0/',
1419                         'http://ns.adobe.com/xap/1.0/g/',
1420                         'http://ns.adobe.com/xap/1.0/g/img/',
1421                         'http://ns.adobe.com/xap/1.0/mm/',
1422                         'http://ns.adobe.com/xap/1.0/rights/',
1423                         'http://ns.adobe.com/xap/1.0/stype/dimensions#',
1424                         'http://ns.adobe.com/xap/1.0/stype/font#',
1425                         'http://ns.adobe.com/xap/1.0/stype/manifestitem#',
1426                         'http://ns.adobe.com/xap/1.0/stype/resourceevent#',
1427                         'http://ns.adobe.com/xap/1.0/stype/resourceref#',
1428                         'http://ns.adobe.com/xap/1.0/t/pg/',
1429                         'http://purl.org/dc/elements/1.1/',
1430                         'http://purl.org/dc/elements/1.1',
1431                         'http://schemas.microsoft.com/visio/2003/svgextensions/',
1432                         'http://sodipodi.sourceforge.net/dtd/sodipodi-0.dtd',
1433                         'http://taptrix.com/inkpad/svg_extensions',
1434                         'http://web.resource.org/cc/',
1435                         'http://www.freesoftware.fsf.org/bkchem/cdml',
1436                         'http://www.inkscape.org/namespaces/inkscape',
1437                         'http://www.opengis.net/gml',
1438                         'http://www.w3.org/1999/02/22-rdf-syntax-ns#',
1439                         'http://www.w3.org/2000/svg',
1440                         'http://www.w3.org/tr/rec-rdf-syntax/',
1441                 ];
1442
1443                 if ( !in_array( $namespace, $validNamespaces ) ) {
1444                         wfDebug( __METHOD__ . ": Non-svg namespace '$namespace' in uploaded file.\n" );
1445                         /** @todo Return a status object to a closure in XmlTypeCheck, for MW1.21+ */
1446                         $this->mSVGNSError = $namespace;
1447
1448                         return true;
1449                 }
1450
1451                 /*
1452                  * check for elements that can contain javascript
1453                  */
1454                 if ( $strippedElement == 'script' ) {
1455                         wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
1456
1457                         return [ 'uploaded-script-svg', $strippedElement ];
1458                 }
1459
1460                 # e.g., <svg xmlns="http://www.w3.org/2000/svg">
1461                 #  <handler xmlns:ev="http://www.w3.org/2001/xml-events" ev:event="load">alert(1)</handler> </svg>
1462                 if ( $strippedElement == 'handler' ) {
1463                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1464
1465                         return [ 'uploaded-script-svg', $strippedElement ];
1466                 }
1467
1468                 # SVG reported in Feb '12 that used xml:stylesheet to generate javascript block
1469                 if ( $strippedElement == 'stylesheet' ) {
1470                         wfDebug( __METHOD__ . ": Found scriptable element '$element' in uploaded file.\n" );
1471
1472                         return [ 'uploaded-script-svg', $strippedElement ];
1473                 }
1474
1475                 # Block iframes, in case they pass the namespace check
1476                 if ( $strippedElement == 'iframe' ) {
1477                         wfDebug( __METHOD__ . ": iframe in uploaded file.\n" );
1478
1479                         return [ 'uploaded-script-svg', $strippedElement ];
1480                 }
1481
1482                 # Check <style> css
1483                 if ( $strippedElement == 'style'
1484                         && self::checkCssFragment( Sanitizer::normalizeCss( $data ) )
1485                 ) {
1486                         wfDebug( __METHOD__ . ": hostile css in style element.\n" );
1487                         return [ 'uploaded-hostile-svg' ];
1488                 }
1489
1490                 foreach ( $attribs as $attrib => $value ) {
1491                         $stripped = $this->stripXmlNamespace( $attrib );
1492                         $value = strtolower( $value );
1493
1494                         if ( substr( $stripped, 0, 2 ) == 'on' ) {
1495                                 wfDebug( __METHOD__
1496                                         . ": Found event-handler attribute '$attrib'='$value' in uploaded file.\n" );
1497
1498                                 return [ 'uploaded-event-handler-on-svg', $attrib, $value ];
1499                         }
1500
1501                         # Do not allow relative links, or unsafe url schemas.
1502                         # For <a> tags, only data:, http: and https: and same-document
1503                         # fragment links are allowed. For all other tags, only data:
1504                         # and fragment are allowed.
1505                         if ( $stripped == 'href'
1506                                 && strpos( $value, 'data:' ) !== 0
1507                                 && strpos( $value, '#' ) !== 0
1508                         ) {
1509                                 if ( !( $strippedElement === 'a'
1510                                         && preg_match( '!^https?://!i', $value ) )
1511                                 ) {
1512                                         wfDebug( __METHOD__ . ": Found href attribute <$strippedElement "
1513                                                 . "'$attrib'='$value' in uploaded file.\n" );
1514
1515                                         return [ 'uploaded-href-attribute-svg', $strippedElement, $attrib, $value ];
1516                                 }
1517                         }
1518
1519                         # only allow data: targets that should be safe. This prevents vectors like,
1520                         # image/svg, text/xml, application/xml, and text/html, which can contain scripts
1521                         if ( $stripped == 'href' && strncasecmp( 'data:', $value, 5 ) === 0 ) {
1522                                 // rfc2397 parameters. This is only slightly slower than (;[\w;]+)*.
1523                                 // @codingStandardsIgnoreStart Generic.Files.LineLength
1524                                 $parameters = '(?>;[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+=(?>[a-zA-Z0-9\!#$&\'*+.^_`{|}~-]+|"(?>[\0-\x0c\x0e-\x21\x23-\x5b\x5d-\x7f]+|\\\\[\0-\x7f])*"))*(?:;base64)?';
1525                                 // @codingStandardsIgnoreEnd
1526
1527                                 if ( !preg_match( "!^data:\s*image/(gif|jpeg|jpg|png)$parameters,!i", $value ) ) {
1528                                         wfDebug( __METHOD__ . ": Found href to unwhitelisted data: uri "
1529                                                 . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1530                                         return [ 'uploaded-href-unsafe-target-svg', $strippedElement, $attrib, $value ];
1531                                 }
1532                         }
1533
1534                         # Change href with animate from (http://html5sec.org/#137).
1535                         if ( $stripped === 'attributename'
1536                                 && $strippedElement === 'animate'
1537                                 && $this->stripXmlNamespace( $value ) == 'href'
1538                         ) {
1539                                 wfDebug( __METHOD__ . ": Found animate that might be changing href using from "
1540                                         . "\"<$strippedElement '$attrib'='$value'...\" in uploaded file.\n" );
1541
1542                                 return [ 'uploaded-animate-svg', $strippedElement, $attrib, $value ];
1543                         }
1544
1545                         # use set/animate to add event-handler attribute to parent
1546                         if ( ( $strippedElement == 'set' || $strippedElement == 'animate' )
1547                                 && $stripped == 'attributename'
1548                                 && substr( $value, 0, 2 ) == 'on'
1549                         ) {
1550                                 wfDebug( __METHOD__ . ": Found svg setting event-handler attribute with "
1551                                         . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1552
1553                                 return [ 'uploaded-setting-event-handler-svg', $strippedElement, $stripped, $value ];
1554                         }
1555
1556                         # use set to add href attribute to parent element
1557                         if ( $strippedElement == 'set'
1558                                 && $stripped == 'attributename'
1559                                 && strpos( $value, 'href' ) !== false
1560                         ) {
1561                                 wfDebug( __METHOD__ . ": Found svg setting href attribute '$value' in uploaded file.\n" );
1562
1563                                 return [ 'uploaded-setting-href-svg' ];
1564                         }
1565
1566                         # use set to add a remote / data / script target to an element
1567                         if ( $strippedElement == 'set'
1568                                 && $stripped == 'to'
1569                                 && preg_match( '!(http|https|data|script):!sim', $value )
1570                         ) {
1571                                 wfDebug( __METHOD__ . ": Found svg setting attribute to '$value' in uploaded file.\n" );
1572
1573                                 return [ 'uploaded-wrong-setting-svg', $value ];
1574                         }
1575
1576                         # use handler attribute with remote / data / script
1577                         if ( $stripped == 'handler' && preg_match( '!(http|https|data|script):!sim', $value ) ) {
1578                                 wfDebug( __METHOD__ . ": Found svg setting handler with remote/data/script "
1579                                         . "'$attrib'='$value' in uploaded file.\n" );
1580
1581                                 return [ 'uploaded-setting-handler-svg', $attrib, $value ];
1582                         }
1583
1584                         # use CSS styles to bring in remote code
1585                         if ( $stripped == 'style'
1586                                 && self::checkCssFragment( Sanitizer::normalizeCss( $value ) )
1587                         ) {
1588                                 wfDebug( __METHOD__ . ": Found svg setting a style with "
1589                                         . "remote url '$attrib'='$value' in uploaded file.\n" );
1590                                 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1591                         }
1592
1593                         # Several attributes can include css, css character escaping isn't allowed
1594                         $cssAttrs = [ 'font', 'clip-path', 'fill', 'filter', 'marker',
1595                                 'marker-end', 'marker-mid', 'marker-start', 'mask', 'stroke' ];
1596                         if ( in_array( $stripped, $cssAttrs )
1597                                 && self::checkCssFragment( $value )
1598                         ) {
1599                                 wfDebug( __METHOD__ . ": Found svg setting a style with "
1600                                         . "remote url '$attrib'='$value' in uploaded file.\n" );
1601                                 return [ 'uploaded-remote-url-svg', $attrib, $value ];
1602                         }
1603
1604                         # image filters can pull in url, which could be svg that executes scripts
1605                         if ( $strippedElement == 'image'
1606                                 && $stripped == 'filter'
1607                                 && preg_match( '!url\s*\(!sim', $value )
1608                         ) {
1609                                 wfDebug( __METHOD__ . ": Found image filter with url: "
1610                                         . "\"<$strippedElement $stripped='$value'...\" in uploaded file.\n" );
1611
1612                                 return [ 'uploaded-image-filter-svg', $strippedElement, $stripped, $value ];
1613                         }
1614                 }
1615
1616                 return false; // No scripts detected
1617         }
1618
1619         /**
1620          * Check a block of CSS or CSS fragment for anything that looks like
1621          * it is bringing in remote code.
1622          * @param string $value a string of CSS
1623          * @param bool $propOnly only check css properties (start regex with :)
1624          * @return bool true if the CSS contains an illegal string, false if otherwise
1625          */
1626         private static function checkCssFragment( $value ) {
1627
1628                 # Forbid external stylesheets, for both reliability and to protect viewer's privacy
1629                 if ( stripos( $value, '@import' ) !== false ) {
1630                         return true;
1631                 }
1632
1633                 # We allow @font-face to embed fonts with data: urls, so we snip the string
1634                 # 'url' out so this case won't match when we check for urls below
1635                 $pattern = '!(@font-face\s*{[^}]*src:)url(\("data:;base64,)!im';
1636                 $value = preg_replace( $pattern, '$1$2', $value );
1637
1638                 # Check for remote and executable CSS. Unlike in Sanitizer::checkCss, the CSS
1639                 # properties filter and accelerator don't seem to be useful for xss in SVG files.
1640                 # Expression and -o-link don't seem to work either, but filtering them here in case.
1641                 # Additionally, we catch remote urls like url("http:..., url('http:..., url(http:...,
1642                 # but not local ones such as url("#..., url('#..., url(#....
1643                 if ( preg_match( '!expression
1644                                 | -o-link\s*:
1645                                 | -o-link-source\s*:
1646                                 | -o-replace\s*:!imx', $value ) ) {
1647                         return true;
1648                 }
1649
1650                 if ( preg_match_all(
1651                                 "!(\s*(url|image|image-set)\s*\(\s*[\"']?\s*[^#]+.*?\))!sim",
1652                                 $value,
1653                                 $matches
1654                         ) !== 0
1655                 ) {
1656                         # TODO: redo this in one regex. Until then, url("#whatever") matches the first
1657                         foreach ( $matches[1] as $match ) {
1658                                 if ( !preg_match( "!\s*(url|image|image-set)\s*\(\s*(#|'#|\"#)!im", $match ) ) {
1659                                         return true;
1660                                 }
1661                         }
1662                 }
1663
1664                 if ( preg_match( '/[\000-\010\013\016-\037\177]/', $value ) ) {
1665                         return true;
1666                 }
1667
1668                 return false;
1669         }
1670
1671         /**
1672          * Divide the element name passed by the xml parser to the callback into URI and prifix.
1673          * @param string $element
1674          * @return array Containing the namespace URI and prefix
1675          */
1676         private static function splitXmlNamespace( $element ) {
1677                 // 'http://www.w3.org/2000/svg:script' -> [ 'http://www.w3.org/2000/svg', 'script' ]
1678                 $parts = explode( ':', strtolower( $element ) );
1679                 $name = array_pop( $parts );
1680                 $ns = implode( ':', $parts );
1681
1682                 return [ $ns, $name ];
1683         }
1684
1685         /**
1686          * @param string $name
1687          * @return string
1688          */
1689         private function stripXmlNamespace( $name ) {
1690                 // 'http://www.w3.org/2000/svg:script' -> 'script'
1691                 $parts = explode( ':', strtolower( $name ) );
1692
1693                 return array_pop( $parts );
1694         }
1695
1696         /**
1697          * Generic wrapper function for a virus scanner program.
1698          * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
1699          * $wgAntivirusRequired may be used to deny upload if the scan fails.
1700          *
1701          * @param string $file Pathname to the temporary upload file
1702          * @return mixed False if not virus is found, null if the scan fails or is disabled,
1703          *   or a string containing feedback from the virus scanner if a virus was found.
1704          *   If textual feedback is missing but a virus was found, this function returns true.
1705          */
1706         public static function detectVirus( $file ) {
1707                 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
1708
1709                 if ( !$wgAntivirus ) {
1710                         wfDebug( __METHOD__ . ": virus scanner disabled\n" );
1711
1712                         return null;
1713                 }
1714
1715                 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
1716                         wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
1717                         $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
1718                                 [ 'virus-badscanner', $wgAntivirus ] );
1719
1720                         return wfMessage( 'virus-unknownscanner' )->text() . " $wgAntivirus";
1721                 }
1722
1723                 # look up scanner configuration
1724                 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
1725                 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
1726                 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
1727                         $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
1728
1729                 if ( strpos( $command, "%f" ) === false ) {
1730                         # simple pattern: append file to scan
1731                         $command .= " " . wfEscapeShellArg( $file );
1732                 } else {
1733                         # complex pattern: replace "%f" with file to scan
1734                         $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
1735                 }
1736
1737                 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
1738
1739                 # execute virus scanner
1740                 $exitCode = false;
1741
1742                 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
1743                 #      that does not seem to be worth the pain.
1744                 #      Ask me (Duesentrieb) about it if it's ever needed.
1745                 $output = wfShellExecWithStderr( $command, $exitCode );
1746
1747                 # map exit code to AV_xxx constants.
1748                 $mappedCode = $exitCode;
1749                 if ( $exitCodeMap ) {
1750                         if ( isset( $exitCodeMap[$exitCode] ) ) {
1751                                 $mappedCode = $exitCodeMap[$exitCode];
1752                         } elseif ( isset( $exitCodeMap["*"] ) ) {
1753                                 $mappedCode = $exitCodeMap["*"];
1754                         }
1755                 }
1756
1757                 /* NB: AV_NO_VIRUS is 0 but AV_SCAN_FAILED is false,
1758                  * so we need the strict equalities === and thus can't use a switch here
1759                  */
1760                 if ( $mappedCode === AV_SCAN_FAILED ) {
1761                         # scan failed (code was mapped to false by $exitCodeMap)
1762                         wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
1763
1764                         $output = $wgAntivirusRequired
1765                                 ? wfMessage( 'virus-scanfailed', [ $exitCode ] )->text()
1766                                 : null;
1767                 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
1768                         # scan failed because filetype is unknown (probably imune)
1769                         wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
1770                         $output = null;
1771                 } elseif ( $mappedCode === AV_NO_VIRUS ) {
1772                         # no virus found
1773                         wfDebug( __METHOD__ . ": file passed virus scan.\n" );
1774                         $output = false;
1775                 } else {
1776                         $output = trim( $output );
1777
1778                         if ( !$output ) {
1779                                 $output = true; # if there's no output, return true
1780                         } elseif ( $msgPattern ) {
1781                                 $groups = [];
1782                                 if ( preg_match( $msgPattern, $output, $groups ) ) {
1783                                         if ( $groups[1] ) {
1784                                                 $output = $groups[1];
1785                                         }
1786                                 }
1787                         }
1788
1789                         wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
1790                 }
1791
1792                 return $output;
1793         }
1794
1795         /**
1796          * Check if there's an overwrite conflict and, if so, if restrictions
1797          * forbid this user from performing the upload.
1798          *
1799          * @param User $user
1800          *
1801          * @return mixed True on success, array on failure
1802          */
1803         private function checkOverwrite( $user ) {
1804                 // First check whether the local file can be overwritten
1805                 $file = $this->getLocalFile();
1806                 $file->load( File::READ_LATEST );
1807                 if ( $file->exists() ) {
1808                         if ( !self::userCanReUpload( $user, $file ) ) {
1809                                 return [ 'fileexists-forbidden', $file->getName() ];
1810                         } else {
1811                                 return true;
1812                         }
1813                 }
1814
1815                 /* Check shared conflicts: if the local file does not exist, but
1816                  * wfFindFile finds a file, it exists in a shared repository.
1817                  */
1818                 $file = wfFindFile( $this->getTitle(), [ 'latest' => true ] );
1819                 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1820                         return [ 'fileexists-shared-forbidden', $file->getName() ];
1821                 }
1822
1823                 return true;
1824         }
1825
1826         /**
1827          * Check if a user is the last uploader
1828          *
1829          * @param User $user
1830          * @param File $img
1831          * @return bool
1832          */
1833         public static function userCanReUpload( User $user, File $img ) {
1834                 if ( $user->isAllowed( 'reupload' ) ) {
1835                         return true; // non-conditional
1836                 } elseif ( !$user->isAllowed( 'reupload-own' ) ) {
1837                         return false;
1838                 }
1839
1840                 if ( !( $img instanceof LocalFile ) ) {
1841                         return false;
1842                 }
1843
1844                 $img->load();
1845
1846                 return $user->getId() == $img->getUser( 'id' );
1847         }
1848
1849         /**
1850          * Helper function that does various existence checks for a file.
1851          * The following checks are performed:
1852          * - The file exists
1853          * - Article with the same name as the file exists
1854          * - File exists with normalized extension
1855          * - The file looks like a thumbnail and the original exists
1856          *
1857          * @param File $file The File object to check
1858          * @return mixed False if the file does not exists, else an array
1859          */
1860         public static function getExistsWarning( $file ) {
1861                 if ( $file->exists() ) {
1862                         return [ 'warning' => 'exists', 'file' => $file ];
1863                 }
1864
1865                 if ( $file->getTitle()->getArticleID() ) {
1866                         return [ 'warning' => 'page-exists', 'file' => $file ];
1867                 }
1868
1869                 if ( strpos( $file->getName(), '.' ) == false ) {
1870                         $partname = $file->getName();
1871                         $extension = '';
1872                 } else {
1873                         $n = strrpos( $file->getName(), '.' );
1874                         $extension = substr( $file->getName(), $n + 1 );
1875                         $partname = substr( $file->getName(), 0, $n );
1876                 }
1877                 $normalizedExtension = File::normalizeExtension( $extension );
1878
1879                 if ( $normalizedExtension != $extension ) {
1880                         // We're not using the normalized form of the extension.
1881                         // Normal form is lowercase, using most common of alternate
1882                         // extensions (eg 'jpg' rather than 'JPEG').
1883
1884                         // Check for another file using the normalized form...
1885                         $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
1886                         $file_lc = wfLocalFile( $nt_lc );
1887
1888                         if ( $file_lc->exists() ) {
1889                                 return [
1890                                         'warning' => 'exists-normalized',
1891                                         'file' => $file,
1892                                         'normalizedFile' => $file_lc
1893                                 ];
1894                         }
1895                 }
1896
1897                 // Check for files with the same name but a different extension
1898                 $similarFiles = RepoGroup::singleton()->getLocalRepo()->findFilesByPrefix(
1899                         "{$partname}.", 1 );
1900                 if ( count( $similarFiles ) ) {
1901                         return [
1902                                 'warning' => 'exists-normalized',
1903                                 'file' => $file,
1904                                 'normalizedFile' => $similarFiles[0],
1905                         ];
1906                 }
1907
1908                 if ( self::isThumbName( $file->getName() ) ) {
1909                         # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1910                         $nt_thb = Title::newFromText(
1911                                 substr( $partname, strpos( $partname, '-' ) + 1 ) . '.' . $extension,
1912                                 NS_FILE
1913                         );
1914                         $file_thb = wfLocalFile( $nt_thb );
1915                         if ( $file_thb->exists() ) {
1916                                 return [
1917                                         'warning' => 'thumb',
1918                                         'file' => $file,
1919                                         'thumbFile' => $file_thb
1920                                 ];
1921                         } else {
1922                                 // File does not exist, but we just don't like the name
1923                                 return [
1924                                         'warning' => 'thumb-name',
1925                                         'file' => $file,
1926                                         'thumbFile' => $file_thb
1927                                 ];
1928                         }
1929                 }
1930
1931                 foreach ( self::getFilenamePrefixBlacklist() as $prefix ) {
1932                         if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1933                                 return [
1934                                         'warning' => 'bad-prefix',
1935                                         'file' => $file,
1936                                         'prefix' => $prefix
1937                                 ];
1938                         }
1939                 }
1940
1941                 return false;
1942         }
1943
1944         /**
1945          * Helper function that checks whether the filename looks like a thumbnail
1946          * @param string $filename
1947          * @return bool
1948          */
1949         public static function isThumbName( $filename ) {
1950                 $n = strrpos( $filename, '.' );
1951                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
1952
1953                 return (
1954                         substr( $partname, 3, 3 ) == 'px-' ||
1955                         substr( $partname, 2, 3 ) == 'px-'
1956                 ) &&
1957                 preg_match( "/[0-9]{2}/", substr( $partname, 0, 2 ) );
1958         }
1959
1960         /**
1961          * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
1962          *
1963          * @return array List of prefixes
1964          */
1965         public static function getFilenamePrefixBlacklist() {
1966                 $blacklist = [];
1967                 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
1968                 if ( !$message->isDisabled() ) {
1969                         $lines = explode( "\n", $message->plain() );
1970                         foreach ( $lines as $line ) {
1971                                 // Remove comment lines
1972                                 $comment = substr( trim( $line ), 0, 1 );
1973                                 if ( $comment == '#' || $comment == '' ) {
1974                                         continue;
1975                                 }
1976                                 // Remove additional comments after a prefix
1977                                 $comment = strpos( $line, '#' );
1978                                 if ( $comment > 0 ) {
1979                                         $line = substr( $line, 0, $comment - 1 );
1980                                 }
1981                                 $blacklist[] = trim( $line );
1982                         }
1983                 }
1984
1985                 return $blacklist;
1986         }
1987
1988         /**
1989          * Gets image info about the file just uploaded.
1990          *
1991          * Also has the effect of setting metadata to be an 'indexed tag name' in
1992          * returned API result if 'metadata' was requested. Oddly, we have to pass
1993          * the "result" object down just so it can do that with the appropriate
1994          * format, presumably.
1995          *
1996          * @param ApiResult $result
1997          * @return array Image info
1998          */
1999         public function getImageInfo( $result ) {
2000                 $localFile = $this->getLocalFile();
2001                 $stashFile = $this->getStashFile();
2002                 // Calling a different API module depending on whether the file was stashed is less than optimal.
2003                 // In fact, calling API modules here at all is less than optimal. Maybe it should be refactored.
2004                 if ( $stashFile ) {
2005                         $imParam = ApiQueryStashImageInfo::getPropertyNames();
2006                         $info = ApiQueryStashImageInfo::getInfo( $stashFile, array_flip( $imParam ), $result );
2007                 } else {
2008                         $imParam = ApiQueryImageInfo::getPropertyNames();
2009                         $info = ApiQueryImageInfo::getInfo( $localFile, array_flip( $imParam ), $result );
2010                 }
2011
2012                 return $info;
2013         }
2014
2015         /**
2016          * @param array $error
2017          * @return Status
2018          */
2019         public function convertVerifyErrorToStatus( $error ) {
2020                 $code = $error['status'];
2021                 unset( $code['status'] );
2022
2023                 return Status::newFatal( $this->getVerificationErrorCode( $code ), $error );
2024         }
2025
2026         /**
2027          * Get the MediaWiki maximum uploaded file size for given type of upload, based on
2028          * $wgMaxUploadSize.
2029          *
2030          * @param null|string $forType
2031          * @return int
2032          */
2033         public static function getMaxUploadSize( $forType = null ) {
2034                 global $wgMaxUploadSize;
2035
2036                 if ( is_array( $wgMaxUploadSize ) ) {
2037                         if ( !is_null( $forType ) && isset( $wgMaxUploadSize[$forType] ) ) {
2038                                 return $wgMaxUploadSize[$forType];
2039                         } else {
2040                                 return $wgMaxUploadSize['*'];
2041                         }
2042                 } else {
2043                         return intval( $wgMaxUploadSize );
2044                 }
2045         }
2046
2047         /**
2048          * Get the PHP maximum uploaded file size, based on ini settings. If there is no limit or the
2049          * limit can't be guessed, returns a very large number (PHP_INT_MAX).
2050          *
2051          * @since 1.27
2052          * @return int
2053          */
2054         public static function getMaxPhpUploadSize() {
2055                 $phpMaxFileSize = wfShorthandToInteger(
2056                         ini_get( 'upload_max_filesize' ) ?: ini_get( 'hhvm.server.upload.upload_max_file_size' ),
2057                         PHP_INT_MAX
2058                 );
2059                 $phpMaxPostSize = wfShorthandToInteger(
2060                         ini_get( 'post_max_size' ) ?: ini_get( 'hhvm.server.max_post_size' ),
2061                         PHP_INT_MAX
2062                 ) ?: PHP_INT_MAX;
2063                 return min( $phpMaxFileSize, $phpMaxPostSize );
2064         }
2065
2066         /**
2067          * Get the current status of a chunked upload (used for polling)
2068          *
2069          * The value will be read from cache.
2070          *
2071          * @param User $user
2072          * @param string $statusKey
2073          * @return Status[]|bool
2074          */
2075         public static function getSessionStatus( User $user, $statusKey ) {
2076                 $key = wfMemcKey( 'uploadstatus', $user->getId() ?: md5( $user->getName() ), $statusKey );
2077
2078                 return ObjectCache::getMainStashInstance()->get( $key );
2079         }
2080
2081         /**
2082          * Set the current status of a chunked upload (used for polling)
2083          *
2084          * The value will be set in cache for 1 day
2085          *
2086          * @param User $user
2087          * @param string $statusKey
2088          * @param array|bool $value
2089          * @return void
2090          */
2091         public static function setSessionStatus( User $user, $statusKey, $value ) {
2092                 $key = wfMemcKey( 'uploadstatus', $user->getId() ?: md5( $user->getName() ), $statusKey );
2093
2094                 $cache = ObjectCache::getMainStashInstance();
2095                 if ( $value === false ) {
2096                         $cache->delete( $key );
2097                 } else {
2098                         $cache->set( $key, $value, $cache::TTL_DAY );
2099                 }
2100         }
2101 }