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