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