includes/upload/UploadBase.php

   1 <?php
   2 /**
   3  * @file
   4  * @ingroup upload
   5  *
   6  * UploadBase and subclasses are the backend of MediaWiki's file uploads.
   7  * The frontends are formed by ApiUpload and SpecialUpload.
   8  *
   9  * See also includes/docs/upload.txt
  10  *
  11  * @author Brion Vibber
  12  * @author Bryan Tong Minh
  13  * @author Michael Dale
  14  */
  15
  16 abstract class UploadBase {
  17         protected $mTempPath;
  18         protected $mDesiredDestName, $mDestName, $mRemoveTempFile, $mSourceType;
  19         protected $mTitle = false, $mTitleError = 0;
  20         protected $mFilteredName, $mFinalExtension;
  21         protected $mLocalFile;
  22
  23         const SUCCESS = 0;
  24         const OK = 0;
  25         const EMPTY_FILE = 3;
  26         const MIN_LENGTH_PARTNAME = 4;
  27         const ILLEGAL_FILENAME = 5;
  28         const OVERWRITE_EXISTING_FILE = 7; # Not used anymore; handled by verifyPermissions()
  29         const FILETYPE_MISSING = 8;
  30         const FILETYPE_BADTYPE = 9;
  31         const VERIFICATION_ERROR = 10;
  32
  33         # HOOK_ABORTED is the new name of UPLOAD_VERIFICATION_ERROR
  34         const UPLOAD_VERIFICATION_ERROR = 11;
  35         const HOOK_ABORTED = 11;
  36         const FILE_TOO_LARGE = 12;
  37
  38         const SESSION_VERSION = 2;
  39         const SESSION_KEYNAME = 'wsUploadData';
  40
  41         static public function getSessionKeyname() {
  42                 return self::SESSION_KEYNAME;
  43         }
  44
  45         public function getVerificationErrorCode( $error ) {
  46                 $code_to_status = array(self::EMPTY_FILE => 'empty-file',
  47                                                                 self::FILE_TOO_LARGE => 'file-too-large',
  48                                                                 self::FILETYPE_MISSING => 'filetype-missing',
  49                                                                 self::FILETYPE_BADTYPE => 'filetype-banned',
  50                                                                 self::MIN_LENGTH_PARTNAME => 'filename-tooshort',
  51                                                                 self::ILLEGAL_FILENAME => 'illegal-filename',
  52                                                                 self::OVERWRITE_EXISTING_FILE => 'overwrite',
  53                                                                 self::VERIFICATION_ERROR => 'verification-error',
  54                                                                 self::HOOK_ABORTED =>  'hookaborted',
  55                 );
  56                 if( isset( $code_to_status[$error] ) ) {
  57                         return $code_to_status[$error];
  58                 }
  59
  60                 return 'unknown-error';
  61         }
  62
  63         /**
  64          * Returns true if uploads are enabled.
  65          * Can be override by subclasses.
  66          */
  67         public static function isEnabled() {
  68                 global $wgEnableUploads;
  69                 if ( !$wgEnableUploads ) {
  70                         return false;
  71                 }
  72
  73                 # Check php's file_uploads setting
  74                 if( !wfIniGetBool( 'file_uploads' ) ) {
  75                         return false;
  76                 }
  77                 return true;
  78         }
  79
  80         /**
  81          * Returns an array of permissions that is required to upload a file
  82          *
  83          * @return array
  84          */
  85         public static function getRequiredPermissions() {
  86                 return array( 'upload', 'edit' );
  87         }
  88         /**
  89          * Returns true if the user can use this upload module or else a string
  90          * identifying the missing permission.
  91          * Can be overriden by subclasses.
  92          */
  93         public static function isAllowed( $user ) {
  94                 foreach ( self::getRequiredPermissions() as $permission ) {
  95                         if ( !$user->isAllowed( $permission ) ) {
  96                                 return $permission;
  97                         }
  98                 }
  99                 return true;
 100         }
 101
 102         // Upload handlers. Should probably just be a global.
 103         static $uploadHandlers = array( 'Stash', 'File', 'Url' );
 104
 105         /**
 106          * Create a form of UploadBase depending on wpSourceType and initializes it
 107          */
 108         public static function createFromRequest( &$request, $type = null ) {
 109                 $type = $type ? $type : $request->getVal( 'wpSourceType', 'File' );
 110
 111                 if( !$type ) {
 112                         return null;
 113                 }
 114
 115                 // Get the upload class
 116                 $type = ucfirst( $type );
 117
 118                 // Give hooks the chance to handle this request
 119                 $className = null;
 120                 wfRunHooks( 'UploadCreateFromRequest', array( $type, &$className ) );
 121                 if ( is_null( $className ) ) {
 122                         $className = 'UploadFrom' . $type;
 123                         wfDebug( __METHOD__ . ": class name: $className\n" );
 124                         if( !in_array( $type, self::$uploadHandlers ) ) {
 125                                 return null;
 126                         }
 127                 }
 128
 129                 // Check whether this upload class is enabled
 130                 if( !call_user_func( array( $className, 'isEnabled' ) ) ) {
 131                         return null;
 132                 }
 133
 134                 // Check whether the request is valid
 135                 if( !call_user_func( array( $className, 'isValidRequest' ), $request ) ) {
 136                         return null;
 137                 }
 138
 139                 $handler = new $className;
 140
 141                 $handler->initializeFromRequest( $request );
 142                 return $handler;
 143         }
 144
 145         /**
 146          * Check whether a request if valid for this handler
 147          */
 148         public static function isValidRequest( $request ) {
 149                 return false;
 150         }
 151
 152         public function __construct() {}
 153
 154         /**
 155          * Initialize the path information
 156          * @param $name string the desired destination name
 157          * @param $tempPath string the temporary path
 158          * @param $fileSize int the file size
 159          * @param $removeTempFile bool (false) remove the temporary file?
 160          * @return null
 161          */
 162         public function initializePathInfo( $name, $tempPath, $fileSize, $removeTempFile = false ) {
 163                 $this->mDesiredDestName = $name;
 164                 $this->mTempPath = $tempPath;
 165                 $this->mFileSize = $fileSize;
 166                 $this->mRemoveTempFile = $removeTempFile;
 167         }
 168
 169         /**
 170          * Initialize from a WebRequest. Override this in a subclass.
 171          */
 172         public abstract function initializeFromRequest( &$request );
 173
 174         /**
 175          * Fetch the file. Usually a no-op
 176          */
 177         public function fetchFile() {
 178                 return Status::newGood();
 179         }
 180
 181         /**
 182          * Return true if the file is empty
 183          * @return bool
 184          */
 185         public function isEmptyFile() {
 186                 return empty( $this->mFileSize );
 187         }
 188
 189         /**
 190          * Return the file size
 191          * @return integer
 192          */
 193         public function getFileSize() {
 194                 return $this->mFileSize;
 195         }
 196
 197         /**
 198          * Append a file to the Repo file
 199          *
 200          * @param $srcPath String: path to source file
 201          * @param $toAppendPath String: path to the Repo file that will be appended to.
 202          * @return Status Status
 203          */
 204         protected function appendToUploadFile( $srcPath, $toAppendPath ) {
 205                 $repo = RepoGroup::singleton()->getLocalRepo();
 206                 $status = $repo->append( $srcPath, $toAppendPath );
 207                 return $status;
 208         }
 209
 210         /**
 211          * @param $srcPath String: the source path
 212          * @return the real path if it was a virtual URL
 213          */
 214         function getRealPath( $srcPath ) {
 215                 $repo = RepoGroup::singleton()->getLocalRepo();
 216                 if ( $repo->isVirtualUrl( $srcPath ) ) {
 217                         return $repo->resolveVirtualUrl( $srcPath );
 218                 }
 219                 return $srcPath;
 220         }
 221
 222         /**
 223          * Verify whether the upload is sane.
 224          * @return mixed self::OK or else an array with error information
 225          */
 226         public function verifyUpload() {
 227                 /**
 228                  * If there was no filename or a zero size given, give up quick.
 229                  */
 230                 if( $this->isEmptyFile() ) {
 231                         return array( 'status' => self::EMPTY_FILE );
 232                 }
 233
 234                 /**
 235                  * Honor $wgMaxUploadSize
 236                  */
 237                 global $wgMaxUploadSize;
 238                 if( $this->mFileSize > $wgMaxUploadSize ) {
 239                         return array(
 240                                 'status' => self::FILE_TOO_LARGE,
 241                                 'max' => $wgMaxUploadSize,
 242                         );
 243                 }
 244
 245                 /**
 246                  * Look at the contents of the file; if we can recognize the
 247                  * type but it's corrupt or data of the wrong type, we should
 248                  * probably not accept it.
 249                  */
 250                 $verification = $this->verifyFile();
 251                 if( $verification !== true ) {
 252                         return array(
 253                                 'status' => self::VERIFICATION_ERROR,
 254                                 'details' => $verification
 255                         );
 256                 }
 257
 258                 /**
 259                  * Make sure this file can be created
 260                  */
 261                 $result = $this->validateName();
 262                 if( $result !== true ) {
 263                         return $result;
 264                 }
 265
 266                 $error = '';
 267                 if( !wfRunHooks( 'UploadVerification',
 268                                 array( $this->mDestName, $this->mTempPath, &$error ) ) ) {
 269                         return array( 'status' => self::HOOK_ABORTED, 'error' => $error );
 270                 }
 271
 272                 return array( 'status' => self::OK );
 273         }
 274
 275         /**
 276          * Verify that the name is valid and, if necessary, that we can overwrite
 277          *
 278          * @return mixed true if valid, otherwise and array with 'status'
 279          * and other keys
 280          **/
 281         protected function validateName() {
 282                 $nt = $this->getTitle();
 283                 if( is_null( $nt ) ) {
 284                         $result = array( 'status' => $this->mTitleError );
 285                         if( $this->mTitleError == self::ILLEGAL_FILENAME ) {
 286                                 $result['filtered'] = $this->mFilteredName;
 287                         }
 288                         if ( $this->mTitleError == self::FILETYPE_BADTYPE ) {
 289                                 $result['finalExt'] = $this->mFinalExtension;
 290                         }
 291                         return $result;
 292                 }
 293                 $this->mDestName = $this->getLocalFile()->getName();
 294
 295                 return true;
 296         }
 297
 298         /**
 299          * Verify the mime type
 300          *
 301          * @param $mime string representing the mime
 302          * @return mixed true if the file is verified, an array otherwise
 303          */
 304         protected function verifyMimeType( $mime ) {
 305                 global $wgVerifyMimeType;
 306                 if ( $wgVerifyMimeType ) {
 307                         wfDebug ( "\n\nmime: <$mime> extension: <{$this->mFinalExtension}>\n\n");
 308                         if ( !$this->verifyExtension( $mime, $this->mFinalExtension ) ) {
 309                                 return array( 'filetype-mime-mismatch' );
 310                         }
 311
 312                         global $wgMimeTypeBlacklist;
 313                         if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
 314                                 return array( 'filetype-badmime', $mime );
 315                         }
 316
 317                         # Check IE type
 318                         $fp = fopen( $this->mTempPath, 'rb' );
 319                         $chunk = fread( $fp, 256 );
 320                         fclose( $fp );
 321
 322                         $magic = MimeMagic::singleton();
 323                         $extMime = $magic->guessTypesForExtension( $this->mFinalExtension );
 324                         $ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime );
 325                         foreach ( $ieTypes as $ieType ) {
 326                                 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
 327                                         return array( 'filetype-bad-ie-mime', $ieType );
 328                                 }
 329                         }
 330                 }
 331
 332                 return true;
 333         }
 334
 335         /**
 336          * Verifies that it's ok to include the uploaded file
 337          *
 338          * @return mixed true of the file is verified, array otherwise.
 339          */
 340         protected function verifyFile() {
 341                 # get the title, even though we are doing nothing with it, because
 342                 # we need to populate mFinalExtension
 343                 $this->getTitle();
 344
 345                 $this->mFileProps = File::getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 346                 $this->checkMacBinary();
 347
 348                 # check mime type, if desired
 349                 $mime = $this->mFileProps[ 'file-mime' ];
 350                 $status = $this->verifyMimeType( $mime );
 351                 if ( $status !== true ) {
 352                         return $status;
 353                 }
 354
 355                 # check for htmlish code and javascript
 356                 if( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
 357                         return array( 'uploadscripted' );
 358                 }
 359                 if( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 360                         if( $this->detectScriptInSvg( $this->mTempPath ) ) {
 361                                 return array( 'uploadscripted' );
 362                         }
 363                 }
 364
 365                 /**
 366                 * Scan the uploaded file for viruses
 367                 */
 368                 $virus = $this->detectVirus( $this->mTempPath );
 369                 if ( $virus ) {
 370                         return array( 'uploadvirus', $virus );
 371                 }
 372
 373                 $handler = MediaHandler::getHandler( $mime );
 374                 if ( $handler ) {
 375                         $handlerStatus = $handler->verifyUpload( $this->mTempPath );
 376                         if ( !$handlerStatus->isOK() ) {
 377                                 $errors = $handlerStatus->getErrorsArray();
 378                                 return reset( $errors );
 379                         }
 380                 }
 381
 382                 wfRunHooks( 'UploadVerifyFile', array( $this, $mime, &$status ) );
 383                 if ( $status !== true ) {
 384                         return $status;
 385                 }
 386
 387                 wfDebug( __METHOD__ . ": all clear; passing.\n" );
 388                 return true;
 389         }
 390
 391         /**
 392          * Check whether the user can edit, upload and create the image. This
 393          * checks only against the current title; if it returns errors, it may
 394          * very well be that another title will not give errors. Therefore
 395          * isAllowed() should be called as well for generic is-user-blocked or
 396          * can-user-upload checking.
 397          *
 398          * @param $user the User object to verify the permissions against
 399          * @return mixed An array as returned by getUserPermissionsErrors or true
 400          *               in case the user has proper permissions.
 401          */
 402         public function verifyPermissions( $user ) {
 403                 /**
 404                  * If the image is protected, non-sysop users won't be able
 405                  * to modify it by uploading a new revision.
 406                  */
 407                 $nt = $this->getTitle();
 408                 if( is_null( $nt ) ) {
 409                         return true;
 410                 }
 411                 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
 412                 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
 413                 if ( $nt->exists() ) {
 414                         $permErrorsCreate = $nt->getUserPermissionsErrors( 'createpage', $user );
 415                 } else {
 416                         $permErrorsCreate = array();
 417                 }
 418                 if( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
 419                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
 420                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
 421                         return $permErrors;
 422                 }
 423
 424                 $overwriteError = $this->checkOverwrite( $user );
 425                 if ( $overwriteError !== true ) {
 426                         return array( array( $overwriteError ) );
 427                 }
 428
 429                 return true;
 430         }
 431
 432         /**
 433          * Check for non fatal problems with the file
 434          *
 435          * @return Array of warnings
 436          */
 437         public function checkWarnings() {
 438                 $warnings = array();
 439
 440                 $localFile = $this->getLocalFile();
 441                 $filename = $localFile->getName();
 442                 $n = strrpos( $filename, '.' );
 443
 444                 /**
 445                  * Check whether the resulting filename is different from the desired one,
 446                  * but ignore things like ucfirst() and spaces/underscore things
 447                  */
 448                 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName );
 449                 $comparableName = Title::capitalize( $comparableName, NS_FILE );
 450
 451                 if( $this->mDesiredDestName != $filename && $comparableName != $filename ) {
 452                         $warnings['badfilename'] = $filename;
 453                 }
 454
 455                 // Check whether the file extension is on the unwanted list
 456                 global $wgCheckFileExtensions, $wgFileExtensions;
 457                 if ( $wgCheckFileExtensions ) {
 458                         if ( !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) ) {
 459                                 $warnings['filetype-unwanted-type'] = $this->mFinalExtension;
 460                         }
 461                 }
 462
 463                 global $wgUploadSizeWarning;
 464                 if ( $wgUploadSizeWarning && ( $this->mFileSize > $wgUploadSizeWarning ) ) {
 465                         $warnings['large-file'] = $wgUploadSizeWarning;
 466                 }
 467
 468                 if ( $this->mFileSize == 0 ) {
 469                         $warnings['emptyfile'] = true;
 470                 }
 471
 472                 $exists = self::getExistsWarning( $localFile );
 473                 if( $exists !== false ) {
 474                         $warnings['exists'] = $exists;
 475                 }
 476
 477                 // Check dupes against existing files
 478                 $hash = File::sha1Base36( $this->mTempPath );
 479                 $dupes = RepoGroup::singleton()->findBySha1( $hash );
 480                 $title = $this->getTitle();
 481                 // Remove all matches against self
 482                 foreach ( $dupes as $key => $dupe ) {
 483                         if( $title->equals( $dupe->getTitle() ) ) {
 484                                 unset( $dupes[$key] );
 485                         }
 486                 }
 487                 if( $dupes ) {
 488                         $warnings['duplicate'] = $dupes;
 489                 }
 490
 491                 // Check dupes against archives
 492                 $archivedImage = new ArchivedFile( null, 0, "{$hash}.{$this->mFinalExtension}" );
 493                 if ( $archivedImage->getID() > 0 ) {
 494                         $warnings['duplicate-archive'] = $archivedImage->getName();
 495                 }
 496
 497                 return $warnings;
 498         }
 499
 500         /**
 501          * Really perform the upload. Stores the file in the local repo, watches
 502          * if necessary and runs the UploadComplete hook.
 503          *
 504          * @return mixed Status indicating the whether the upload succeeded.
 505          */
 506         public function performUpload( $comment, $pageText, $watch, $user ) {
 507                 $status = $this->getLocalFile()->upload(
 508                         $this->mTempPath,
 509                         $comment,
 510                         $pageText,
 511                         File::DELETE_SOURCE,
 512                         $this->mFileProps,
 513                         false,
 514                         $user
 515                 );
 516
 517                 if( $status->isGood() ) {
 518                         if ( $watch ) {
 519                                 $user->addWatch( $this->getLocalFile()->getTitle() );
 520                         }
 521
 522                         wfRunHooks( 'UploadComplete', array( &$this ) );
 523                 }
 524
 525                 return $status;
 526         }
 527
 528         /**
 529          * Returns the title of the file to be uploaded. Sets mTitleError in case
 530          * the name was illegal.
 531          *
 532          * @return Title The title of the file or null in case the name was illegal
 533          */
 534         public function getTitle() {
 535                 if ( $this->mTitle !== false ) {
 536                         return $this->mTitle;
 537                 }
 538
 539                 /**
 540                  * Chop off any directories in the given filename. Then
 541                  * filter out illegal characters, and try to make a legible name
 542                  * out of it. We'll strip some silently that Title would die on.
 543                  */
 544                 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mDesiredDestName );
 545                 /* Normalize to title form before we do any further processing */
 546                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 547                 if( is_null( $nt ) ) {
 548                         $this->mTitleError = self::ILLEGAL_FILENAME;
 549                         return $this->mTitle = null;
 550                 }
 551                 $this->mFilteredName = $nt->getDBkey();
 552
 553                 /**
 554                  * We'll want to blacklist against *any* 'extension', and use
 555                  * only the final one for the whitelist.
 556                  */
 557                 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
 558
 559                 if( count( $ext ) ) {
 560                         $this->mFinalExtension = trim( $ext[count( $ext ) - 1] );
 561                 } else {
 562                         $this->mFinalExtension = '';
 563                 }
 564
 565                 /* Don't allow users to override the blacklist (check file extension) */
 566                 global $wgCheckFileExtensions, $wgStrictFileExtensions;
 567                 global $wgFileExtensions, $wgFileBlacklist;
 568                 if ( $this->mFinalExtension == '' ) {
 569                         $this->mTitleError = self::FILETYPE_MISSING;
 570                         return $this->mTitle = null;
 571                 } elseif ( $this->checkFileExtensionList( $ext, $wgFileBlacklist ) ||
 572                                 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
 573                                         !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) ) ) {
 574                         $this->mTitleError = self::FILETYPE_BADTYPE;
 575                         return $this->mTitle = null;
 576                 }
 577
 578                 # If there was more than one "extension", reassemble the base
 579                 # filename to prevent bogus complaints about length
 580                 if( count( $ext ) > 1 ) {
 581                         for( $i = 0; $i < count( $ext ) - 1; $i++ ) {
 582                                 $partname .= '.' . $ext[$i];
 583                         }
 584                 }
 585
 586                 if( strlen( $partname ) < 1 ) {
 587                         $this->mTitleError =  self::MIN_LENGTH_PARTNAME;
 588                         return $this->mTitle = null;
 589                 }
 590
 591                 return $this->mTitle = $nt;
 592         }
 593
 594         /**
 595          * Return the local file and initializes if necessary.
 596          */
 597         public function getLocalFile() {
 598                 if( is_null( $this->mLocalFile ) ) {
 599                         $nt = $this->getTitle();
 600                         $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
 601                 }
 602                 return $this->mLocalFile;
 603         }
 604
 605         /**
 606          * NOTE: Probably should be deprecated in favor of UploadStash, but this is sometimes
 607          * called outside that context.
 608          *
 609          * Stash a file in a temporary directory for later processing
 610          * after the user has confirmed it.
 611          *
 612          * If the user doesn't explicitly cancel or accept, these files
 613          * can accumulate in the temp directory.
 614          *
 615          * @param $saveName String: the destination filename
 616          * @param $tempSrc String: the source temporary file to save
 617          * @return String: full path the stashed file, or false on failure
 618          */
 619         protected function saveTempUploadedFile( $saveName, $tempSrc ) {
 620                 $repo = RepoGroup::singleton()->getLocalRepo();
 621                 $status = $repo->storeTemp( $saveName, $tempSrc );
 622                 return $status;
 623         }
 624
 625         /**
 626          * If the user does not supply all necessary information in the first upload form submission (either by accident or
 627          * by design) then we may want to stash the file temporarily, get more information, and publish the file later.
 628          *
 629          * This method will stash a file in a temporary directory for later processing, and save the necessary descriptive info
 630          * into the user's session.
 631          * This method returns the file object, which also has a 'sessionKey' property which can be passed through a form or
 632          * API request to find this stashed file again.
 633          *
 634          * @param $key String: (optional) the session key used to find the file info again. If not supplied, a key will be autogenerated.
 635          * @return File: stashed file
 636          */
 637         public function stashSessionFile( $key = null ) {
 638                 $stash = new UploadStash();
 639                 $data = array(
 640                         'mFileProps' => $this->mFileProps
 641                 );
 642                 $file = $stash->stashFile( $this->mTempPath, $data, $key );
 643                 $this->mLocalFile = $file;
 644                 return $file;
 645         }
 646
 647         /**
 648          * Stash a file in a temporary directory, returning a key which can be used to find the file again. See stashSessionFile().
 649          *
 650          * @param $key String: (optional) the session key used to find the file info again. If not supplied, a key will be autogenerated.
 651          * @return String: session key
 652          */
 653         public function stashSession( $key = null ) {
 654                 return $this->stashSessionFile( $key )->getSessionKey();
 655         }
 656
 657         /**
 658          * If we've modified the upload file we need to manually remove it
 659          * on exit to clean up.
 660          */
 661         public function cleanupTempFile() {
 662                 if ( $this->mRemoveTempFile && $this->mTempPath && file_exists( $this->mTempPath ) ) {
 663                         wfDebug( __METHOD__ . ": Removing temporary file {$this->mTempPath}\n" );
 664                         unlink( $this->mTempPath );
 665                 }
 666         }
 667
 668         public function getTempPath() {
 669                 return $this->mTempPath;
 670         }
 671
 672         /**
 673          * Split a file into a base name and all dot-delimited 'extensions'
 674          * on the end. Some web server configurations will fall back to
 675          * earlier pseudo-'extensions' to determine type and execute
 676          * scripts, so the blacklist needs to check them all.
 677          *
 678          * @return array
 679          */
 680         public static function splitExtensions( $filename ) {
 681                 $bits = explode( '.', $filename );
 682                 $basename = array_shift( $bits );
 683                 return array( $basename, $bits );
 684         }
 685
 686         /**
 687          * Perform case-insensitive match against a list of file extensions.
 688          * Returns true if the extension is in the list.
 689          *
 690          * @param $ext String
 691          * @param $list Array
 692          * @return Boolean
 693          */
 694         public static function checkFileExtension( $ext, $list ) {
 695                 return in_array( strtolower( $ext ), $list );
 696         }
 697
 698         /**
 699          * Perform case-insensitive match against a list of file extensions.
 700          * Returns true if any of the extensions are in the list.
 701          *
 702          * @param $ext Array
 703          * @param $list Array
 704          * @return Boolean
 705          */
 706         public static function checkFileExtensionList( $ext, $list ) {
 707                 foreach( $ext as $e ) {
 708                         if( in_array( strtolower( $e ), $list ) ) {
 709                                 return true;
 710                         }
 711                 }
 712                 return false;
 713         }
 714
 715         /**
 716          * Checks if the mime type of the uploaded file matches the file extension.
 717          *
 718          * @param $mime String: the mime type of the uploaded file
 719          * @param $extension String: the filename extension that the file is to be served with
 720          * @return Boolean
 721          */
 722         public static function verifyExtension( $mime, $extension ) {
 723                 $magic = MimeMagic::singleton();
 724
 725                 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' )
 726                         if ( !$magic->isRecognizableExtension( $extension ) ) {
 727                                 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
 728                                         "unrecognized extension '$extension', can't verify\n" );
 729                                 return true;
 730                         } else {
 731                                 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; ".
 732                                         "recognized extension '$extension', so probably invalid file\n" );
 733                                 return false;
 734                         }
 735
 736                 $match = $magic->isMatchingExtension( $extension, $mime );
 737
 738                 if ( $match === null ) {
 739                         wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
 740                         return true;
 741                 } elseif( $match === true ) {
 742                         wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
 743
 744                         #TODO: if it's a bitmap, make sure PHP or ImageMagic resp. can handle it!
 745                         return true;
 746
 747                 } else {
 748                         wfDebug( __METHOD__ . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
 749                         return false;
 750                 }
 751         }
 752
 753         /**
 754          * Heuristic for detecting files that *could* contain JavaScript instructions or
 755          * things that may look like HTML to a browser and are thus
 756          * potentially harmful. The present implementation will produce false
 757          * positives in some situations.
 758          *
 759          * @param $file String: pathname to the temporary upload file
 760          * @param $mime String: the mime type of the file
 761          * @param $extension String: the extension of the file
 762          * @return Boolean: true if the file contains something looking like embedded scripts
 763          */
 764         public static function detectScript( $file, $mime, $extension ) {
 765                 global $wgAllowTitlesInSVG;
 766
 767                 # ugly hack: for text files, always look at the entire file.
 768                 # For binary field, just check the first K.
 769
 770                 if( strpos( $mime,'text/' ) === 0 ) {
 771                         $chunk = file_get_contents( $file );
 772                 } else {
 773                         $fp = fopen( $file, 'rb' );
 774                         $chunk = fread( $fp, 1024 );
 775                         fclose( $fp );
 776                 }
 777
 778                 $chunk = strtolower( $chunk );
 779
 780                 if( !$chunk ) {
 781                         return false;
 782                 }
 783
 784                 # decode from UTF-16 if needed (could be used for obfuscation).
 785                 if( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
 786                         $enc = 'UTF-16BE';
 787                 } elseif( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
 788                         $enc = 'UTF-16LE';
 789                 } else {
 790                         $enc = null;
 791                 }
 792
 793                 if( $enc ) {
 794                         $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
 795                 }
 796
 797                 $chunk = trim( $chunk );
 798
 799                 # FIXME: convert from UTF-16 if necessarry!
 800                 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
 801
 802                 # check for HTML doctype
 803                 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
 804                         return true;
 805                 }
 806
 807                 /**
 808                  * Internet Explorer for Windows performs some really stupid file type
 809                  * autodetection which can cause it to interpret valid image files as HTML
 810                  * and potentially execute JavaScript, creating a cross-site scripting
 811                  * attack vectors.
 812                  *
 813                  * Apple's Safari browser also performs some unsafe file type autodetection
 814                  * which can cause legitimate files to be interpreted as HTML if the
 815                  * web server is not correctly configured to send the right content-type
 816                  * (or if you're really uploading plain text and octet streams!)
 817                  *
 818                  * Returns true if IE is likely to mistake the given file for HTML.
 819                  * Also returns true if Safari would mistake the given file for HTML
 820                  * when served with a generic content-type.
 821                  */
 822                 $tags = array(
 823                         '<a href',
 824                         '<body',
 825                         '<head',
 826                         '<html',   #also in safari
 827                         '<img',
 828                         '<pre',
 829                         '<script', #also in safari
 830                         '<table'
 831                 );
 832
 833                 if( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
 834                         $tags[] = '<title';
 835                 }
 836
 837                 foreach( $tags as $tag ) {
 838                         if( false !== strpos( $chunk, $tag ) ) {
 839                                 return true;
 840                         }
 841                 }
 842
 843                 /*
 844                  * look for JavaScript
 845                  */
 846
 847                 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
 848                 $chunk = Sanitizer::decodeCharReferences( $chunk );
 849
 850                 # look for script-types
 851                 if( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
 852                         return true;
 853                 }
 854
 855                 # look for html-style script-urls
 856                 if( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
 857                         return true;
 858                 }
 859
 860                 # look for css-style script-urls
 861                 if( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
 862                         return true;
 863                 }
 864
 865                 wfDebug( __METHOD__ . ": no scripts found\n" );
 866                 return false;
 867         }
 868
 869         protected function detectScriptInSvg( $filename ) {
 870                 $check = new XmlTypeCheck( $filename, array( $this, 'checkSvgScriptCallback' ) );
 871                 return $check->filterMatch;
 872         }
 873
 874         /**
 875          * @todo Replace this with a whitelist filter!
 876          */
 877         public function checkSvgScriptCallback( $element, $attribs ) {
 878                 $stripped = $this->stripXmlNamespace( $element );
 879
 880                 if( $stripped == 'script' ) {
 881                         wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
 882                         return true;
 883                 }
 884
 885                 foreach( $attribs as $attrib => $value ) {
 886                         $stripped = $this->stripXmlNamespace( $attrib );
 887                         if( substr( $stripped, 0, 2 ) == 'on' ) {
 888                                 wfDebug( __METHOD__ . ": Found script attribute '$attrib'='value' in uploaded file.\n" );
 889                                 return true;
 890                         }
 891                         if( $stripped == 'href' && strpos( strtolower( $value ), 'javascript:' ) !== false ) {
 892                                 wfDebug( __METHOD__ . ": Found script href attribute '$attrib'='$value' in uploaded file.\n" );
 893                                 return true;
 894                         }
 895                 }
 896         }
 897
 898         private function stripXmlNamespace( $name ) {
 899                 // 'http://www.w3.org/2000/svg:script' -> 'script'
 900                 $parts = explode( ':', strtolower( $name ) );
 901                 return array_pop( $parts );
 902         }
 903
 904         /**
 905          * Generic wrapper function for a virus scanner program.
 906          * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
 907          * $wgAntivirusRequired may be used to deny upload if the scan fails.
 908          *
 909          * @param $file String: pathname to the temporary upload file
 910          * @return mixed false if not virus is found, NULL if the scan fails or is disabled,
 911          *         or a string containing feedback from the virus scanner if a virus was found.
 912          *         If textual feedback is missing but a virus was found, this function returns true.
 913          */
 914         public static function detectVirus( $file ) {
 915                 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
 916
 917                 if ( !$wgAntivirus ) {
 918                         wfDebug( __METHOD__ . ": virus scanner disabled\n" );
 919                         return null;
 920                 }
 921
 922                 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
 923                         wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
 924                         $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
 925                                 array( 'virus-badscanner', $wgAntivirus ) );
 926                         return wfMsg( 'virus-unknownscanner' ) . " $wgAntivirus";
 927                 }
 928
 929                 # look up scanner configuration
 930                 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
 931                 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
 932                 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
 933                         $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
 934
 935                 if ( strpos( $command, "%f" ) === false ) {
 936                         # simple pattern: append file to scan
 937                         $command .= " " . wfEscapeShellArg( $file );
 938                 } else {
 939                         # complex pattern: replace "%f" with file to scan
 940                         $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
 941                 }
 942
 943                 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
 944
 945                 # execute virus scanner
 946                 $exitCode = false;
 947
 948                 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
 949                 #      that does not seem to be worth the pain.
 950                 #      Ask me (Duesentrieb) about it if it's ever needed.
 951                 $output = wfShellExec( "$command 2>&1", $exitCode );
 952
 953                 # map exit code to AV_xxx constants.
 954                 $mappedCode = $exitCode;
 955                 if ( $exitCodeMap ) {
 956                         if ( isset( $exitCodeMap[$exitCode] ) ) {
 957                                 $mappedCode = $exitCodeMap[$exitCode];
 958                         } elseif ( isset( $exitCodeMap["*"] ) ) {
 959                                 $mappedCode = $exitCodeMap["*"];
 960                         }
 961                 }
 962
 963                 if ( $mappedCode === AV_SCAN_FAILED ) {
 964                         # scan failed (code was mapped to false by $exitCodeMap)
 965                         wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
 966
 967                         if ( $wgAntivirusRequired ) {
 968                                 return wfMsg( 'virus-scanfailed', array( $exitCode ) );
 969                         } else {
 970                                 return null;
 971                         }
 972                 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
 973                         # scan failed because filetype is unknown (probably imune)
 974                         wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
 975                         return null;
 976                 } elseif ( $mappedCode === AV_NO_VIRUS ) {
 977                         # no virus found
 978                         wfDebug( __METHOD__ . ": file passed virus scan.\n" );
 979                         return false;
 980                 } else {
 981                         $output = trim( $output );
 982
 983                         if ( !$output ) {
 984                                 $output = true; #if there's no output, return true
 985                         } elseif ( $msgPattern ) {
 986                                 $groups = array();
 987                                 if ( preg_match( $msgPattern, $output, $groups ) ) {
 988                                         if ( $groups[1] ) {
 989                                                 $output = $groups[1];
 990                                         }
 991                                 }
 992                         }
 993
 994                         wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
 995                         return $output;
 996                 }
 997         }
 998
 999         /**
1000          * Check if the temporary file is MacBinary-encoded, as some uploads
1001          * from Internet Explorer on Mac OS Classic and Mac OS X will be.
1002          * If so, the data fork will be extracted to a second temporary file,
1003          * which will then be checked for validity and either kept or discarded.
1004          */
1005         private function checkMacBinary() {
1006                 $macbin = new MacBinary( $this->mTempPath );
1007                 if( $macbin->isValid() ) {
1008                         $dataFile = tempnam( wfTempDir(), 'WikiMacBinary' );
1009                         $dataHandle = fopen( $dataFile, 'wb' );
1010
1011                         wfDebug( __METHOD__ . ": Extracting MacBinary data fork to $dataFile\n" );
1012                         $macbin->extractData( $dataHandle );
1013
1014                         $this->mTempPath = $dataFile;
1015                         $this->mFileSize = $macbin->dataForkLength();
1016
1017                         // We'll have to manually remove the new file if it's not kept.
1018                         $this->mRemoveTempFile = true;
1019                 }
1020                 $macbin->close();
1021         }
1022
1023         /**
1024          * Check if there's an overwrite conflict and, if so, if restrictions
1025          * forbid this user from performing the upload.
1026          *
1027          * @return mixed true on success, error string on failure
1028          */
1029         private function checkOverwrite( $user ) {
1030                 // First check whether the local file can be overwritten
1031                 $file = $this->getLocalFile();
1032                 if( $file->exists() ) {
1033                         if( !self::userCanReUpload( $user, $file ) ) {
1034                                 return 'fileexists-forbidden';
1035                         } else {
1036                                 return true;
1037                         }
1038                 }
1039
1040                 /* Check shared conflicts: if the local file does not exist, but
1041                  * wfFindFile finds a file, it exists in a shared repository.
1042                  */
1043                 $file = wfFindFile( $this->getTitle() );
1044                 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1045                         return 'fileexists-shared-forbidden';
1046                 }
1047
1048                 return true;
1049         }
1050
1051         /**
1052          * Check if a user is the last uploader
1053          *
1054          * @param $user User object
1055          * @param $img String: image name
1056          * @return Boolean
1057          */
1058         public static function userCanReUpload( User $user, $img ) {
1059                 if( $user->isAllowed( 'reupload' ) ) {
1060                         return true; // non-conditional
1061                 }
1062                 if( !$user->isAllowed( 'reupload-own' ) ) {
1063                         return false;
1064                 }
1065                 if( is_string( $img ) ) {
1066                         $img = wfLocalFile( $img );
1067                 }
1068                 if ( !( $img instanceof LocalFile ) ) {
1069                         return false;
1070                 }
1071
1072                 return $user->getId() == $img->getUser( 'id' );
1073         }
1074
1075         /**
1076          * Helper function that does various existence checks for a file.
1077          * The following checks are performed:
1078          * - The file exists
1079          * - Article with the same name as the file exists
1080          * - File exists with normalized extension
1081          * - The file looks like a thumbnail and the original exists
1082          *
1083          * @param $file The File object to check
1084          * @return mixed False if the file does not exists, else an array
1085          */
1086         public static function getExistsWarning( $file ) {
1087                 if( $file->exists() ) {
1088                         return array( 'warning' => 'exists', 'file' => $file );
1089                 }
1090
1091                 if( $file->getTitle()->getArticleID() ) {
1092                         return array( 'warning' => 'page-exists', 'file' => $file );
1093                 }
1094
1095                 if ( $file->wasDeleted() && !$file->exists() ) {
1096                         return array( 'warning' => 'was-deleted', 'file' => $file );
1097                 }
1098
1099                 if( strpos( $file->getName(), '.' ) == false ) {
1100                         $partname = $file->getName();
1101                         $extension = '';
1102                 } else {
1103                         $n = strrpos( $file->getName(), '.' );
1104                         $extension = substr( $file->getName(), $n + 1 );
1105                         $partname = substr( $file->getName(), 0, $n );
1106                 }
1107                 $normalizedExtension = File::normalizeExtension( $extension );
1108
1109                 if ( $normalizedExtension != $extension ) {
1110                         // We're not using the normalized form of the extension.
1111                         // Normal form is lowercase, using most common of alternate
1112                         // extensions (eg 'jpg' rather than 'JPEG').
1113                         //
1114                         // Check for another file using the normalized form...
1115                         $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
1116                         $file_lc = wfLocalFile( $nt_lc );
1117
1118                         if( $file_lc->exists() ) {
1119                                 return array(
1120                                         'warning' => 'exists-normalized',
1121                                         'file' => $file,
1122                                         'normalizedFile' => $file_lc
1123                                 );
1124                         }
1125                 }
1126
1127                 if ( self::isThumbName( $file->getName() ) ) {
1128                         # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1129                         $nt_thb = Title::newFromText( substr( $partname , strpos( $partname , '-' ) +1 ) . '.' . $extension, NS_FILE );
1130                         $file_thb = wfLocalFile( $nt_thb );
1131                         if( $file_thb->exists() ) {
1132                                 return array(
1133                                         'warning' => 'thumb',
1134                                         'file' => $file,
1135                                         'thumbFile' => $file_thb
1136                                 );
1137                         } else {
1138                                 // File does not exist, but we just don't like the name
1139                                 return array(
1140                                         'warning' => 'thumb-name',
1141                                         'file' => $file,
1142                                         'thumbFile' => $file_thb
1143                                 );
1144                         }
1145                 }
1146
1147
1148                 foreach( self::getFilenamePrefixBlacklist() as $prefix ) {
1149                         if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1150                                 return array(
1151                                         'warning' => 'bad-prefix',
1152                                         'file' => $file,
1153                                         'prefix' => $prefix
1154                                 );
1155                         }
1156                 }
1157
1158                 return false;
1159         }
1160
1161         /**
1162          * Helper function that checks whether the filename looks like a thumbnail
1163          */
1164         public static function isThumbName( $filename ) {
1165                 $n = strrpos( $filename, '.' );
1166                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
1167                 return (
1168                                         substr( $partname , 3, 3 ) == 'px-' ||
1169                                         substr( $partname , 2, 3 ) == 'px-'
1170                                 ) &&
1171                                 preg_match( "/[0-9]{2}/" , substr( $partname , 0, 2 ) );
1172         }
1173
1174         /**
1175          * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
1176          *
1177          * @return array list of prefixes
1178          */
1179         public static function getFilenamePrefixBlacklist() {
1180                 $blacklist = array();
1181                 $message = wfMsgForContent( 'filename-prefix-blacklist' );
1182                 if( $message && !( wfEmptyMsg( 'filename-prefix-blacklist', $message ) || $message == '-' ) ) {
1183                         $lines = explode( "\n", $message );
1184                         foreach( $lines as $line ) {
1185                                 // Remove comment lines
1186                                 $comment = substr( trim( $line ), 0, 1 );
1187                                 if ( $comment == '#' || $comment == '' ) {
1188                                         continue;
1189                                 }
1190                                 // Remove additional comments after a prefix
1191                                 $comment = strpos( $line, '#' );
1192                                 if ( $comment > 0 ) {
1193                                         $line = substr( $line, 0, $comment-1 );
1194                                 }
1195                                 $blacklist[] = trim( $line );
1196                         }
1197                 }
1198                 return $blacklist;
1199         }
1200
1201         /**
1202          * Gets image info about the file just uploaded.
1203          *
1204          * Also has the effect of setting metadata to be an 'indexed tag name' in returned API result if
1205          * 'metadata' was requested. Oddly, we have to pass the "result" object down just so it can do that
1206          * with the appropriate format, presumably.
1207          *
1208          * @param $result ApiResult:
1209          * @return Array: image info
1210          */
1211         public function getImageInfo( $result ) {
1212                 $file = $this->getLocalFile();
1213                 // TODO This cries out for refactoring. We really want to say $file->getAllInfo(); here.
1214                 // Perhaps "info" methods should be moved into files, and the API should just wrap them in queries.
1215                 if ( $file instanceof UploadStashFile ) {
1216                         $imParam = ApiQueryStashImageInfo::getPropertyNames();
1217                         $info = ApiQueryStashImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1218                 } else {
1219                         $imParam = ApiQueryImageInfo::getPropertyNames();
1220                         $info = ApiQueryImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1221                 }
1222                 return $info;
1223         }
1224
1225
1226         public function convertVerifyErrorToStatus( $error ) {
1227                 $code = $error['status'];
1228                 unset( $code['status'] );
1229                 return Status::newFatal( $this->getVerificationErrorCode( $code ), $error );
1230         }
1231 }