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                         // @fixme This status needs another name...
 270                         return array( 'status' => self::HOOK_ABORTED, 'error' => $error );
 271                 }
 272
 273                 return array( 'status' => self::OK );
 274         }
 275
 276         /**
 277          * Verify that the name is valid and, if necessary, that we can overwrite
 278          *
 279          * @return mixed true if valid, otherwise and array with 'status'
 280          * and other keys
 281          **/
 282         protected function validateName() {
 283                 $nt = $this->getTitle();
 284                 if( is_null( $nt ) ) {
 285                         $result = array( 'status' => $this->mTitleError );
 286                         if( $this->mTitleError == self::ILLEGAL_FILENAME ) {
 287                                 $result['filtered'] = $this->mFilteredName;
 288                         }
 289                         if ( $this->mTitleError == self::FILETYPE_BADTYPE ) {
 290                                 $result['finalExt'] = $this->mFinalExtension;
 291                         }
 292                         return $result;
 293                 }
 294                 $this->mDestName = $this->getLocalFile()->getName();
 295
 296                 return true;
 297         }
 298
 299         /**
 300          * Verify the mime type
 301          * @param $magic MagicMime object
 302          * @param $mime string representing the mime
 303          * @return mixed true if the file is verified, an array otherwise
 304          */
 305         protected function verifyMimeType( $magic, $mime ) {
 306                 global $wgVerifyMimeType;
 307                 if ( $wgVerifyMimeType ) {
 308                         wfDebug ( "\n\nmime: <$mime> extension: <{$this->mFinalExtension}>\n\n");
 309                         if ( !$this->verifyExtension( $mime, $this->mFinalExtension ) ) {
 310                                 return array( 'filetype-mime-mismatch' );
 311                         }
 312
 313                         global $wgMimeTypeBlacklist;
 314                         if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
 315                                 return array( 'filetype-badmime', $mime );
 316                         }
 317
 318                         # Check IE type
 319                         $fp = fopen( $this->mTempPath, 'rb' );
 320                         $chunk = fread( $fp, 256 );
 321                         fclose( $fp );
 322                         $extMime = $magic->guessTypesForExtension( $this->mFinalExtension );
 323                         $ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime );
 324                         foreach ( $ieTypes as $ieType ) {
 325                                 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
 326                                         return array( 'filetype-bad-ie-mime', $ieType );
 327                                 }
 328                         }
 329                 }
 330
 331                 return true;
 332         }
 333
 334         /**
 335          * Verifies that it's ok to include the uploaded file
 336          *
 337          * @return mixed true of the file is verified, array otherwise.
 338          */
 339         protected function verifyFile() {
 340                 # get the title, even though we are doing nothing with it, because
 341                 # we need to populate mFinalExtension
 342                 $nt = $this->getTitle();
 343
 344                 $this->mFileProps = File::getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 345                 $this->checkMacBinary();
 346
 347                 # magically determine mime type
 348                 $magic = MimeMagic::singleton();
 349                 $mime = $magic->guessMimeType( $this->mTempPath, false );
 350
 351                 # check mime type, if desired
 352                 $status = $this->verifyMimeType( $magic, $mime );
 353                 if ( $status !== true ) {
 354                         return $status;
 355                 }
 356
 357                 # check for htmlish code and javascript
 358                 if( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
 359                         return array( 'uploadscripted' );
 360                 }
 361                 if( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 362                         if( $this->detectScriptInSvg( $this->mTempPath ) ) {
 363                                 return array( 'uploadscripted' );
 364                         }
 365                 }
 366
 367                 /**
 368                 * Scan the uploaded file for viruses
 369                 */
 370                 $virus = $this->detectVirus( $this->mTempPath );
 371                 if ( $virus ) {
 372                         return array( 'uploadvirus', $virus );
 373                 }
 374
 375                 wfRunHooks( 'UploadVerifyFile', array( $this, $mime, &$status ) );
 376                 if ( $status !== true ) {
 377                         return $status;
 378                 }
 379
 380                 wfDebug( __METHOD__ . ": all clear; passing.\n" );
 381                 return true;
 382         }
 383
 384         /**
 385          * Check whether the user can edit, upload and create the image. This
 386          * checks only against the current title; if it returns errors, it may
 387          * very well be that another title will not give errors. Therefore
 388          * isAllowed() should be called as well for generic is-user-blocked or
 389          * can-user-upload checking.
 390          *
 391          * @param $user the User object to verify the permissions against
 392          * @return mixed An array as returned by getUserPermissionsErrors or true
 393          *               in case the user has proper permissions.
 394          */
 395         public function verifyPermissions( $user ) {
 396                 /**
 397                  * If the image is protected, non-sysop users won't be able
 398                  * to modify it by uploading a new revision.
 399                  */
 400                 $nt = $this->getTitle();
 401                 if( is_null( $nt ) ) {
 402                         return true;
 403                 }
 404                 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
 405                 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
 406                 if ( $nt->exists() ) {
 407                         $permErrorsCreate = $nt->getUserPermissionsErrors( 'createpage', $user );
 408                 } else {
 409                         $permErrorsCreate = array();
 410                 }
 411                 if( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
 412                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
 413                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
 414                         return $permErrors;
 415                 }
 416
 417                 $overwriteError = $this->checkOverwrite( $user );
 418                 if ( $overwriteError !== true ) {
 419                         return array( array( $overwriteError ) );
 420                 }
 421
 422                 return true;
 423         }
 424
 425         /**
 426          * Check for non fatal problems with the file
 427          *
 428          * @return Array of warnings
 429          */
 430         public function checkWarnings() {
 431                 $warnings = array();
 432
 433                 $localFile = $this->getLocalFile();
 434                 $filename = $localFile->getName();
 435                 $n = strrpos( $filename, '.' );
 436                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
 437
 438                 /**
 439                  * Check whether the resulting filename is different from the desired one,
 440                  * but ignore things like ucfirst() and spaces/underscore things
 441                  */
 442                 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName );
 443                 $comparableName = Title::capitalize( $comparableName, NS_FILE );
 444
 445                 if( $this->mDesiredDestName != $filename && $comparableName != $filename ) {
 446                         $warnings['badfilename'] = $filename;
 447                 }
 448
 449                 // Check whether the file extension is on the unwanted list
 450                 global $wgCheckFileExtensions, $wgFileExtensions;
 451                 if ( $wgCheckFileExtensions ) {
 452                         if ( !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) ) {
 453                                 $warnings['filetype-unwanted-type'] = $this->mFinalExtension;
 454                         }
 455                 }
 456
 457                 global $wgUploadSizeWarning;
 458                 if ( $wgUploadSizeWarning && ( $this->mFileSize > $wgUploadSizeWarning ) ) {
 459                         $warnings['large-file'] = $wgUploadSizeWarning;
 460                 }
 461
 462                 if ( $this->mFileSize == 0 ) {
 463                         $warnings['emptyfile'] = true;
 464                 }
 465
 466                 $exists = self::getExistsWarning( $localFile );
 467                 if( $exists !== false ) {
 468                         $warnings['exists'] = $exists;
 469                 }
 470
 471                 // Check dupes against existing files
 472                 $hash = File::sha1Base36( $this->mTempPath );
 473                 $dupes = RepoGroup::singleton()->findBySha1( $hash );
 474                 $title = $this->getTitle();
 475                 // Remove all matches against self
 476                 foreach ( $dupes as $key => $dupe ) {
 477                         if( $title->equals( $dupe->getTitle() ) ) {
 478                                 unset( $dupes[$key] );
 479                         }
 480                 }
 481                 if( $dupes ) {
 482                         $warnings['duplicate'] = $dupes;
 483                 }
 484
 485                 // Check dupes against archives
 486                 $archivedImage = new ArchivedFile( null, 0, "{$hash}.{$this->mFinalExtension}" );
 487                 if ( $archivedImage->getID() > 0 ) {
 488                         $warnings['duplicate-archive'] = $archivedImage->getName();
 489                 }
 490
 491                 return $warnings;
 492         }
 493
 494         /**
 495          * Really perform the upload. Stores the file in the local repo, watches
 496          * if necessary and runs the UploadComplete hook.
 497          *
 498          * @return mixed Status indicating the whether the upload succeeded.
 499          */
 500         public function performUpload( $comment, $pageText, $watch, $user ) {
 501                 wfDebug( "\n\n\performUpload: sum: " . $comment . ' c: ' . $pageText .
 502                         ' w: ' . $watch );
 503                 $status = $this->getLocalFile()->upload( $this->mTempPath, $comment, $pageText,
 504                         File::DELETE_SOURCE, $this->mFileProps, false, $user );
 505
 506                 if( $status->isGood() ) {
 507                         if ( $watch ) {
 508                                 $user->addWatch( $this->getLocalFile()->getTitle() );
 509                         }
 510
 511                         wfRunHooks( 'UploadComplete', array( &$this ) );
 512                 }
 513
 514                 return $status;
 515         }
 516
 517         /**
 518          * Returns the title of the file to be uploaded. Sets mTitleError in case
 519          * the name was illegal.
 520          *
 521          * @return Title The title of the file or null in case the name was illegal
 522          */
 523         public function getTitle() {
 524                 if ( $this->mTitle !== false ) {
 525                         return $this->mTitle;
 526                 }
 527
 528                 /**
 529                  * Chop off any directories in the given filename. Then
 530                  * filter out illegal characters, and try to make a legible name
 531                  * out of it. We'll strip some silently that Title would die on.
 532                  */
 533                 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mDesiredDestName );
 534                 /* Normalize to title form before we do any further processing */
 535                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 536                 if( is_null( $nt ) ) {
 537                         $this->mTitleError = self::ILLEGAL_FILENAME;
 538                         return $this->mTitle = null;
 539                 }
 540                 $this->mFilteredName = $nt->getDBkey();
 541
 542                 /**
 543                  * We'll want to blacklist against *any* 'extension', and use
 544                  * only the final one for the whitelist.
 545                  */
 546                 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
 547
 548                 if( count( $ext ) ) {
 549                         $this->mFinalExtension = trim( $ext[count( $ext ) - 1] );
 550                 } else {
 551                         $this->mFinalExtension = '';
 552                 }
 553
 554                 /* Don't allow users to override the blacklist (check file extension) */
 555                 global $wgCheckFileExtensions, $wgStrictFileExtensions;
 556                 global $wgFileExtensions, $wgFileBlacklist;
 557                 if ( $this->mFinalExtension == '' ) {
 558                         $this->mTitleError = self::FILETYPE_MISSING;
 559                         return $this->mTitle = null;
 560                 } elseif ( $this->checkFileExtensionList( $ext, $wgFileBlacklist ) ||
 561                                 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
 562                                         !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) ) ) {
 563                         $this->mTitleError = self::FILETYPE_BADTYPE;
 564                         return $this->mTitle = null;
 565                 }
 566
 567                 # If there was more than one "extension", reassemble the base
 568                 # filename to prevent bogus complaints about length
 569                 if( count( $ext ) > 1 ) {
 570                         for( $i = 0; $i < count( $ext ) - 1; $i++ ) {
 571                                 $partname .= '.' . $ext[$i];
 572                         }
 573                 }
 574
 575                 if( strlen( $partname ) < 1 ) {
 576                         $this->mTitleError =  self::MIN_LENGTH_PARTNAME;
 577                         return $this->mTitle = null;
 578                 }
 579
 580                 return $this->mTitle = $nt;
 581         }
 582
 583         /**
 584          * Return the local file and initializes if necessary.
 585          */
 586         public function getLocalFile() {
 587                 if( is_null( $this->mLocalFile ) ) {
 588                         $nt = $this->getTitle();
 589                         $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
 590                 }
 591                 return $this->mLocalFile;
 592         }
 593
 594         /**
 595          * Stash a file in a temporary directory for later processing
 596          * after the user has confirmed it.
 597          *
 598          * If the user doesn't explicitly cancel or accept, these files
 599          * can accumulate in the temp directory.
 600          *
 601          * @param $saveName String: the destination filename
 602          * @param $tempSrc String: the source temporary file to save
 603          * @return String: full path the stashed file, or false on failure
 604          */
 605         protected function saveTempUploadedFile( $saveName, $tempSrc ) {
 606                 $repo = RepoGroup::singleton()->getLocalRepo();
 607                 $status = $repo->storeTemp( $saveName, $tempSrc );
 608                 return $status;
 609         }
 610
 611         /**
 612          * Stash a file in a temporary directory for later processing,
 613          * and save the necessary descriptive info into the session.
 614          * Returns a key value which will be passed through a form
 615          * to pick up the path info on a later invocation.
 616          *
 617          * @return Integer: session key
 618          */
 619         public function stashSession( $key = null ) {
 620                 $status = $this->saveTempUploadedFile( $this->mDestName, $this->mTempPath );
 621                 if( !$status->isOK() ) {
 622                         # Couldn't save the file.
 623                         return false;
 624                 }
 625
 626                 if ( is_null( $key ) ) {
 627                         $key = $this->getSessionKey();
 628                 }
 629                 $_SESSION[self::SESSION_KEYNAME][$key] = array(
 630                         'mTempPath'       => $status->value,
 631                         'mFileSize'       => $this->mFileSize,
 632                         'mFileProps'      => $this->mFileProps,
 633                         'version'         => self::SESSION_VERSION,
 634                 );
 635                 return $key;
 636         }
 637
 638         /**
 639          * Generate a random session key from stash in cases where we want
 640          * to start an upload without much information
 641          */
 642         protected function getSessionKey() {
 643                 $key = mt_rand( 0, 0x7fffffff );
 644                 $_SESSION[self::SESSION_KEYNAME][$key] = array();
 645                 return $key;
 646         }
 647
 648         /**
 649          * If we've modified the upload file we need to manually remove it
 650          * on exit to clean up.
 651          */
 652         public function cleanupTempFile() {
 653                 if ( $this->mRemoveTempFile && $this->mTempPath && file_exists( $this->mTempPath ) ) {
 654                         wfDebug( __METHOD__ . ": Removing temporary file {$this->mTempPath}\n" );
 655                         unlink( $this->mTempPath );
 656                 }
 657         }
 658
 659         public function getTempPath() {
 660                 return $this->mTempPath;
 661         }
 662
 663         /**
 664          * Split a file into a base name and all dot-delimited 'extensions'
 665          * on the end. Some web server configurations will fall back to
 666          * earlier pseudo-'extensions' to determine type and execute
 667          * scripts, so the blacklist needs to check them all.
 668          *
 669          * @return array
 670          */
 671         public static function splitExtensions( $filename ) {
 672                 $bits = explode( '.', $filename );
 673                 $basename = array_shift( $bits );
 674                 return array( $basename, $bits );
 675         }
 676
 677         /**
 678          * Perform case-insensitive match against a list of file extensions.
 679          * Returns true if the extension is in the list.
 680          *
 681          * @param $ext String
 682          * @param $list Array
 683          * @return Boolean
 684          */
 685         public static function checkFileExtension( $ext, $list ) {
 686                 return in_array( strtolower( $ext ), $list );
 687         }
 688
 689         /**
 690          * Perform case-insensitive match against a list of file extensions.
 691          * Returns true if any of the extensions are in the list.
 692          *
 693          * @param $ext Array
 694          * @param $list Array
 695          * @return Boolean
 696          */
 697         public static function checkFileExtensionList( $ext, $list ) {
 698                 foreach( $ext as $e ) {
 699                         if( in_array( strtolower( $e ), $list ) ) {
 700                                 return true;
 701                         }
 702                 }
 703                 return false;
 704         }
 705
 706         /**
 707          * Checks if the mime type of the uploaded file matches the file extension.
 708          *
 709          * @param $mime String: the mime type of the uploaded file
 710          * @param $extension String: the filename extension that the file is to be served with
 711          * @return Boolean
 712          */
 713         public static function verifyExtension( $mime, $extension ) {
 714                 $magic = MimeMagic::singleton();
 715
 716                 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' )
 717                         if ( !$magic->isRecognizableExtension( $extension ) ) {
 718                                 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
 719                                         "unrecognized extension '$extension', can't verify\n" );
 720                                 return true;
 721                         } else {
 722                                 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; ".
 723                                         "recognized extension '$extension', so probably invalid file\n" );
 724                                 return false;
 725                         }
 726
 727                 $match = $magic->isMatchingExtension( $extension, $mime );
 728
 729                 if ( $match === null ) {
 730                         wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
 731                         return true;
 732                 } elseif( $match === true ) {
 733                         wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
 734
 735                         #TODO: if it's a bitmap, make sure PHP or ImageMagic resp. can handle it!
 736                         return true;
 737
 738                 } else {
 739                         wfDebug( __METHOD__ . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
 740                         return false;
 741                 }
 742         }
 743
 744         /**
 745          * Heuristic for detecting files that *could* contain JavaScript instructions or
 746          * things that may look like HTML to a browser and are thus
 747          * potentially harmful. The present implementation will produce false
 748          * positives in some situations.
 749          *
 750          * @param $file String: pathname to the temporary upload file
 751          * @param $mime String: the mime type of the file
 752          * @param $extension String: the extension of the file
 753          * @return Boolean: true if the file contains something looking like embedded scripts
 754          */
 755         public static function detectScript( $file, $mime, $extension ) {
 756                 global $wgAllowTitlesInSVG;
 757
 758                 # ugly hack: for text files, always look at the entire file.
 759                 # For binary field, just check the first K.
 760
 761                 if( strpos( $mime,'text/' ) === 0 ) {
 762                         $chunk = file_get_contents( $file );
 763                 } else {
 764                         $fp = fopen( $file, 'rb' );
 765                         $chunk = fread( $fp, 1024 );
 766                         fclose( $fp );
 767                 }
 768
 769                 $chunk = strtolower( $chunk );
 770
 771                 if( !$chunk ) {
 772                         return false;
 773                 }
 774
 775                 # decode from UTF-16 if needed (could be used for obfuscation).
 776                 if( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
 777                         $enc = 'UTF-16BE';
 778                 } elseif( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
 779                         $enc = 'UTF-16LE';
 780                 } else {
 781                         $enc = null;
 782                 }
 783
 784                 if( $enc ) {
 785                         $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
 786                 }
 787
 788                 $chunk = trim( $chunk );
 789
 790                 # FIXME: convert from UTF-16 if necessarry!
 791                 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
 792
 793                 # check for HTML doctype
 794                 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
 795                         return true;
 796                 }
 797
 798                 /**
 799                  * Internet Explorer for Windows performs some really stupid file type
 800                  * autodetection which can cause it to interpret valid image files as HTML
 801                  * and potentially execute JavaScript, creating a cross-site scripting
 802                  * attack vectors.
 803                  *
 804                  * Apple's Safari browser also performs some unsafe file type autodetection
 805                  * which can cause legitimate files to be interpreted as HTML if the
 806                  * web server is not correctly configured to send the right content-type
 807                  * (or if you're really uploading plain text and octet streams!)
 808                  *
 809                  * Returns true if IE is likely to mistake the given file for HTML.
 810                  * Also returns true if Safari would mistake the given file for HTML
 811                  * when served with a generic content-type.
 812                  */
 813                 $tags = array(
 814                         '<a href',
 815                         '<body',
 816                         '<head',
 817                         '<html',   #also in safari
 818                         '<img',
 819                         '<pre',
 820                         '<script', #also in safari
 821                         '<table'
 822                 );
 823
 824                 if( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
 825                         $tags[] = '<title';
 826                 }
 827
 828                 foreach( $tags as $tag ) {
 829                         if( false !== strpos( $chunk, $tag ) ) {
 830                                 return true;
 831                         }
 832                 }
 833
 834                 /*
 835                  * look for JavaScript
 836                  */
 837
 838                 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
 839                 $chunk = Sanitizer::decodeCharReferences( $chunk );
 840
 841                 # look for script-types
 842                 if( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
 843                         return true;
 844                 }
 845
 846                 # look for html-style script-urls
 847                 if( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
 848                         return true;
 849                 }
 850
 851                 # look for css-style script-urls
 852                 if( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
 853                         return true;
 854                 }
 855
 856                 wfDebug( __METHOD__ . ": no scripts found\n" );
 857                 return false;
 858         }
 859
 860         protected function detectScriptInSvg( $filename ) {
 861                 $check = new XmlTypeCheck( $filename, array( $this, 'checkSvgScriptCallback' ) );
 862                 return $check->filterMatch;
 863         }
 864
 865         /**
 866          * @todo Replace this with a whitelist filter!
 867          */
 868         public function checkSvgScriptCallback( $element, $attribs ) {
 869                 $stripped = $this->stripXmlNamespace( $element );
 870
 871                 if( $stripped == 'script' ) {
 872                         wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
 873                         return true;
 874                 }
 875
 876                 foreach( $attribs as $attrib => $value ) {
 877                         $stripped = $this->stripXmlNamespace( $attrib );
 878                         if( substr( $stripped, 0, 2 ) == 'on' ) {
 879                                 wfDebug( __METHOD__ . ": Found script attribute '$attrib'='value' in uploaded file.\n" );
 880                                 return true;
 881                         }
 882                         if( $stripped == 'href' && strpos( strtolower( $value ), 'javascript:' ) !== false ) {
 883                                 wfDebug( __METHOD__ . ": Found script href attribute '$attrib'='$value' in uploaded file.\n" );
 884                                 return true;
 885                         }
 886                 }
 887         }
 888
 889         private function stripXmlNamespace( $name ) {
 890                 // 'http://www.w3.org/2000/svg:script' -> 'script'
 891                 $parts = explode( ':', strtolower( $name ) );
 892                 return array_pop( $parts );
 893         }
 894
 895         /**
 896          * Generic wrapper function for a virus scanner program.
 897          * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
 898          * $wgAntivirusRequired may be used to deny upload if the scan fails.
 899          *
 900          * @param $file String: pathname to the temporary upload file
 901          * @return mixed false if not virus is found, NULL if the scan fails or is disabled,
 902          *         or a string containing feedback from the virus scanner if a virus was found.
 903          *         If textual feedback is missing but a virus was found, this function returns true.
 904          */
 905         public static function detectVirus( $file ) {
 906                 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
 907
 908                 if ( !$wgAntivirus ) {
 909                         wfDebug( __METHOD__ . ": virus scanner disabled\n" );
 910                         return null;
 911                 }
 912
 913                 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
 914                         wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
 915                         $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
 916                                 array( 'virus-badscanner', $wgAntivirus ) );
 917                         return wfMsg( 'virus-unknownscanner' ) . " $wgAntivirus";
 918                 }
 919
 920                 # look up scanner configuration
 921                 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
 922                 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
 923                 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
 924                         $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
 925
 926                 if ( strpos( $command, "%f" ) === false ) {
 927                         # simple pattern: append file to scan
 928                         $command .= " " . wfEscapeShellArg( $file );
 929                 } else {
 930                         # complex pattern: replace "%f" with file to scan
 931                         $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
 932                 }
 933
 934                 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
 935
 936                 # execute virus scanner
 937                 $exitCode = false;
 938
 939                 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
 940                 #      that does not seem to be worth the pain.
 941                 #      Ask me (Duesentrieb) about it if it's ever needed.
 942                 $output = wfShellExec( "$command 2>&1", $exitCode );
 943
 944                 # map exit code to AV_xxx constants.
 945                 $mappedCode = $exitCode;
 946                 if ( $exitCodeMap ) {
 947                         if ( isset( $exitCodeMap[$exitCode] ) ) {
 948                                 $mappedCode = $exitCodeMap[$exitCode];
 949                         } elseif ( isset( $exitCodeMap["*"] ) ) {
 950                                 $mappedCode = $exitCodeMap["*"];
 951                         }
 952                 }
 953
 954                 if ( $mappedCode === AV_SCAN_FAILED ) {
 955                         # scan failed (code was mapped to false by $exitCodeMap)
 956                         wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
 957
 958                         if ( $wgAntivirusRequired ) {
 959                                 return wfMsg( 'virus-scanfailed', array( $exitCode ) );
 960                         } else {
 961                                 return null;
 962                         }
 963                 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
 964                         # scan failed because filetype is unknown (probably imune)
 965                         wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
 966                         return null;
 967                 } elseif ( $mappedCode === AV_NO_VIRUS ) {
 968                         # no virus found
 969                         wfDebug( __METHOD__ . ": file passed virus scan.\n" );
 970                         return false;
 971                 } else {
 972                         $output = trim( $output );
 973
 974                         if ( !$output ) {
 975                                 $output = true; #if there's no output, return true
 976                         } elseif ( $msgPattern ) {
 977                                 $groups = array();
 978                                 if ( preg_match( $msgPattern, $output, $groups ) ) {
 979                                         if ( $groups[1] ) {
 980                                                 $output = $groups[1];
 981                                         }
 982                                 }
 983                         }
 984
 985                         wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
 986                         return $output;
 987                 }
 988         }
 989
 990         /**
 991          * Check if the temporary file is MacBinary-encoded, as some uploads
 992          * from Internet Explorer on Mac OS Classic and Mac OS X will be.
 993          * If so, the data fork will be extracted to a second temporary file,
 994          * which will then be checked for validity and either kept or discarded.
 995          */
 996         private function checkMacBinary() {
 997                 $macbin = new MacBinary( $this->mTempPath );
 998                 if( $macbin->isValid() ) {
 999                         $dataFile = tempnam( wfTempDir(), 'WikiMacBinary' );
1000                         $dataHandle = fopen( $dataFile, 'wb' );
1001
1002                         wfDebug( __METHOD__ . ": Extracting MacBinary data fork to $dataFile\n" );
1003                         $macbin->extractData( $dataHandle );
1004
1005                         $this->mTempPath = $dataFile;
1006                         $this->mFileSize = $macbin->dataForkLength();
1007
1008                         // We'll have to manually remove the new file if it's not kept.
1009                         $this->mRemoveTempFile = true;
1010                 }
1011                 $macbin->close();
1012         }
1013
1014         /**
1015          * Check if there's an overwrite conflict and, if so, if restrictions
1016          * forbid this user from performing the upload.
1017          *
1018          * @return mixed true on success, error string on failure
1019          */
1020         private function checkOverwrite( $user ) {
1021                 // First check whether the local file can be overwritten
1022                 $file = $this->getLocalFile();
1023                 if( $file->exists() ) {
1024                         if( !self::userCanReUpload( $user, $file ) ) {
1025                                 return 'fileexists-forbidden';
1026                         } else {
1027                                 return true;
1028                         }
1029                 }
1030
1031                 /* Check shared conflicts: if the local file does not exist, but
1032                  * wfFindFile finds a file, it exists in a shared repository.
1033                  */
1034                 $file = wfFindFile( $this->getTitle() );
1035                 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1036                         return 'fileexists-shared-forbidden';
1037                 }
1038
1039                 return true;
1040         }
1041
1042         /**
1043          * Check if a user is the last uploader
1044          *
1045          * @param $user User object
1046          * @param $img String: image name
1047          * @return Boolean
1048          */
1049         public static function userCanReUpload( User $user, $img ) {
1050                 if( $user->isAllowed( 'reupload' ) ) {
1051                         return true; // non-conditional
1052                 }
1053                 if( !$user->isAllowed( 'reupload-own' ) ) {
1054                         return false;
1055                 }
1056                 if( is_string( $img ) ) {
1057                         $img = wfLocalFile( $img );
1058                 }
1059                 if ( !( $img instanceof LocalFile ) ) {
1060                         return false;
1061                 }
1062
1063                 return $user->getId() == $img->getUser( 'id' );
1064         }
1065
1066         /**
1067          * Helper function that does various existence checks for a file.
1068          * The following checks are performed:
1069          * - The file exists
1070          * - Article with the same name as the file exists
1071          * - File exists with normalized extension
1072          * - The file looks like a thumbnail and the original exists
1073          *
1074          * @param $file The File object to check
1075          * @return mixed False if the file does not exists, else an array
1076          */
1077         public static function getExistsWarning( $file ) {
1078                 if( $file->exists() ) {
1079                         return array( 'warning' => 'exists', 'file' => $file );
1080                 }
1081
1082                 if( $file->getTitle()->getArticleID() ) {
1083                         return array( 'warning' => 'page-exists', 'file' => $file );
1084                 }
1085
1086                 if ( $file->wasDeleted() && !$file->exists() ) {
1087                         return array( 'warning' => 'was-deleted', 'file' => $file );
1088                 }
1089
1090                 if( strpos( $file->getName(), '.' ) == false ) {
1091                         $partname = $file->getName();
1092                         $extension = '';
1093                 } else {
1094                         $n = strrpos( $file->getName(), '.' );
1095                         $extension = substr( $file->getName(), $n + 1 );
1096                         $partname = substr( $file->getName(), 0, $n );
1097                 }
1098                 $normalizedExtension = File::normalizeExtension( $extension );
1099
1100                 if ( $normalizedExtension != $extension ) {
1101                         // We're not using the normalized form of the extension.
1102                         // Normal form is lowercase, using most common of alternate
1103                         // extensions (eg 'jpg' rather than 'JPEG').
1104                         //
1105                         // Check for another file using the normalized form...
1106                         $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
1107                         $file_lc = wfLocalFile( $nt_lc );
1108
1109                         if( $file_lc->exists() ) {
1110                                 return array(
1111                                         'warning' => 'exists-normalized',
1112                                         'file' => $file,
1113                                         'normalizedFile' => $file_lc
1114                                 );
1115                         }
1116                 }
1117
1118                 if ( self::isThumbName( $file->getName() ) ) {
1119                         # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1120                         $nt_thb = Title::newFromText( substr( $partname , strpos( $partname , '-' ) +1 ) . '.' . $extension, NS_FILE );
1121                         $file_thb = wfLocalFile( $nt_thb );
1122                         if( $file_thb->exists() ) {
1123                                 return array(
1124                                         'warning' => 'thumb',
1125                                         'file' => $file,
1126                                         'thumbFile' => $file_thb
1127                                 );
1128                         } else {
1129                                 // File does not exist, but we just don't like the name
1130                                 return array(
1131                                         'warning' => 'thumb-name',
1132                                         'file' => $file,
1133                                         'thumbFile' => $file_thb
1134                                 );
1135                         }
1136                 }
1137
1138
1139                 foreach( self::getFilenamePrefixBlacklist() as $prefix ) {
1140                         if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1141                                 return array(
1142                                         'warning' => 'bad-prefix',
1143                                         'file' => $file,
1144                                         'prefix' => $prefix
1145                                 );
1146                         }
1147                 }
1148
1149                 return false;
1150         }
1151
1152         /**
1153          * Helper function that checks whether the filename looks like a thumbnail
1154          */
1155         public static function isThumbName( $filename ) {
1156                 $n = strrpos( $filename, '.' );
1157                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
1158                 return (
1159                                         substr( $partname , 3, 3 ) == 'px-' ||
1160                                         substr( $partname , 2, 3 ) == 'px-'
1161                                 ) &&
1162                                 preg_match( "/[0-9]{2}/" , substr( $partname , 0, 2 ) );
1163         }
1164
1165         /**
1166          * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
1167          *
1168          * @return array list of prefixes
1169          */
1170         public static function getFilenamePrefixBlacklist() {
1171                 $blacklist = array();
1172                 $message = wfMsgForContent( 'filename-prefix-blacklist' );
1173                 if( $message && !( wfEmptyMsg( 'filename-prefix-blacklist', $message ) || $message == '-' ) ) {
1174                         $lines = explode( "\n", $message );
1175                         foreach( $lines as $line ) {
1176                                 // Remove comment lines
1177                                 $comment = substr( trim( $line ), 0, 1 );
1178                                 if ( $comment == '#' || $comment == '' ) {
1179                                         continue;
1180                                 }
1181                                 // Remove additional comments after a prefix
1182                                 $comment = strpos( $line, '#' );
1183                                 if ( $comment > 0 ) {
1184                                         $line = substr( $line, 0, $comment-1 );
1185                                 }
1186                                 $blacklist[] = trim( $line );
1187                         }
1188                 }
1189                 return $blacklist;
1190         }
1191
1192         public function getImageInfo( $result ) {
1193                 $file = $this->getLocalFile();
1194                 $imParam = ApiQueryImageInfo::getPropertyNames();
1195                 return ApiQueryImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1196         }
1197
1198         public function convertVerifyErrorToStatus( $error ) {
1199                 $code = $error['status'];
1200                 unset( $code['status'] );
1201                 return Status::newFatal( $this->getVerificationErrorCode( $code ), $error );
1202         }
1203 }