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 true if the user can use this upload module or else a string
  82          * identifying the missing permission.
  83          * Can be overriden by subclasses.
  84          */
  85         public static function isAllowed( $user ) {
  86                 foreach ( array( 'upload', 'edit' ) as $permission ) {
  87                         if ( !$user->isAllowed( $permission ) ) {
  88                                 return $permission;
  89                         }
  90                 }
  91                 return true;
  92         }
  93
  94         // Upload handlers. Should probably just be a global.
  95         static $uploadHandlers = array( 'Stash', 'File', 'Url' );
  96
  97         /**
  98          * Create a form of UploadBase depending on wpSourceType and initializes it
  99          */
 100         public static function createFromRequest( &$request, $type = null ) {
 101                 $type = $type ? $type : $request->getVal( 'wpSourceType', 'File' );
 102
 103                 if( !$type ) {
 104                         return null;
 105                 }
 106
 107                 // Get the upload class
 108                 $type = ucfirst( $type );
 109
 110                 // Give hooks the chance to handle this request
 111                 $className = null;
 112                 wfRunHooks( 'UploadCreateFromRequest', array( $type, &$className ) );
 113                 if ( is_null( $className ) ) {
 114                         $className = 'UploadFrom' . $type;
 115                         wfDebug( __METHOD__ . ": class name: $className\n" );
 116                         if( !in_array( $type, self::$uploadHandlers ) ) {
 117                                 return null;
 118                         }
 119                 }
 120
 121                 // Check whether this upload class is enabled
 122                 if( !call_user_func( array( $className, 'isEnabled' ) ) ) {
 123                         return null;
 124                 }
 125
 126                 // Check whether the request is valid
 127                 if( !call_user_func( array( $className, 'isValidRequest' ), $request ) ) {
 128                         return null;
 129                 }
 130
 131                 $handler = new $className;
 132
 133                 $handler->initializeFromRequest( $request );
 134                 return $handler;
 135         }
 136
 137         /**
 138          * Check whether a request if valid for this handler
 139          */
 140         public static function isValidRequest( $request ) {
 141                 return false;
 142         }
 143
 144         public function __construct() {}
 145
 146         /**
 147          * Returns the upload type. Should be overridden by child classes
 148          *
 149          * @since 1.18
 150          * @return string
 151          */
 152         public function getSourceType() { return null; }
 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                 $maxSize = self::getMaxUploadSize( $this->getSourceType() );
 238                 if( $this->mFileSize > $maxSize ) {
 239                         return array(
 240                                 'status' => self::FILE_TOO_LARGE,
 241                                 'max' => $maxSize,
 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                                 if ( count( $this->mBlackListedExtensions ) ) {
 291                                         $result['blacklistedExt'] = $this->mBlackListedExtensions;
 292                                 }
 293                         }
 294                         return $result;
 295                 }
 296                 $this->mDestName = $this->getLocalFile()->getName();
 297
 298                 return true;
 299         }
 300
 301         /**
 302          * Verify the mime type
 303          *
 304          * @param $mime string representing the mime
 305          * @return mixed true if the file is verified, an array otherwise
 306          */
 307         protected function verifyMimeType( $mime ) {
 308                 global $wgVerifyMimeType;
 309                 if ( $wgVerifyMimeType ) {
 310                         wfDebug ( "\n\nmime: <$mime> extension: <{$this->mFinalExtension}>\n\n");
 311                         global $wgMimeTypeBlacklist;
 312                         if ( $this->checkFileExtension( $mime, $wgMimeTypeBlacklist ) ) {
 313                                 return array( 'filetype-badmime', $mime );
 314                         }
 315
 316                         # XXX: Missing extension will be caught by validateName() via getTitle()
 317                         if ( $this->mFinalExtension != '' && !$this->verifyExtension( $mime, $this->mFinalExtension ) ) {
 318                                 return array( 'filetype-mime-mismatch', $this->mFinalExtension, $mime );
 319                         }
 320
 321                         # Check IE type
 322                         $fp = fopen( $this->mTempPath, 'rb' );
 323                         $chunk = fread( $fp, 256 );
 324                         fclose( $fp );
 325
 326                         $magic = MimeMagic::singleton();
 327                         $extMime = $magic->guessTypesForExtension( $this->mFinalExtension );
 328                         $ieTypes = $magic->getIEMimeTypes( $this->mTempPath, $chunk, $extMime );
 329                         foreach ( $ieTypes as $ieType ) {
 330                                 if ( $this->checkFileExtension( $ieType, $wgMimeTypeBlacklist ) ) {
 331                                         return array( 'filetype-bad-ie-mime', $ieType );
 332                                 }
 333                         }
 334                 }
 335
 336                 return true;
 337         }
 338
 339         /**
 340          * Verifies that it's ok to include the uploaded file
 341          *
 342          * @return mixed true of the file is verified, array otherwise.
 343          */
 344         protected function verifyFile() {
 345                 # get the title, even though we are doing nothing with it, because
 346                 # we need to populate mFinalExtension
 347                 $this->getTitle();
 348
 349                 $this->mFileProps = File::getPropsFromPath( $this->mTempPath, $this->mFinalExtension );
 350                 $this->checkMacBinary();
 351
 352                 # check mime type, if desired
 353                 $mime = $this->mFileProps[ 'file-mime' ];
 354                 $status = $this->verifyMimeType( $mime );
 355                 if ( $status !== true ) {
 356                         return $status;
 357                 }
 358
 359                 # check for htmlish code and javascript
 360                 if( self::detectScript( $this->mTempPath, $mime, $this->mFinalExtension ) ) {
 361                         return array( 'uploadscripted' );
 362                 }
 363                 if( $this->mFinalExtension == 'svg' || $mime == 'image/svg+xml' ) {
 364                         if( $this->detectScriptInSvg( $this->mTempPath ) ) {
 365                                 return array( 'uploadscripted' );
 366                         }
 367                 }
 368
 369                 /**
 370                 * Scan the uploaded file for viruses
 371                 */
 372                 $virus = $this->detectVirus( $this->mTempPath );
 373                 if ( $virus ) {
 374                         return array( 'uploadvirus', $virus );
 375                 }
 376
 377                 $handler = MediaHandler::getHandler( $mime );
 378                 if ( $handler ) {
 379                         $handlerStatus = $handler->verifyUpload( $this->mTempPath );
 380                         if ( !$handlerStatus->isOK() ) {
 381                                 $errors = $handlerStatus->getErrorsArray();
 382                                 return reset( $errors );
 383                         }
 384                 }
 385
 386                 wfRunHooks( 'UploadVerifyFile', array( $this, $mime, &$status ) );
 387                 if ( $status !== true ) {
 388                         return $status;
 389                 }
 390
 391                 wfDebug( __METHOD__ . ": all clear; passing.\n" );
 392                 return true;
 393         }
 394
 395         /**
 396          * Check whether the user can edit, upload and create the image. This
 397          * checks only against the current title; if it returns errors, it may
 398          * very well be that another title will not give errors. Therefore
 399          * isAllowed() should be called as well for generic is-user-blocked or
 400          * can-user-upload checking.
 401          *
 402          * @param $user the User object to verify the permissions against
 403          * @return mixed An array as returned by getUserPermissionsErrors or true
 404          *               in case the user has proper permissions.
 405          */
 406         public function verifyPermissions( $user ) {
 407                 /**
 408                  * If the image is protected, non-sysop users won't be able
 409                  * to modify it by uploading a new revision.
 410                  */
 411                 $nt = $this->getTitle();
 412                 if( is_null( $nt ) ) {
 413                         return true;
 414                 }
 415                 $permErrors = $nt->getUserPermissionsErrors( 'edit', $user );
 416                 $permErrorsUpload = $nt->getUserPermissionsErrors( 'upload', $user );
 417                 if ( $nt->exists() ) {
 418                         $permErrorsCreate = $nt->getUserPermissionsErrors( 'createpage', $user );
 419                 } else {
 420                         $permErrorsCreate = array();
 421                 }
 422                 if( $permErrors || $permErrorsUpload || $permErrorsCreate ) {
 423                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsUpload, $permErrors ) );
 424                         $permErrors = array_merge( $permErrors, wfArrayDiff2( $permErrorsCreate, $permErrors ) );
 425                         return $permErrors;
 426                 }
 427
 428                 $overwriteError = $this->checkOverwrite( $user );
 429                 if ( $overwriteError !== true ) {
 430                         return array( array( $overwriteError ) );
 431                 }
 432
 433                 return true;
 434         }
 435
 436         /**
 437          * Check for non fatal problems with the file
 438          *
 439          * @return Array of warnings
 440          */
 441         public function checkWarnings() {
 442                 $warnings = array();
 443
 444                 $localFile = $this->getLocalFile();
 445                 $filename = $localFile->getName();
 446                 $n = strrpos( $filename, '.' );
 447
 448                 /**
 449                  * Check whether the resulting filename is different from the desired one,
 450                  * but ignore things like ucfirst() and spaces/underscore things
 451                  */
 452                 $comparableName = str_replace( ' ', '_', $this->mDesiredDestName );
 453                 $comparableName = Title::capitalize( $comparableName, NS_FILE );
 454
 455                 if( $this->mDesiredDestName != $filename && $comparableName != $filename ) {
 456                         $warnings['badfilename'] = $filename;
 457                 }
 458
 459                 // Check whether the file extension is on the unwanted list
 460                 global $wgCheckFileExtensions, $wgFileExtensions;
 461                 if ( $wgCheckFileExtensions ) {
 462                         if ( !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) ) {
 463                                 $warnings['filetype-unwanted-type'] = $this->mFinalExtension;
 464                         }
 465                 }
 466
 467                 global $wgUploadSizeWarning;
 468                 if ( $wgUploadSizeWarning && ( $this->mFileSize > $wgUploadSizeWarning ) ) {
 469                         $warnings['large-file'] = $wgUploadSizeWarning;
 470                 }
 471
 472                 if ( $this->mFileSize == 0 ) {
 473                         $warnings['emptyfile'] = true;
 474                 }
 475
 476                 $exists = self::getExistsWarning( $localFile );
 477                 if( $exists !== false ) {
 478                         $warnings['exists'] = $exists;
 479                 }
 480
 481                 // Check dupes against existing files
 482                 $hash = File::sha1Base36( $this->mTempPath );
 483                 $dupes = RepoGroup::singleton()->findBySha1( $hash );
 484                 $title = $this->getTitle();
 485                 // Remove all matches against self
 486                 foreach ( $dupes as $key => $dupe ) {
 487                         if( $title->equals( $dupe->getTitle() ) ) {
 488                                 unset( $dupes[$key] );
 489                         }
 490                 }
 491                 if( $dupes ) {
 492                         $warnings['duplicate'] = $dupes;
 493                 }
 494
 495                 // Check dupes against archives
 496                 $archivedImage = new ArchivedFile( null, 0, "{$hash}.{$this->mFinalExtension}" );
 497                 if ( $archivedImage->getID() > 0 ) {
 498                         $warnings['duplicate-archive'] = $archivedImage->getName();
 499                 }
 500
 501                 return $warnings;
 502         }
 503
 504         /**
 505          * Really perform the upload. Stores the file in the local repo, watches
 506          * if necessary and runs the UploadComplete hook.
 507          *
 508          * @return mixed Status indicating the whether the upload succeeded.
 509          */
 510         public function performUpload( $comment, $pageText, $watch, $user ) {
 511                 $status = $this->getLocalFile()->upload(
 512                         $this->mTempPath,
 513                         $comment,
 514                         $pageText,
 515                         File::DELETE_SOURCE,
 516                         $this->mFileProps,
 517                         false,
 518                         $user
 519                 );
 520
 521                 if( $status->isGood() ) {
 522                         if ( $watch ) {
 523                                 $user->addWatch( $this->getLocalFile()->getTitle() );
 524                         }
 525
 526                         wfRunHooks( 'UploadComplete', array( &$this ) );
 527                 }
 528
 529                 return $status;
 530         }
 531
 532         /**
 533          * Returns the title of the file to be uploaded. Sets mTitleError in case
 534          * the name was illegal.
 535          *
 536          * @return Title The title of the file or null in case the name was illegal
 537          */
 538         public function getTitle() {
 539                 if ( $this->mTitle !== false ) {
 540                         return $this->mTitle;
 541                 }
 542
 543                 /**
 544                  * Chop off any directories in the given filename. Then
 545                  * filter out illegal characters, and try to make a legible name
 546                  * out of it. We'll strip some silently that Title would die on.
 547                  */
 548                 $this->mFilteredName = wfStripIllegalFilenameChars( $this->mDesiredDestName );
 549                 /* Normalize to title form before we do any further processing */
 550                 $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 551                 if( is_null( $nt ) ) {
 552                         $this->mTitleError = self::ILLEGAL_FILENAME;
 553                         return $this->mTitle = null;
 554                 }
 555                 $this->mFilteredName = $nt->getDBkey();
 556
 557                 /**
 558                  * We'll want to blacklist against *any* 'extension', and use
 559                  * only the final one for the whitelist.
 560                  */
 561                 list( $partname, $ext ) = $this->splitExtensions( $this->mFilteredName );
 562
 563                 if( count( $ext ) ) {
 564                         $this->mFinalExtension = trim( $ext[count( $ext ) - 1] );
 565                 } else {
 566                         $this->mFinalExtension = '';
 567
 568                         # No extension, try guessing one
 569                         $magic = MimeMagic::singleton();
 570                         $mime = $magic->guessMimeType( $this->mTempPath );
 571                         if ( $mime !== 'unknown/unknown' ) {
 572                                 # Get a space separated list of extensions
 573                                 $extList = $magic->getExtensionsForType( $mime );
 574                                 if ( $extList ) {
 575                                         # Set the extension to the canonical extension
 576                                         $this->mFinalExtension = strtok( $extList, ' ' );
 577
 578                                         # Fix up the other variables
 579                                         $this->mFilteredName .= ".{$this->mFinalExtension}";
 580                                         $nt = Title::makeTitleSafe( NS_FILE, $this->mFilteredName );
 581                                         $ext = array( $this->mFinalExtension );
 582                                 }
 583                         }
 584
 585                 }
 586
 587                 /* Don't allow users to override the blacklist (check file extension) */
 588                 global $wgCheckFileExtensions, $wgStrictFileExtensions;
 589                 global $wgFileExtensions, $wgFileBlacklist;
 590
 591                 $blackListedExtensions = $this->checkFileExtensionList( $ext, $wgFileBlacklist );
 592
 593                 if ( $this->mFinalExtension == '' ) {
 594                         $this->mTitleError = self::FILETYPE_MISSING;
 595                         return $this->mTitle = null;
 596                 } elseif ( $blackListedExtensions ||
 597                                 ( $wgCheckFileExtensions && $wgStrictFileExtensions &&
 598                                         !$this->checkFileExtension( $this->mFinalExtension, $wgFileExtensions ) ) ) {
 599                         $this->mBlackListedExtensions = $blackListedExtensions;
 600                         $this->mTitleError = self::FILETYPE_BADTYPE;
 601                         return $this->mTitle = null;
 602                 }
 603
 604                 # If there was more than one "extension", reassemble the base
 605                 # filename to prevent bogus complaints about length
 606                 if( count( $ext ) > 1 ) {
 607                         for( $i = 0; $i < count( $ext ) - 1; $i++ ) {
 608                                 $partname .= '.' . $ext[$i];
 609                         }
 610                 }
 611
 612                 if( strlen( $partname ) < 1 ) {
 613                         $this->mTitleError =  self::MIN_LENGTH_PARTNAME;
 614                         return $this->mTitle = null;
 615                 }
 616
 617                 return $this->mTitle = $nt;
 618         }
 619
 620         /**
 621          * Return the local file and initializes if necessary.
 622          */
 623         public function getLocalFile() {
 624                 if( is_null( $this->mLocalFile ) ) {
 625                         $nt = $this->getTitle();
 626                         $this->mLocalFile = is_null( $nt ) ? null : wfLocalFile( $nt );
 627                 }
 628                 return $this->mLocalFile;
 629         }
 630
 631         /**
 632          * NOTE: Probably should be deprecated in favor of UploadStash, but this is sometimes
 633          * called outside that context.
 634          *
 635          * Stash a file in a temporary directory for later processing
 636          * after the user has confirmed it.
 637          *
 638          * If the user doesn't explicitly cancel or accept, these files
 639          * can accumulate in the temp directory.
 640          *
 641          * @param $saveName String: the destination filename
 642          * @param $tempSrc String: the source temporary file to save
 643          * @return String: full path the stashed file, or false on failure
 644          */
 645         protected function saveTempUploadedFile( $saveName, $tempSrc ) {
 646                 $repo = RepoGroup::singleton()->getLocalRepo();
 647                 $status = $repo->storeTemp( $saveName, $tempSrc );
 648                 return $status;
 649         }
 650
 651         /**
 652          * If the user does not supply all necessary information in the first upload form submission (either by accident or
 653          * by design) then we may want to stash the file temporarily, get more information, and publish the file later.
 654          *
 655          * This method will stash a file in a temporary directory for later processing, and save the necessary descriptive info
 656          * into the user's session.
 657          * This method returns the file object, which also has a 'sessionKey' property which can be passed through a form or
 658          * API request to find this stashed file again.
 659          *
 660          * @param $key String: (optional) the session key used to find the file info again. If not supplied, a key will be autogenerated.
 661          * @return File: stashed file
 662          */
 663         public function stashSessionFile( $key = null ) {
 664                 $stash = RepoGroup::singleton()->getLocalRepo()->getUploadStash();
 665                 $data = array(
 666                         'mFileProps' => $this->mFileProps,
 667                         'mSourceType' => $this->getSourceType(),
 668                 );
 669                 $file = $stash->stashFile( $this->mTempPath, $data, $key );
 670                 $this->mLocalFile = $file;
 671                 return $file;
 672         }
 673
 674         /**
 675          * Stash a file in a temporary directory, returning a key which can be used to find the file again. See stashSessionFile().
 676          *
 677          * @param $key String: (optional) the session key used to find the file info again. If not supplied, a key will be autogenerated.
 678          * @return String: session key
 679          */
 680         public function stashSession( $key = null ) {
 681                 return $this->stashSessionFile( $key )->getSessionKey();
 682         }
 683
 684         /**
 685          * If we've modified the upload file we need to manually remove it
 686          * on exit to clean up.
 687          */
 688         public function cleanupTempFile() {
 689                 if ( $this->mRemoveTempFile && $this->mTempPath && file_exists( $this->mTempPath ) ) {
 690                         wfDebug( __METHOD__ . ": Removing temporary file {$this->mTempPath}\n" );
 691                         unlink( $this->mTempPath );
 692                 }
 693         }
 694
 695         public function getTempPath() {
 696                 return $this->mTempPath;
 697         }
 698
 699         /**
 700          * Split a file into a base name and all dot-delimited 'extensions'
 701          * on the end. Some web server configurations will fall back to
 702          * earlier pseudo-'extensions' to determine type and execute
 703          * scripts, so the blacklist needs to check them all.
 704          *
 705          * @return array
 706          */
 707         public static function splitExtensions( $filename ) {
 708                 $bits = explode( '.', $filename );
 709                 $basename = array_shift( $bits );
 710                 return array( $basename, $bits );
 711         }
 712
 713         /**
 714          * Perform case-insensitive match against a list of file extensions.
 715          * Returns true if the extension is in the list.
 716          *
 717          * @param $ext String
 718          * @param $list Array
 719          * @return Boolean
 720          */
 721         public static function checkFileExtension( $ext, $list ) {
 722                 return in_array( strtolower( $ext ), $list );
 723         }
 724
 725         /**
 726          * Perform case-insensitive match against a list of file extensions.
 727          * Returns an array of matching extensions.
 728          *
 729          * @param $ext Array
 730          * @param $list Array
 731          * @return Boolean
 732          */
 733         public static function checkFileExtensionList( $ext, $list ) {
 734                 return array_intersect( array_map( 'strtolower', $ext ), $list );
 735         }
 736
 737         /**
 738          * Checks if the mime type of the uploaded file matches the file extension.
 739          *
 740          * @param $mime String: the mime type of the uploaded file
 741          * @param $extension String: the filename extension that the file is to be served with
 742          * @return Boolean
 743          */
 744         public static function verifyExtension( $mime, $extension ) {
 745                 $magic = MimeMagic::singleton();
 746
 747                 if ( !$mime || $mime == 'unknown' || $mime == 'unknown/unknown' )
 748                         if ( !$magic->isRecognizableExtension( $extension ) ) {
 749                                 wfDebug( __METHOD__ . ": passing file with unknown detected mime type; " .
 750                                         "unrecognized extension '$extension', can't verify\n" );
 751                                 return true;
 752                         } else {
 753                                 wfDebug( __METHOD__ . ": rejecting file with unknown detected mime type; ".
 754                                         "recognized extension '$extension', so probably invalid file\n" );
 755                                 return false;
 756                         }
 757
 758                 $match = $magic->isMatchingExtension( $extension, $mime );
 759
 760                 if ( $match === null ) {
 761                         wfDebug( __METHOD__ . ": no file extension known for mime type $mime, passing file\n" );
 762                         return true;
 763                 } elseif( $match === true ) {
 764                         wfDebug( __METHOD__ . ": mime type $mime matches extension $extension, passing file\n" );
 765
 766                         #TODO: if it's a bitmap, make sure PHP or ImageMagic resp. can handle it!
 767                         return true;
 768
 769                 } else {
 770                         wfDebug( __METHOD__ . ": mime type $mime mismatches file extension $extension, rejecting file\n" );
 771                         return false;
 772                 }
 773         }
 774
 775         /**
 776          * Heuristic for detecting files that *could* contain JavaScript instructions or
 777          * things that may look like HTML to a browser and are thus
 778          * potentially harmful. The present implementation will produce false
 779          * positives in some situations.
 780          *
 781          * @param $file String: pathname to the temporary upload file
 782          * @param $mime String: the mime type of the file
 783          * @param $extension String: the extension of the file
 784          * @return Boolean: true if the file contains something looking like embedded scripts
 785          */
 786         public static function detectScript( $file, $mime, $extension ) {
 787                 global $wgAllowTitlesInSVG;
 788
 789                 # ugly hack: for text files, always look at the entire file.
 790                 # For binary field, just check the first K.
 791
 792                 if( strpos( $mime,'text/' ) === 0 ) {
 793                         $chunk = file_get_contents( $file );
 794                 } else {
 795                         $fp = fopen( $file, 'rb' );
 796                         $chunk = fread( $fp, 1024 );
 797                         fclose( $fp );
 798                 }
 799
 800                 $chunk = strtolower( $chunk );
 801
 802                 if( !$chunk ) {
 803                         return false;
 804                 }
 805
 806                 # decode from UTF-16 if needed (could be used for obfuscation).
 807                 if( substr( $chunk, 0, 2 ) == "\xfe\xff" ) {
 808                         $enc = 'UTF-16BE';
 809                 } elseif( substr( $chunk, 0, 2 ) == "\xff\xfe" ) {
 810                         $enc = 'UTF-16LE';
 811                 } else {
 812                         $enc = null;
 813                 }
 814
 815                 if( $enc ) {
 816                         $chunk = iconv( $enc, "ASCII//IGNORE", $chunk );
 817                 }
 818
 819                 $chunk = trim( $chunk );
 820
 821                 # FIXME: convert from UTF-16 if necessarry!
 822                 wfDebug( __METHOD__ . ": checking for embedded scripts and HTML stuff\n" );
 823
 824                 # check for HTML doctype
 825                 if ( preg_match( "/<!DOCTYPE *X?HTML/i", $chunk ) ) {
 826                         return true;
 827                 }
 828
 829                 /**
 830                  * Internet Explorer for Windows performs some really stupid file type
 831                  * autodetection which can cause it to interpret valid image files as HTML
 832                  * and potentially execute JavaScript, creating a cross-site scripting
 833                  * attack vectors.
 834                  *
 835                  * Apple's Safari browser also performs some unsafe file type autodetection
 836                  * which can cause legitimate files to be interpreted as HTML if the
 837                  * web server is not correctly configured to send the right content-type
 838                  * (or if you're really uploading plain text and octet streams!)
 839                  *
 840                  * Returns true if IE is likely to mistake the given file for HTML.
 841                  * Also returns true if Safari would mistake the given file for HTML
 842                  * when served with a generic content-type.
 843                  */
 844                 $tags = array(
 845                         '<a href',
 846                         '<body',
 847                         '<head',
 848                         '<html',   #also in safari
 849                         '<img',
 850                         '<pre',
 851                         '<script', #also in safari
 852                         '<table'
 853                 );
 854
 855                 if( !$wgAllowTitlesInSVG && $extension !== 'svg' && $mime !== 'image/svg' ) {
 856                         $tags[] = '<title';
 857                 }
 858
 859                 foreach( $tags as $tag ) {
 860                         if( false !== strpos( $chunk, $tag ) ) {
 861                                 return true;
 862                         }
 863                 }
 864
 865                 /*
 866                  * look for JavaScript
 867                  */
 868
 869                 # resolve entity-refs to look at attributes. may be harsh on big files... cache result?
 870                 $chunk = Sanitizer::decodeCharReferences( $chunk );
 871
 872                 # look for script-types
 873                 if( preg_match( '!type\s*=\s*[\'"]?\s*(?:\w*/)?(?:ecma|java)!sim', $chunk ) ) {
 874                         return true;
 875                 }
 876
 877                 # look for html-style script-urls
 878                 if( preg_match( '!(?:href|src|data)\s*=\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
 879                         return true;
 880                 }
 881
 882                 # look for css-style script-urls
 883                 if( preg_match( '!url\s*\(\s*[\'"]?\s*(?:ecma|java)script:!sim', $chunk ) ) {
 884                         return true;
 885                 }
 886
 887                 wfDebug( __METHOD__ . ": no scripts found\n" );
 888                 return false;
 889         }
 890
 891         protected function detectScriptInSvg( $filename ) {
 892                 $check = new XmlTypeCheck( $filename, array( $this, 'checkSvgScriptCallback' ) );
 893                 return $check->filterMatch;
 894         }
 895
 896         /**
 897          * @todo Replace this with a whitelist filter!
 898          */
 899         public function checkSvgScriptCallback( $element, $attribs ) {
 900                 $stripped = $this->stripXmlNamespace( $element );
 901
 902                 if( $stripped == 'script' ) {
 903                         wfDebug( __METHOD__ . ": Found script element '$element' in uploaded file.\n" );
 904                         return true;
 905                 }
 906
 907                 foreach( $attribs as $attrib => $value ) {
 908                         $stripped = $this->stripXmlNamespace( $attrib );
 909                         if( substr( $stripped, 0, 2 ) == 'on' ) {
 910                                 wfDebug( __METHOD__ . ": Found script attribute '$attrib'='value' in uploaded file.\n" );
 911                                 return true;
 912                         }
 913                         if( $stripped == 'href' && strpos( strtolower( $value ), 'javascript:' ) !== false ) {
 914                                 wfDebug( __METHOD__ . ": Found script href attribute '$attrib'='$value' in uploaded file.\n" );
 915                                 return true;
 916                         }
 917                 }
 918         }
 919
 920         private function stripXmlNamespace( $name ) {
 921                 // 'http://www.w3.org/2000/svg:script' -> 'script'
 922                 $parts = explode( ':', strtolower( $name ) );
 923                 return array_pop( $parts );
 924         }
 925
 926         /**
 927          * Generic wrapper function for a virus scanner program.
 928          * This relies on the $wgAntivirus and $wgAntivirusSetup variables.
 929          * $wgAntivirusRequired may be used to deny upload if the scan fails.
 930          *
 931          * @param $file String: pathname to the temporary upload file
 932          * @return mixed false if not virus is found, NULL if the scan fails or is disabled,
 933          *         or a string containing feedback from the virus scanner if a virus was found.
 934          *         If textual feedback is missing but a virus was found, this function returns true.
 935          */
 936         public static function detectVirus( $file ) {
 937                 global $wgAntivirus, $wgAntivirusSetup, $wgAntivirusRequired, $wgOut;
 938
 939                 if ( !$wgAntivirus ) {
 940                         wfDebug( __METHOD__ . ": virus scanner disabled\n" );
 941                         return null;
 942                 }
 943
 944                 if ( !$wgAntivirusSetup[$wgAntivirus] ) {
 945                         wfDebug( __METHOD__ . ": unknown virus scanner: $wgAntivirus\n" );
 946                         $wgOut->wrapWikiMsg( "<div class=\"error\">\n$1\n</div>",
 947                                 array( 'virus-badscanner', $wgAntivirus ) );
 948                         return wfMsg( 'virus-unknownscanner' ) . " $wgAntivirus";
 949                 }
 950
 951                 # look up scanner configuration
 952                 $command = $wgAntivirusSetup[$wgAntivirus]['command'];
 953                 $exitCodeMap = $wgAntivirusSetup[$wgAntivirus]['codemap'];
 954                 $msgPattern = isset( $wgAntivirusSetup[$wgAntivirus]['messagepattern'] ) ?
 955                         $wgAntivirusSetup[$wgAntivirus]['messagepattern'] : null;
 956
 957                 if ( strpos( $command, "%f" ) === false ) {
 958                         # simple pattern: append file to scan
 959                         $command .= " " . wfEscapeShellArg( $file );
 960                 } else {
 961                         # complex pattern: replace "%f" with file to scan
 962                         $command = str_replace( "%f", wfEscapeShellArg( $file ), $command );
 963                 }
 964
 965                 wfDebug( __METHOD__ . ": running virus scan: $command \n" );
 966
 967                 # execute virus scanner
 968                 $exitCode = false;
 969
 970                 # NOTE: there's a 50 line workaround to make stderr redirection work on windows, too.
 971                 #      that does not seem to be worth the pain.
 972                 #      Ask me (Duesentrieb) about it if it's ever needed.
 973                 $output = wfShellExec( "$command 2>&1", $exitCode );
 974
 975                 # map exit code to AV_xxx constants.
 976                 $mappedCode = $exitCode;
 977                 if ( $exitCodeMap ) {
 978                         if ( isset( $exitCodeMap[$exitCode] ) ) {
 979                                 $mappedCode = $exitCodeMap[$exitCode];
 980                         } elseif ( isset( $exitCodeMap["*"] ) ) {
 981                                 $mappedCode = $exitCodeMap["*"];
 982                         }
 983                 }
 984
 985                 if ( $mappedCode === AV_SCAN_FAILED ) {
 986                         # scan failed (code was mapped to false by $exitCodeMap)
 987                         wfDebug( __METHOD__ . ": failed to scan $file (code $exitCode).\n" );
 988
 989                         if ( $wgAntivirusRequired ) {
 990                                 return wfMsg( 'virus-scanfailed', array( $exitCode ) );
 991                         } else {
 992                                 return null;
 993                         }
 994                 } elseif ( $mappedCode === AV_SCAN_ABORTED ) {
 995                         # scan failed because filetype is unknown (probably imune)
 996                         wfDebug( __METHOD__ . ": unsupported file type $file (code $exitCode).\n" );
 997                         return null;
 998                 } elseif ( $mappedCode === AV_NO_VIRUS ) {
 999                         # no virus found
1000                         wfDebug( __METHOD__ . ": file passed virus scan.\n" );
1001                         return false;
1002                 } else {
1003                         $output = trim( $output );
1004
1005                         if ( !$output ) {
1006                                 $output = true; #if there's no output, return true
1007                         } elseif ( $msgPattern ) {
1008                                 $groups = array();
1009                                 if ( preg_match( $msgPattern, $output, $groups ) ) {
1010                                         if ( $groups[1] ) {
1011                                                 $output = $groups[1];
1012                                         }
1013                                 }
1014                         }
1015
1016                         wfDebug( __METHOD__ . ": FOUND VIRUS! scanner feedback: $output \n" );
1017                         return $output;
1018                 }
1019         }
1020
1021         /**
1022          * Check if the temporary file is MacBinary-encoded, as some uploads
1023          * from Internet Explorer on Mac OS Classic and Mac OS X will be.
1024          * If so, the data fork will be extracted to a second temporary file,
1025          * which will then be checked for validity and either kept or discarded.
1026          */
1027         private function checkMacBinary() {
1028                 $macbin = new MacBinary( $this->mTempPath );
1029                 if( $macbin->isValid() ) {
1030                         $dataFile = tempnam( wfTempDir(), 'WikiMacBinary' );
1031                         $dataHandle = fopen( $dataFile, 'wb' );
1032
1033                         wfDebug( __METHOD__ . ": Extracting MacBinary data fork to $dataFile\n" );
1034                         $macbin->extractData( $dataHandle );
1035
1036                         $this->mTempPath = $dataFile;
1037                         $this->mFileSize = $macbin->dataForkLength();
1038
1039                         // We'll have to manually remove the new file if it's not kept.
1040                         $this->mRemoveTempFile = true;
1041                 }
1042                 $macbin->close();
1043         }
1044
1045         /**
1046          * Check if there's an overwrite conflict and, if so, if restrictions
1047          * forbid this user from performing the upload.
1048          *
1049          * @return mixed true on success, error string on failure
1050          */
1051         private function checkOverwrite( $user ) {
1052                 // First check whether the local file can be overwritten
1053                 $file = $this->getLocalFile();
1054                 if( $file->exists() ) {
1055                         if( !self::userCanReUpload( $user, $file ) ) {
1056                                 return 'fileexists-forbidden';
1057                         } else {
1058                                 return true;
1059                         }
1060                 }
1061
1062                 /* Check shared conflicts: if the local file does not exist, but
1063                  * wfFindFile finds a file, it exists in a shared repository.
1064                  */
1065                 $file = wfFindFile( $this->getTitle() );
1066                 if ( $file && !$user->isAllowed( 'reupload-shared' ) ) {
1067                         return 'fileexists-shared-forbidden';
1068                 }
1069
1070                 return true;
1071         }
1072
1073         /**
1074          * Check if a user is the last uploader
1075          *
1076          * @param $user User object
1077          * @param $img String: image name
1078          * @return Boolean
1079          */
1080         public static function userCanReUpload( User $user, $img ) {
1081                 if( $user->isAllowed( 'reupload' ) ) {
1082                         return true; // non-conditional
1083                 }
1084                 if( !$user->isAllowed( 'reupload-own' ) ) {
1085                         return false;
1086                 }
1087                 if( is_string( $img ) ) {
1088                         $img = wfLocalFile( $img );
1089                 }
1090                 if ( !( $img instanceof LocalFile ) ) {
1091                         return false;
1092                 }
1093
1094                 return $user->getId() == $img->getUser( 'id' );
1095         }
1096
1097         /**
1098          * Helper function that does various existence checks for a file.
1099          * The following checks are performed:
1100          * - The file exists
1101          * - Article with the same name as the file exists
1102          * - File exists with normalized extension
1103          * - The file looks like a thumbnail and the original exists
1104          *
1105          * @param $file The File object to check
1106          * @return mixed False if the file does not exists, else an array
1107          */
1108         public static function getExistsWarning( $file ) {
1109                 if( $file->exists() ) {
1110                         return array( 'warning' => 'exists', 'file' => $file );
1111                 }
1112
1113                 if( $file->getTitle()->getArticleID() ) {
1114                         return array( 'warning' => 'page-exists', 'file' => $file );
1115                 }
1116
1117                 if ( $file->wasDeleted() && !$file->exists() ) {
1118                         return array( 'warning' => 'was-deleted', 'file' => $file );
1119                 }
1120
1121                 if( strpos( $file->getName(), '.' ) == false ) {
1122                         $partname = $file->getName();
1123                         $extension = '';
1124                 } else {
1125                         $n = strrpos( $file->getName(), '.' );
1126                         $extension = substr( $file->getName(), $n + 1 );
1127                         $partname = substr( $file->getName(), 0, $n );
1128                 }
1129                 $normalizedExtension = File::normalizeExtension( $extension );
1130
1131                 if ( $normalizedExtension != $extension ) {
1132                         // We're not using the normalized form of the extension.
1133                         // Normal form is lowercase, using most common of alternate
1134                         // extensions (eg 'jpg' rather than 'JPEG').
1135                         //
1136                         // Check for another file using the normalized form...
1137                         $nt_lc = Title::makeTitle( NS_FILE, "{$partname}.{$normalizedExtension}" );
1138                         $file_lc = wfLocalFile( $nt_lc );
1139
1140                         if( $file_lc->exists() ) {
1141                                 return array(
1142                                         'warning' => 'exists-normalized',
1143                                         'file' => $file,
1144                                         'normalizedFile' => $file_lc
1145                                 );
1146                         }
1147                 }
1148
1149                 if ( self::isThumbName( $file->getName() ) ) {
1150                         # Check for filenames like 50px- or 180px-, these are mostly thumbnails
1151                         $nt_thb = Title::newFromText( substr( $partname , strpos( $partname , '-' ) +1 ) . '.' . $extension, NS_FILE );
1152                         $file_thb = wfLocalFile( $nt_thb );
1153                         if( $file_thb->exists() ) {
1154                                 return array(
1155                                         'warning' => 'thumb',
1156                                         'file' => $file,
1157                                         'thumbFile' => $file_thb
1158                                 );
1159                         } else {
1160                                 // File does not exist, but we just don't like the name
1161                                 return array(
1162                                         'warning' => 'thumb-name',
1163                                         'file' => $file,
1164                                         'thumbFile' => $file_thb
1165                                 );
1166                         }
1167                 }
1168
1169
1170                 foreach( self::getFilenamePrefixBlacklist() as $prefix ) {
1171                         if ( substr( $partname, 0, strlen( $prefix ) ) == $prefix ) {
1172                                 return array(
1173                                         'warning' => 'bad-prefix',
1174                                         'file' => $file,
1175                                         'prefix' => $prefix
1176                                 );
1177                         }
1178                 }
1179
1180                 return false;
1181         }
1182
1183         /**
1184          * Helper function that checks whether the filename looks like a thumbnail
1185          */
1186         public static function isThumbName( $filename ) {
1187                 $n = strrpos( $filename, '.' );
1188                 $partname = $n ? substr( $filename, 0, $n ) : $filename;
1189                 return (
1190                                         substr( $partname , 3, 3 ) == 'px-' ||
1191                                         substr( $partname , 2, 3 ) == 'px-'
1192                                 ) &&
1193                                 preg_match( "/[0-9]{2}/" , substr( $partname , 0, 2 ) );
1194         }
1195
1196         /**
1197          * Get a list of blacklisted filename prefixes from [[MediaWiki:Filename-prefix-blacklist]]
1198          *
1199          * @return array list of prefixes
1200          */
1201         public static function getFilenamePrefixBlacklist() {
1202                 $blacklist = array();
1203                 $message = wfMessage( 'filename-prefix-blacklist' )->inContentLanguage();
1204                 if( !$message->isDisabled() ) {
1205                         $lines = explode( "\n", $message->plain() );
1206                         foreach( $lines as $line ) {
1207                                 // Remove comment lines
1208                                 $comment = substr( trim( $line ), 0, 1 );
1209                                 if ( $comment == '#' || $comment == '' ) {
1210                                         continue;
1211                                 }
1212                                 // Remove additional comments after a prefix
1213                                 $comment = strpos( $line, '#' );
1214                                 if ( $comment > 0 ) {
1215                                         $line = substr( $line, 0, $comment-1 );
1216                                 }
1217                                 $blacklist[] = trim( $line );
1218                         }
1219                 }
1220                 return $blacklist;
1221         }
1222
1223         /**
1224          * Gets image info about the file just uploaded.
1225          *
1226          * Also has the effect of setting metadata to be an 'indexed tag name' in returned API result if
1227          * 'metadata' was requested. Oddly, we have to pass the "result" object down just so it can do that
1228          * with the appropriate format, presumably.
1229          *
1230          * @param $result ApiResult:
1231          * @return Array: image info
1232          */
1233         public function getImageInfo( $result ) {
1234                 $file = $this->getLocalFile();
1235                 // TODO This cries out for refactoring. We really want to say $file->getAllInfo(); here.
1236                 // Perhaps "info" methods should be moved into files, and the API should just wrap them in queries.
1237                 if ( $file instanceof UploadStashFile ) {
1238                         $imParam = ApiQueryStashImageInfo::getPropertyNames();
1239                         $info = ApiQueryStashImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1240                 } else {
1241                         $imParam = ApiQueryImageInfo::getPropertyNames();
1242                         $info = ApiQueryImageInfo::getInfo( $file, array_flip( $imParam ), $result );
1243                 }
1244                 return $info;
1245         }
1246
1247
1248         public function convertVerifyErrorToStatus( $error ) {
1249                 $code = $error['status'];
1250                 unset( $code['status'] );
1251                 return Status::newFatal( $this->getVerificationErrorCode( $code ), $error );
1252         }
1253
1254         public static function getMaxUploadSize( $forType = null ) {
1255                 global $wgMaxUploadSize;
1256
1257                 if ( is_array( $wgMaxUploadSize ) ) {
1258                         if ( !is_null( $forType) && isset( $wgMaxUploadSize[$forType] ) ) {
1259                                 return $wgMaxUploadSize[$forType];
1260                         } else {
1261                                 return $wgMaxUploadSize['*'];
1262                         }
1263                 } else {
1264                         return intval( $wgMaxUploadSize );
1265                 }
1266
1267         }
1268 }