includes/Import.php

   1 <?php
   2 /**
   3  * MediaWiki page data importer.
   4  *
   5  * Copyright © 2003,2005 Brion Vibber <brion@pobox.com>
   6  * http://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup SpecialPage
  25  */
  26
  27 /**
  28  * XML file reader for the page data importer
  29  *
  30  * implements Special:Import
  31  * @ingroup SpecialPage
  32  */
  33 class WikiImporter {
  34         private $reader = null;
  35         private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
  36         private $mSiteInfoCallback, $mTargetNamespace, $mTargetRootPage, $mPageOutCallback;
  37         private $mNoticeCallback, $mDebug;
  38         private $mImportUploads, $mImageBasePath;
  39         private $mNoUpdates = false;
  40
  41         /**
  42          * Creates an ImportXMLReader drawing from the source provided
  43          * @param $source
  44          */
  45         function __construct( $source ) {
  46                 $this->reader = new XMLReader();
  47
  48                 stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  49                 $id = UploadSourceAdapter::registerSource( $source );
  50                 if (defined( 'LIBXML_PARSEHUGE' ) ) {
  51                         $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  52                 } else {
  53                         $this->reader->open( "uploadsource://$id" );
  54                 }
  55
  56                 // Default callbacks
  57                 $this->setRevisionCallback( array( $this, "importRevision" ) );
  58                 $this->setUploadCallback( array( $this, 'importUpload' ) );
  59                 $this->setLogItemCallback( array( $this, 'importLogItem' ) );
  60                 $this->setPageOutCallback( array( $this, 'finishImportPage' ) );
  61         }
  62
  63         private function throwXmlError( $err ) {
  64                 $this->debug( "FAILURE: $err" );
  65                 wfDebug( "WikiImporter XML error: $err\n" );
  66         }
  67
  68         private function debug( $data ) {
  69                 if( $this->mDebug ) {
  70                         wfDebug( "IMPORT: $data\n" );
  71                 }
  72         }
  73
  74         private function warn( $data ) {
  75                 wfDebug( "IMPORT: $data\n" );
  76         }
  77
  78         private function notice( $msg /*, $param, ...*/ ) {
  79                 $params = func_get_args();
  80                 array_shift( $params );
  81
  82                 if ( is_callable( $this->mNoticeCallback ) ) {
  83                         call_user_func( $this->mNoticeCallback, $msg, $params );
  84                 } else { # No ImportReporter -> CLI
  85                         echo wfMessage( $msg, $params )->text() . "\n";
  86                 }
  87         }
  88
  89         /**
  90          * Set debug mode...
  91          * @param $debug bool
  92          */
  93         function setDebug( $debug ) {
  94                 $this->mDebug = $debug;
  95         }
  96
  97         /**
  98          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
  99          * @param $noupdates bool
 100          */
 101         function setNoUpdates( $noupdates ) {
 102                 $this->mNoUpdates = $noupdates;
 103         }
 104
 105         /**
 106          * Set a callback that displays notice messages
 107          *
 108          * @param $callback callback
 109          * @return callback
 110          */
 111         public function setNoticeCallback( $callback ) {
 112                 return wfSetVar( $this->mNoticeCallback, $callback );
 113         }
 114
 115         /**
 116          * Sets the action to perform as each new page in the stream is reached.
 117          * @param $callback callback
 118          * @return callback
 119          */
 120         public function setPageCallback( $callback ) {
 121                 $previous = $this->mPageCallback;
 122                 $this->mPageCallback = $callback;
 123                 return $previous;
 124         }
 125
 126         /**
 127          * Sets the action to perform as each page in the stream is completed.
 128          * Callback accepts the page title (as a Title object), a second object
 129          * with the original title form (in case it's been overridden into a
 130          * local namespace), and a count of revisions.
 131          *
 132          * @param $callback callback
 133          * @return callback
 134          */
 135         public function setPageOutCallback( $callback ) {
 136                 $previous = $this->mPageOutCallback;
 137                 $this->mPageOutCallback = $callback;
 138                 return $previous;
 139         }
 140
 141         /**
 142          * Sets the action to perform as each page revision is reached.
 143          * @param $callback callback
 144          * @return callback
 145          */
 146         public function setRevisionCallback( $callback ) {
 147                 $previous = $this->mRevisionCallback;
 148                 $this->mRevisionCallback = $callback;
 149                 return $previous;
 150         }
 151
 152         /**
 153          * Sets the action to perform as each file upload version is reached.
 154          * @param $callback callback
 155          * @return callback
 156          */
 157         public function setUploadCallback( $callback ) {
 158                 $previous = $this->mUploadCallback;
 159                 $this->mUploadCallback = $callback;
 160                 return $previous;
 161         }
 162
 163         /**
 164          * Sets the action to perform as each log item reached.
 165          * @param $callback callback
 166          * @return callback
 167          */
 168         public function setLogItemCallback( $callback ) {
 169                 $previous = $this->mLogItemCallback;
 170                 $this->mLogItemCallback = $callback;
 171                 return $previous;
 172         }
 173
 174         /**
 175          * Sets the action to perform when site info is encountered
 176          * @param $callback callback
 177          * @return callback
 178          */
 179         public function setSiteInfoCallback( $callback ) {
 180                 $previous = $this->mSiteInfoCallback;
 181                 $this->mSiteInfoCallback = $callback;
 182                 return $previous;
 183         }
 184
 185         /**
 186          * Set a target namespace to override the defaults
 187          * @param $namespace
 188          * @return bool
 189          */
 190         public function setTargetNamespace( $namespace ) {
 191                 if( is_null( $namespace ) ) {
 192                         // Don't override namespaces
 193                         $this->mTargetNamespace = null;
 194                 } elseif( $namespace >= 0 ) {
 195                         // @todo FIXME: Check for validity
 196                         $this->mTargetNamespace = intval( $namespace );
 197                 } else {
 198                         return false;
 199                 }
 200         }
 201
 202         /**
 203          * Set a target root page under which all pages are imported
 204          * @param $rootpage
 205          * @return status object
 206          */
 207         public function setTargetRootPage( $rootpage ) {
 208                 $status = Status::newGood();
 209                 if( is_null( $rootpage ) ) {
 210                         // No rootpage
 211                         $this->mTargetRootPage = null;
 212                 } elseif( $rootpage !== '' ) {
 213                         $rootpage = rtrim( $rootpage, '/' ); //avoid double slashes
 214                         $title = Title::newFromText( $rootpage, !is_null( $this->mTargetNamespace ) ? $this->mTargetNamespace : NS_MAIN );
 215                         if( !$title || $title->isExternal() ) {
 216                                 $status->fatal( 'import-rootpage-invalid' );
 217                         } else {
 218                                 if( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 219                                         global $wgContLang;
 220
 221                                         $displayNSText = $title->getNamespace() == NS_MAIN
 222                                                 ? wfMessage( 'blanknamespace' )->text()
 223                                                 : $wgContLang->getNsText( $title->getNamespace() );
 224                                         $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 225                                 } else {
 226                                         // set namespace to 'all', so the namespace check in processTitle() can passed
 227                                         $this->setTargetNamespace( null );
 228                                         $this->mTargetRootPage = $title->getPrefixedDBKey();
 229                                 }
 230                         }
 231                 }
 232                 return $status;
 233         }
 234
 235         /**
 236          * @param $dir
 237          */
 238         public function setImageBasePath( $dir ) {
 239                 $this->mImageBasePath = $dir;
 240         }
 241
 242         /**
 243          * @param $import
 244          */
 245         public function setImportUploads( $import ) {
 246                 $this->mImportUploads = $import;
 247         }
 248
 249         /**
 250          * Default per-revision callback, performs the import.
 251          * @param $revision WikiRevision
 252          * @return bool
 253          */
 254         public function importRevision( $revision ) {
 255                 $dbw = wfGetDB( DB_MASTER );
 256                 return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
 257         }
 258
 259         /**
 260          * Default per-revision callback, performs the import.
 261          * @param $rev WikiRevision
 262          * @return bool
 263          */
 264         public function importLogItem( $rev ) {
 265                 $dbw = wfGetDB( DB_MASTER );
 266                 return $dbw->deadlockLoop( array( $rev, 'importLogItem' ) );
 267         }
 268
 269         /**
 270          * Dummy for now...
 271          * @param $revision
 272          * @return bool
 273          */
 274         public function importUpload( $revision ) {
 275                 $dbw = wfGetDB( DB_MASTER );
 276                 return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
 277         }
 278
 279         /**
 280          * Mostly for hook use
 281          * @param $title
 282          * @param $origTitle
 283          * @param $revCount
 284          * @param $sRevCount
 285          * @param $pageInfo
 286          * @return
 287          */
 288         public function finishImportPage( $title, $origTitle, $revCount, $sRevCount, $pageInfo ) {
 289                 $args = func_get_args();
 290                 return wfRunHooks( 'AfterImportPage', $args );
 291         }
 292
 293         /**
 294          * Alternate per-revision callback, for debugging.
 295          * @param $revision WikiRevision
 296          */
 297         public function debugRevisionHandler( &$revision ) {
 298                 $this->debug( "Got revision:" );
 299                 if( is_object( $revision->title ) ) {
 300                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 301                 } else {
 302                         $this->debug( "-- Title: <invalid>" );
 303                 }
 304                 $this->debug( "-- User: " . $revision->user_text );
 305                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 306                 $this->debug( "-- Comment: " . $revision->comment );
 307                 $this->debug( "-- Text: " . $revision->text );
 308         }
 309
 310         /**
 311          * Notify the callback function when a new "<page>" is reached.
 312          * @param $title Title
 313          */
 314         function pageCallback( $title ) {
 315                 if( isset( $this->mPageCallback ) ) {
 316                         call_user_func( $this->mPageCallback, $title );
 317                 }
 318         }
 319
 320         /**
 321          * Notify the callback function when a "</page>" is closed.
 322          * @param $title Title
 323          * @param $origTitle Title
 324          * @param $revCount Integer
 325          * @param $sucCount Int: number of revisions for which callback returned true
 326          * @param $pageInfo Array: associative array of page information
 327          */
 328         private function pageOutCallback( $title, $origTitle, $revCount, $sucCount, $pageInfo ) {
 329                 if( isset( $this->mPageOutCallback ) ) {
 330                         $args = func_get_args();
 331                         call_user_func_array( $this->mPageOutCallback, $args );
 332                 }
 333         }
 334
 335         /**
 336          * Notify the callback function of a revision
 337          * @param $revision WikiRevision object
 338          * @return bool|mixed
 339          */
 340         private function revisionCallback( $revision ) {
 341                 if ( isset( $this->mRevisionCallback ) ) {
 342                         return call_user_func_array( $this->mRevisionCallback,
 343                                         array( $revision, $this ) );
 344                 } else {
 345                         return false;
 346                 }
 347         }
 348
 349         /**
 350          * Notify the callback function of a new log item
 351          * @param $revision WikiRevision object
 352          * @return bool|mixed
 353          */
 354         private function logItemCallback( $revision ) {
 355                 if ( isset( $this->mLogItemCallback ) ) {
 356                         return call_user_func_array( $this->mLogItemCallback,
 357                                         array( $revision, $this ) );
 358                 } else {
 359                         return false;
 360                 }
 361         }
 362
 363         /**
 364          * Shouldn't something like this be built-in to XMLReader?
 365          * Fetches text contents of the current element, assuming
 366          * no sub-elements or such scary things.
 367          * @return string
 368          * @access private
 369          */
 370         private function nodeContents() {
 371                 if( $this->reader->isEmptyElement ) {
 372                         return "";
 373                 }
 374                 $buffer = "";
 375                 while( $this->reader->read() ) {
 376                         switch( $this->reader->nodeType ) {
 377                         case XmlReader::TEXT:
 378                         case XmlReader::SIGNIFICANT_WHITESPACE:
 379                                 $buffer .= $this->reader->value;
 380                                 break;
 381                         case XmlReader::END_ELEMENT:
 382                                 return $buffer;
 383                         }
 384                 }
 385
 386                 $this->reader->close();
 387                 return '';
 388         }
 389
 390         # --------------
 391
 392         /** Left in for debugging */
 393         private function dumpElement() {
 394                 static $lookup = null;
 395                 if (!$lookup) {
 396                         $xmlReaderConstants = array(
 397                                 "NONE",
 398                                 "ELEMENT",
 399                                 "ATTRIBUTE",
 400                                 "TEXT",
 401                                 "CDATA",
 402                                 "ENTITY_REF",
 403                                 "ENTITY",
 404                                 "PI",
 405                                 "COMMENT",
 406                                 "DOC",
 407                                 "DOC_TYPE",
 408                                 "DOC_FRAGMENT",
 409                                 "NOTATION",
 410                                 "WHITESPACE",
 411                                 "SIGNIFICANT_WHITESPACE",
 412                                 "END_ELEMENT",
 413                                 "END_ENTITY",
 414                                 "XML_DECLARATION",
 415                                 );
 416                         $lookup = array();
 417
 418                         foreach( $xmlReaderConstants as $name ) {
 419                                 $lookup[constant("XmlReader::$name")] = $name;
 420                         }
 421                 }
 422
 423                 print( var_dump(
 424                         $lookup[$this->reader->nodeType],
 425                         $this->reader->name,
 426                         $this->reader->value
 427                 )."\n\n" );
 428         }
 429
 430         /**
 431          * Primary entry point
 432          * @throws MWException
 433          * @return bool
 434          */
 435         public function doImport() {
 436                 $this->reader->read();
 437
 438                 if ( $this->reader->name != 'mediawiki' ) {
 439                         throw new MWException( "Expected <mediawiki> tag, got ".
 440                                 $this->reader->name );
 441                 }
 442                 $this->debug( "<mediawiki> tag is correct." );
 443
 444                 $this->debug( "Starting primary dump processing loop." );
 445
 446                 $keepReading = $this->reader->read();
 447                 $skip = false;
 448                 while ( $keepReading ) {
 449                         $tag = $this->reader->name;
 450                         $type = $this->reader->nodeType;
 451
 452                         if ( !wfRunHooks( 'ImportHandleToplevelXMLTag', $this ) ) {
 453                                 // Do nothing
 454                         } elseif ( $tag == 'mediawiki' && $type == XmlReader::END_ELEMENT ) {
 455                                 break;
 456                         } elseif ( $tag == 'siteinfo' ) {
 457                                 $this->handleSiteInfo();
 458                         } elseif ( $tag == 'page' ) {
 459                                 $this->handlePage();
 460                         } elseif ( $tag == 'logitem' ) {
 461                                 $this->handleLogItem();
 462                         } elseif ( $tag != '#text' ) {
 463                                 $this->warn( "Unhandled top-level XML tag $tag" );
 464
 465                                 $skip = true;
 466                         }
 467
 468                         if ($skip) {
 469                                 $keepReading = $this->reader->next();
 470                                 $skip = false;
 471                                 $this->debug( "Skip" );
 472                         } else {
 473                                 $keepReading = $this->reader->read();
 474                         }
 475                 }
 476
 477                 return true;
 478         }
 479
 480         /**
 481          * @return bool
 482          * @throws MWException
 483          */
 484         private function handleSiteInfo() {
 485                 // Site info is useful, but not actually used for dump imports.
 486                 // Includes a quick short-circuit to save performance.
 487                 if ( ! $this->mSiteInfoCallback ) {
 488                         $this->reader->next();
 489                         return true;
 490                 }
 491                 throw new MWException( "SiteInfo tag is not yet handled, do not set mSiteInfoCallback" );
 492         }
 493
 494         private function handleLogItem() {
 495                 $this->debug( "Enter log item handler." );
 496                 $logInfo = array();
 497
 498                 // Fields that can just be stuffed in the pageInfo object
 499                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 500                                         'logtitle', 'params' );
 501
 502                 while ( $this->reader->read() ) {
 503                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 504                                         $this->reader->name == 'logitem') {
 505                                 break;
 506                         }
 507
 508                         $tag = $this->reader->name;
 509
 510                         if ( !wfRunHooks( 'ImportHandleLogItemXMLTag',
 511                                                 $this, $logInfo ) ) {
 512                                 // Do nothing
 513                         } elseif ( in_array( $tag, $normalFields ) ) {
 514                                 $logInfo[$tag] = $this->nodeContents();
 515                         } elseif ( $tag == 'contributor' ) {
 516                                 $logInfo['contributor'] = $this->handleContributor();
 517                         } elseif ( $tag != '#text' ) {
 518                                 $this->warn( "Unhandled log-item XML tag $tag" );
 519                         }
 520                 }
 521
 522                 $this->processLogItem( $logInfo );
 523         }
 524
 525         /**
 526          * @param $logInfo
 527          * @return bool|mixed
 528          */
 529         private function processLogItem( $logInfo ) {
 530                 $revision = new WikiRevision;
 531
 532                 $revision->setID( $logInfo['id'] );
 533                 $revision->setType( $logInfo['type'] );
 534                 $revision->setAction( $logInfo['action'] );
 535                 $revision->setTimestamp( $logInfo['timestamp'] );
 536                 $revision->setParams( $logInfo['params'] );
 537                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 538                 $revision->setNoUpdates( $this->mNoUpdates );
 539
 540                 if ( isset( $logInfo['comment'] ) ) {
 541                         $revision->setComment( $logInfo['comment'] );
 542                 }
 543
 544                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 545                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 546                 }
 547                 if ( isset( $logInfo['contributor']['username'] ) ) {
 548                         $revision->setUserName( $logInfo['contributor']['username'] );
 549                 }
 550
 551                 return $this->logItemCallback( $revision );
 552         }
 553
 554         private function handlePage() {
 555                 // Handle page data.
 556                 $this->debug( "Enter page handler." );
 557                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 558
 559                 // Fields that can just be stuffed in the pageInfo object
 560                 $normalFields = array( 'title', 'id', 'redirect', 'restrictions' );
 561
 562                 $skip = false;
 563                 $badTitle = false;
 564
 565                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 566                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 567                                         $this->reader->name == 'page') {
 568                                 break;
 569                         }
 570
 571                         $tag = $this->reader->name;
 572
 573                         if ( $badTitle ) {
 574                                 // The title is invalid, bail out of this page
 575                                 $skip = true;
 576                         } elseif ( !wfRunHooks( 'ImportHandlePageXMLTag', array( $this,
 577                                                 &$pageInfo ) ) ) {
 578                                 // Do nothing
 579                         } elseif ( in_array( $tag, $normalFields ) ) {
 580                                 $pageInfo[$tag] = $this->nodeContents();
 581                                 if ( $tag == 'title' ) {
 582                                         $title = $this->processTitle( $pageInfo['title'] );
 583
 584                                         if ( !$title ) {
 585                                                 $badTitle = true;
 586                                                 $skip = true;
 587                                         }
 588
 589                                         $this->pageCallback( $title );
 590                                         list( $pageInfo['_title'], $origTitle ) = $title;
 591                                 }
 592                         } elseif ( $tag == 'revision' ) {
 593                                 $this->handleRevision( $pageInfo );
 594                         } elseif ( $tag == 'upload' ) {
 595                                 $this->handleUpload( $pageInfo );
 596                         } elseif ( $tag != '#text' ) {
 597                                 $this->warn( "Unhandled page XML tag $tag" );
 598                                 $skip = true;
 599                         }
 600                 }
 601
 602                 $this->pageOutCallback( $pageInfo['_title'], $origTitle,
 603                                         $pageInfo['revisionCount'],
 604                                         $pageInfo['successfulRevisionCount'],
 605                                         $pageInfo );
 606         }
 607
 608         /**
 609          * @param $pageInfo array
 610          */
 611         private function handleRevision( &$pageInfo ) {
 612                 $this->debug( "Enter revision handler" );
 613                 $revisionInfo = array();
 614
 615                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' );
 616
 617                 $skip = false;
 618
 619                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 620                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 621                                         $this->reader->name == 'revision') {
 622                                 break;
 623                         }
 624
 625                         $tag = $this->reader->name;
 626
 627                         if ( !wfRunHooks( 'ImportHandleRevisionXMLTag', $this,
 628                                                 $pageInfo, $revisionInfo ) ) {
 629                                 // Do nothing
 630                         } elseif ( in_array( $tag, $normalFields ) ) {
 631                                 $revisionInfo[$tag] = $this->nodeContents();
 632                         } elseif ( $tag == 'contributor' ) {
 633                                 $revisionInfo['contributor'] = $this->handleContributor();
 634                         } elseif ( $tag != '#text' ) {
 635                                 $this->warn( "Unhandled revision XML tag $tag" );
 636                                 $skip = true;
 637                         }
 638                 }
 639
 640                 $pageInfo['revisionCount']++;
 641                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 642                         $pageInfo['successfulRevisionCount']++;
 643                 }
 644         }
 645
 646         /**
 647          * @param $pageInfo
 648          * @param $revisionInfo
 649          * @return bool|mixed
 650          */
 651         private function processRevision( $pageInfo, $revisionInfo ) {
 652                 $revision = new WikiRevision;
 653
 654                 if( isset( $revisionInfo['id'] ) ) {
 655                         $revision->setID( $revisionInfo['id'] );
 656                 }
 657                 if ( isset( $revisionInfo['text'] ) ) {
 658                         $revision->setText( $revisionInfo['text'] );
 659                 }
 660                 if ( isset( $revisionInfo['model'] ) ) {
 661                         $revision->setModel( $revisionInfo['model'] );
 662                 }
 663                 if ( isset( $revisionInfo['format'] ) ) {
 664                         $revision->setFormat( $revisionInfo['format'] );
 665                 }
 666                 $revision->setTitle( $pageInfo['_title'] );
 667
 668                 if ( isset( $revisionInfo['timestamp'] ) ) {
 669                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 670                 } else {
 671                         $revision->setTimestamp( wfTimestampNow() );
 672                 }
 673
 674                 if ( isset( $revisionInfo['comment'] ) ) {
 675                         $revision->setComment( $revisionInfo['comment'] );
 676                 }
 677
 678                 if ( isset( $revisionInfo['minor'] ) ) {
 679                         $revision->setMinor( true );
 680                 }
 681                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 682                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 683                 }
 684                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 685                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 686                 }
 687                 $revision->setNoUpdates( $this->mNoUpdates );
 688
 689                 return $this->revisionCallback( $revision );
 690         }
 691
 692         /**
 693          * @param $pageInfo
 694          * @return mixed
 695          */
 696         private function handleUpload( &$pageInfo ) {
 697                 $this->debug( "Enter upload handler" );
 698                 $uploadInfo = array();
 699
 700                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 701                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 702
 703                 $skip = false;
 704
 705                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 706                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 707                                         $this->reader->name == 'upload') {
 708                                 break;
 709                         }
 710
 711                         $tag = $this->reader->name;
 712
 713                         if ( !wfRunHooks( 'ImportHandleUploadXMLTag', $this,
 714                                                 $pageInfo ) ) {
 715                                 // Do nothing
 716                         } elseif ( in_array( $tag, $normalFields ) ) {
 717                                 $uploadInfo[$tag] = $this->nodeContents();
 718                         } elseif ( $tag == 'contributor' ) {
 719                                 $uploadInfo['contributor'] = $this->handleContributor();
 720                         } elseif ( $tag == 'contents' ) {
 721                                 $contents = $this->nodeContents();
 722                                 $encoding = $this->reader->getAttribute( 'encoding' );
 723                                 if ( $encoding === 'base64' ) {
 724                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 725                                         $uploadInfo['isTempSrc'] = true;
 726                                 }
 727                         } elseif ( $tag != '#text' ) {
 728                                 $this->warn( "Unhandled upload XML tag $tag" );
 729                                 $skip = true;
 730                         }
 731                 }
 732
 733                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 734                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 735                         if ( file_exists( $path ) ) {
 736                                 $uploadInfo['fileSrc'] = $path;
 737                                 $uploadInfo['isTempSrc'] = false;
 738                         }
 739                 }
 740
 741                 if ( $this->mImportUploads ) {
 742                         return $this->processUpload( $pageInfo, $uploadInfo );
 743                 }
 744         }
 745
 746         /**
 747          * @param $contents
 748          * @return string
 749          */
 750         private function dumpTemp( $contents ) {
 751                 $filename = tempnam( wfTempDir(), 'importupload' );
 752                 file_put_contents( $filename, $contents );
 753                 return $filename;
 754         }
 755
 756         /**
 757          * @param $pageInfo
 758          * @param $uploadInfo
 759          * @return mixed
 760          */
 761         private function processUpload( $pageInfo, $uploadInfo ) {
 762                 $revision = new WikiRevision;
 763                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 764
 765                 $revision->setTitle( $pageInfo['_title'] );
 766                 $revision->setID( $pageInfo['id'] );
 767                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 768                 $revision->setText( $text );
 769                 $revision->setFilename( $uploadInfo['filename'] );
 770                 if ( isset( $uploadInfo['archivename'] ) ) {
 771                         $revision->setArchiveName( $uploadInfo['archivename'] );
 772                 }
 773                 $revision->setSrc( $uploadInfo['src'] );
 774                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 775                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 776                                 !empty( $uploadInfo['isTempSrc'] ) );
 777                 }
 778                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 779                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 780                 }
 781                 $revision->setSize( intval( $uploadInfo['size'] ) );
 782                 $revision->setComment( $uploadInfo['comment'] );
 783
 784                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 785                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 786                 }
 787                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 788                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 789                 }
 790                 $revision->setNoUpdates( $this->mNoUpdates );
 791
 792                 return call_user_func( $this->mUploadCallback, $revision );
 793         }
 794
 795         /**
 796          * @return array
 797          */
 798         private function handleContributor() {
 799                 $fields = array( 'id', 'ip', 'username' );
 800                 $info = array();
 801
 802                 while ( $this->reader->read() ) {
 803                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 804                                         $this->reader->name == 'contributor') {
 805                                 break;
 806                         }
 807
 808                         $tag = $this->reader->name;
 809
 810                         if ( in_array( $tag, $fields ) ) {
 811                                 $info[$tag] = $this->nodeContents();
 812                         }
 813                 }
 814
 815                 return $info;
 816         }
 817
 818         /**
 819          * @param $text string
 820          * @return Array or false
 821          */
 822         private function processTitle( $text ) {
 823                 global $wgCommandLineMode;
 824
 825                 $workTitle = $text;
 826                 $origTitle = Title::newFromText( $workTitle );
 827
 828                 if( !is_null( $this->mTargetNamespace ) && !is_null( $origTitle ) ) {
 829                         # makeTitleSafe, because $origTitle can have a interwiki (different setting of interwiki map)
 830                         # and than dbKey can begin with a lowercase char
 831                         $title = Title::makeTitleSafe( $this->mTargetNamespace,
 832                                 $origTitle->getDBkey() );
 833                 } else {
 834                         if( !is_null( $this->mTargetRootPage ) ) {
 835                                 $workTitle = $this->mTargetRootPage . '/' . $workTitle;
 836                         }
 837                         $title = Title::newFromText( $workTitle );
 838                 }
 839
 840                 if( is_null( $title ) ) {
 841                         # Invalid page title? Ignore the page
 842                         $this->notice( 'import-error-invalid', $workTitle );
 843                         return false;
 844                 } elseif( $title->isExternal() ) {
 845                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
 846                         return false;
 847                 } elseif( !$title->canExist() ) {
 848                         $this->notice( 'import-error-special', $title->getPrefixedText() );
 849                         return false;
 850                 } elseif( !$title->userCan( 'edit' ) && !$wgCommandLineMode ) {
 851                         # Do not import if the importing wiki user cannot edit this page
 852                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
 853                         return false;
 854                 } elseif( !$title->exists() && !$title->userCan( 'create' ) && !$wgCommandLineMode ) {
 855                         # Do not import if the importing wiki user cannot create this page
 856                         $this->notice( 'import-error-create', $title->getPrefixedText() );
 857                         return false;
 858                 }
 859
 860                 return array( $title, $origTitle );
 861         }
 862 }
 863
 864 /** This is a horrible hack used to keep source compatibility */
 865 class UploadSourceAdapter {
 866         static $sourceRegistrations = array();
 867
 868         private $mSource;
 869         private $mBuffer;
 870         private $mPosition;
 871
 872         /**
 873          * @param $source
 874          * @return string
 875          */
 876         static function registerSource( $source ) {
 877                 $id = wfRandomString();
 878
 879                 self::$sourceRegistrations[$id] = $source;
 880
 881                 return $id;
 882         }
 883
 884         /**
 885          * @param $path
 886          * @param $mode
 887          * @param $options
 888          * @param $opened_path
 889          * @return bool
 890          */
 891         function stream_open( $path, $mode, $options, &$opened_path ) {
 892                 $url = parse_url($path);
 893                 $id = $url['host'];
 894
 895                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
 896                         return false;
 897                 }
 898
 899                 $this->mSource = self::$sourceRegistrations[$id];
 900
 901                 return true;
 902         }
 903
 904         /**
 905          * @param $count
 906          * @return string
 907          */
 908         function stream_read( $count ) {
 909                 $return = '';
 910                 $leave = false;
 911
 912                 while ( !$leave && !$this->mSource->atEnd() &&
 913                                 strlen($this->mBuffer) < $count ) {
 914                         $read = $this->mSource->readChunk();
 915
 916                         if ( !strlen($read) ) {
 917                                 $leave = true;
 918                         }
 919
 920                         $this->mBuffer .= $read;
 921                 }
 922
 923                 if ( strlen($this->mBuffer) ) {
 924                         $return = substr( $this->mBuffer, 0, $count );
 925                         $this->mBuffer = substr( $this->mBuffer, $count );
 926                 }
 927
 928                 $this->mPosition += strlen($return);
 929
 930                 return $return;
 931         }
 932
 933         /**
 934          * @param $data
 935          * @return bool
 936          */
 937         function stream_write( $data ) {
 938                 return false;
 939         }
 940
 941         /**
 942          * @return mixed
 943          */
 944         function stream_tell() {
 945                 return $this->mPosition;
 946         }
 947
 948         /**
 949          * @return bool
 950          */
 951         function stream_eof() {
 952                 return $this->mSource->atEnd();
 953         }
 954
 955         /**
 956          * @return array
 957          */
 958         function url_stat() {
 959                 $result = array();
 960
 961                 $result['dev'] = $result[0] = 0;
 962                 $result['ino'] = $result[1] = 0;
 963                 $result['mode'] = $result[2] = 0;
 964                 $result['nlink'] = $result[3] = 0;
 965                 $result['uid'] = $result[4] = 0;
 966                 $result['gid'] = $result[5] = 0;
 967                 $result['rdev'] = $result[6] = 0;
 968                 $result['size'] = $result[7] = 0;
 969                 $result['atime'] = $result[8] = 0;
 970                 $result['mtime'] = $result[9] = 0;
 971                 $result['ctime'] = $result[10] = 0;
 972                 $result['blksize'] = $result[11] = 0;
 973                 $result['blocks'] = $result[12] = 0;
 974
 975                 return $result;
 976         }
 977 }
 978
 979 class XMLReader2 extends XMLReader {
 980
 981         /**
 982          * @return bool|string
 983          */
 984         function nodeContents() {
 985                 if( $this->isEmptyElement ) {
 986                         return "";
 987                 }
 988                 $buffer = "";
 989                 while( $this->read() ) {
 990                         switch( $this->nodeType ) {
 991                         case XmlReader::TEXT:
 992                         case XmlReader::SIGNIFICANT_WHITESPACE:
 993                                 $buffer .= $this->value;
 994                                 break;
 995                         case XmlReader::END_ELEMENT:
 996                                 return $buffer;
 997                         }
 998                 }
 999                 return $this->close();
1000         }
1001 }
1002
1003 /**
1004  * @todo document (e.g. one-sentence class description).
1005  * @ingroup SpecialPage
1006  */
1007 class WikiRevision {
1008         var $importer = null;
1009
1010         /**
1011          * @var Title
1012          */
1013         var $title = null;
1014         var $id = 0;
1015         var $timestamp = "20010115000000";
1016         var $user = 0;
1017         var $user_text = "";
1018         var $model = null;
1019         var $format = null;
1020         var $text = "";
1021         var $content = null;
1022         var $comment = "";
1023         var $minor = false;
1024         var $type = "";
1025         var $action = "";
1026         var $params = "";
1027         var $fileSrc = '';
1028         var $sha1base36 = false;
1029         var $isTemp = false;
1030         var $archiveName = '';
1031         var $fileIsTemp;
1032         private $mNoUpdates = false;
1033
1034         /**
1035          * @param $title
1036          * @throws MWException
1037          */
1038         function setTitle( $title ) {
1039                 if( is_object( $title ) ) {
1040                         $this->title = $title;
1041                 } elseif( is_null( $title ) ) {
1042                         throw new MWException( "WikiRevision given a null title in import. You may need to adjust \$wgLegalTitleChars." );
1043                 } else {
1044                         throw new MWException( "WikiRevision given non-object title in import." );
1045                 }
1046         }
1047
1048         /**
1049          * @param $id
1050          */
1051         function setID( $id ) {
1052                 $this->id = $id;
1053         }
1054
1055         /**
1056          * @param $ts
1057          */
1058         function setTimestamp( $ts ) {
1059                 # 2003-08-05T18:30:02Z
1060                 $this->timestamp = wfTimestamp( TS_MW, $ts );
1061         }
1062
1063         /**
1064          * @param $user
1065          */
1066         function setUsername( $user ) {
1067                 $this->user_text = $user;
1068         }
1069
1070         /**
1071          * @param $ip
1072          */
1073         function setUserIP( $ip ) {
1074                 $this->user_text = $ip;
1075         }
1076
1077         /**
1078          * @param $model
1079          */
1080         function setModel( $model ) {
1081                 $this->model = $model;
1082         }
1083
1084         /**
1085          * @param $format
1086          */
1087         function setFormat( $format ) {
1088                 $this->format = $format;
1089         }
1090
1091         /**
1092          * @param $text
1093          */
1094         function setText( $text ) {
1095                 $this->text = $text;
1096         }
1097
1098         /**
1099          * @param $text
1100          */
1101         function setComment( $text ) {
1102                 $this->comment = $text;
1103         }
1104
1105         /**
1106          * @param $minor
1107          */
1108         function setMinor( $minor ) {
1109                 $this->minor = (bool)$minor;
1110         }
1111
1112         /**
1113          * @param $src
1114          */
1115         function setSrc( $src ) {
1116                 $this->src = $src;
1117         }
1118
1119         /**
1120          * @param $src
1121          * @param $isTemp
1122          */
1123         function setFileSrc( $src, $isTemp ) {
1124                 $this->fileSrc = $src;
1125                 $this->fileIsTemp = $isTemp;
1126         }
1127
1128         /**
1129          * @param $sha1base36
1130          */
1131         function setSha1Base36( $sha1base36 ) {
1132                 $this->sha1base36 = $sha1base36;
1133         }
1134
1135         /**
1136          * @param $filename
1137          */
1138         function setFilename( $filename ) {
1139                 $this->filename = $filename;
1140         }
1141
1142         /**
1143          * @param $archiveName
1144          */
1145         function setArchiveName( $archiveName ) {
1146                 $this->archiveName = $archiveName;
1147         }
1148
1149         /**
1150          * @param $size
1151          */
1152         function setSize( $size ) {
1153                 $this->size = intval( $size );
1154         }
1155
1156         /**
1157          * @param $type
1158          */
1159         function setType( $type ) {
1160                 $this->type = $type;
1161         }
1162
1163         /**
1164          * @param $action
1165          */
1166         function setAction( $action ) {
1167                 $this->action = $action;
1168         }
1169
1170         /**
1171          * @param $params
1172          */
1173         function setParams( $params ) {
1174                 $this->params = $params;
1175         }
1176
1177         /**
1178          * @param $noupdates
1179          */
1180         public function setNoUpdates( $noupdates ) {
1181                 $this->mNoUpdates = $noupdates;
1182         }
1183
1184         /**
1185          * @return Title
1186          */
1187         function getTitle() {
1188                 return $this->title;
1189         }
1190
1191         /**
1192          * @return int
1193          */
1194         function getID() {
1195                 return $this->id;
1196         }
1197
1198         /**
1199          * @return string
1200          */
1201         function getTimestamp() {
1202                 return $this->timestamp;
1203         }
1204
1205         /**
1206          * @return string
1207          */
1208         function getUser() {
1209                 return $this->user_text;
1210         }
1211
1212         /**
1213          * @return string
1214          *
1215          * @deprecated Since 1.21, use getContent() instead.
1216          */
1217         function getText() {
1218                 ContentHandler::deprecated( __METHOD__, '1.21' );
1219
1220                 return $this->text;
1221         }
1222
1223         /**
1224          * @return Content
1225          */
1226         function getContent() {
1227                 if ( is_null( $this->content ) ) {
1228                         $this->content =
1229                                 ContentHandler::makeContent(
1230                                         $this->text,
1231                                         $this->getTitle(),
1232                                         $this->getModel(),
1233                                         $this->getFormat()
1234                                 );
1235                 }
1236
1237                 return $this->content;
1238         }
1239
1240         /**
1241          * @return String
1242          */
1243         function getModel() {
1244                 if ( is_null( $this->model ) ) {
1245                         $this->model = $this->getTitle()->getContentModel();
1246                 }
1247
1248                 return $this->model;
1249         }
1250
1251         /**
1252          * @return String
1253          */
1254         function getFormat() {
1255                 if ( is_null( $this->model ) ) {
1256                         $this->format = ContentHandler::getForTitle( $this->getTitle() )->getDefaultFormat();
1257                 }
1258
1259                 return $this->format;
1260         }
1261
1262         /**
1263          * @return string
1264          */
1265         function getComment() {
1266                 return $this->comment;
1267         }
1268
1269         /**
1270          * @return bool
1271          */
1272         function getMinor() {
1273                 return $this->minor;
1274         }
1275
1276         /**
1277          * @return mixed
1278          */
1279         function getSrc() {
1280                 return $this->src;
1281         }
1282
1283         /**
1284          * @return bool|String
1285          */
1286         function getSha1() {
1287                 if ( $this->sha1base36 ) {
1288                         return wfBaseConvert( $this->sha1base36, 36, 16 );
1289                 }
1290                 return false;
1291         }
1292
1293         /**
1294          * @return string
1295          */
1296         function getFileSrc() {
1297                 return $this->fileSrc;
1298         }
1299
1300         /**
1301          * @return bool
1302          */
1303         function isTempSrc() {
1304                 return $this->isTemp;
1305         }
1306
1307         /**
1308          * @return mixed
1309          */
1310         function getFilename() {
1311                 return $this->filename;
1312         }
1313
1314         /**
1315          * @return string
1316          */
1317         function getArchiveName() {
1318                 return $this->archiveName;
1319         }
1320
1321         /**
1322          * @return mixed
1323          */
1324         function getSize() {
1325                 return $this->size;
1326         }
1327
1328         /**
1329          * @return string
1330          */
1331         function getType() {
1332                 return $this->type;
1333         }
1334
1335         /**
1336          * @return string
1337          */
1338         function getAction() {
1339                 return $this->action;
1340         }
1341
1342         /**
1343          * @return string
1344          */
1345         function getParams() {
1346                 return $this->params;
1347         }
1348
1349         /**
1350          * @return bool
1351          */
1352         function importOldRevision() {
1353                 $dbw = wfGetDB( DB_MASTER );
1354
1355                 # Sneak a single revision into place
1356                 $user = User::newFromName( $this->getUser() );
1357                 if( $user ) {
1358                         $userId = intval( $user->getId() );
1359                         $userText = $user->getName();
1360                         $userObj = $user;
1361                 } else {
1362                         $userId = 0;
1363                         $userText = $this->getUser();
1364                         $userObj = new User;
1365                 }
1366
1367                 // avoid memory leak...?
1368                 $linkCache = LinkCache::singleton();
1369                 $linkCache->clear();
1370
1371                 $page = WikiPage::factory( $this->title );
1372                 if( !$page->exists() ) {
1373                         # must create the page...
1374                         $pageId = $page->insertOn( $dbw );
1375                         $created = true;
1376                         $oldcountable = null;
1377                 } else {
1378                         $pageId = $page->getId();
1379                         $created = false;
1380
1381                         $prior = $dbw->selectField( 'revision', '1',
1382                                 array( 'rev_page' => $pageId,
1383                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1384                                         'rev_user_text' => $userText,
1385                                         'rev_comment'   => $this->getComment() ),
1386                                 __METHOD__
1387                         );
1388                         if( $prior ) {
1389                                 // @todo FIXME: This could fail slightly for multiple matches :P
1390                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1391                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1392                                 return false;
1393                         }
1394                         $oldcountable = $page->isCountable();
1395                 }
1396
1397                 # @todo FIXME: Use original rev_id optionally (better for backups)
1398                 # Insert the row
1399                 $revision = new Revision( array(
1400                         'title'      => $this->title,
1401                         'page'       => $pageId,
1402                         'content_model'  => $this->getModel(),
1403                         'content_format' => $this->getFormat(),
1404                         'text'       => $this->getContent()->serialize( $this->getFormat() ), //XXX: just set 'content' => $this->getContent()?
1405                         'comment'    => $this->getComment(),
1406                         'user'       => $userId,
1407                         'user_text'  => $userText,
1408                         'timestamp'  => $this->timestamp,
1409                         'minor_edit' => $this->minor,
1410                         ) );
1411                 $revision->insertOn( $dbw );
1412                 $changed = $page->updateIfNewerOn( $dbw, $revision );
1413
1414                 if ( $changed !== false && !$this->mNoUpdates ) {
1415                         wfDebug( __METHOD__ . ": running updates\n" );
1416                         $page->doEditUpdates( $revision, $userObj, array( 'created' => $created, 'oldcountable' => $oldcountable ) );
1417                 }
1418
1419                 return true;
1420         }
1421
1422         /**
1423          * @return mixed
1424          */
1425         function importLogItem() {
1426                 $dbw = wfGetDB( DB_MASTER );
1427                 # @todo FIXME: This will not record autoblocks
1428                 if( !$this->getTitle() ) {
1429                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1430                                 $this->timestamp . "\n" );
1431                         return;
1432                 }
1433                 # Check if it exists already
1434                 // @todo FIXME: Use original log ID (better for backups)
1435                 $prior = $dbw->selectField( 'logging', '1',
1436                         array( 'log_type' => $this->getType(),
1437                                 'log_action'    => $this->getAction(),
1438                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1439                                 'log_namespace' => $this->getTitle()->getNamespace(),
1440                                 'log_title'     => $this->getTitle()->getDBkey(),
1441                                 'log_comment'   => $this->getComment(),
1442                                 #'log_user_text' => $this->user_text,
1443                                 'log_params'    => $this->params ),
1444                         __METHOD__
1445                 );
1446                 // @todo FIXME: This could fail slightly for multiple matches :P
1447                 if( $prior ) {
1448                         wfDebug( __METHOD__ . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp " .
1449                                 $this->timestamp . "\n" );
1450                         return;
1451                 }
1452                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1453                 $data = array(
1454                         'log_id' => $log_id,
1455                         'log_type' => $this->type,
1456                         'log_action' => $this->action,
1457                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1458                         'log_user' => User::idFromName( $this->user_text ),
1459                         #'log_user_text' => $this->user_text,
1460                         'log_namespace' => $this->getTitle()->getNamespace(),
1461                         'log_title' => $this->getTitle()->getDBkey(),
1462                         'log_comment' => $this->getComment(),
1463                         'log_params' => $this->params
1464                 );
1465                 $dbw->insert( 'logging', $data, __METHOD__ );
1466         }
1467
1468         /**
1469          * @return bool
1470          */
1471         function importUpload() {
1472                 # Construct a file
1473                 $archiveName = $this->getArchiveName();
1474                 if ( $archiveName ) {
1475                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1476                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1477                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1478                 } else {
1479                         $file = wfLocalFile( $this->getTitle() );
1480                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1481                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1482                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1483                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1484                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1485                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1486                         }
1487                 }
1488                 if( !$file ) {
1489                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1490                         return false;
1491                 }
1492
1493                 # Get the file source or download if necessary
1494                 $source = $this->getFileSrc();
1495                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1496                 if ( !$source ) {
1497                         $source = $this->downloadSource();
1498                         $flags |= File::DELETE_SOURCE;
1499                 }
1500                 if( !$source ) {
1501                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1502                         return false;
1503                 }
1504                 $sha1 = $this->getSha1();
1505                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1506                         if ( $flags & File::DELETE_SOURCE ) {
1507                                 # Broken file; delete it if it is a temporary file
1508                                 unlink( $source );
1509                         }
1510                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1511                         return false;
1512                 }
1513
1514                 $user = User::newFromName( $this->user_text );
1515
1516                 # Do the actual upload
1517                 if ( $archiveName ) {
1518                         $status = $file->uploadOld( $source, $archiveName,
1519                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1520                 } else {
1521                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1522                                 $flags, false, $this->getTimestamp(), $user );
1523                 }
1524
1525                 if ( $status->isGood() ) {
1526                         wfDebug( __METHOD__ . ": Succesful\n" );
1527                         return true;
1528                 } else {
1529                         wfDebug( __METHOD__ . ': failed: ' . $status->getXml() . "\n" );
1530                         return false;
1531                 }
1532         }
1533
1534         /**
1535          * @return bool|string
1536          */
1537         function downloadSource() {
1538                 global $wgEnableUploads;
1539                 if( !$wgEnableUploads ) {
1540                         return false;
1541                 }
1542
1543                 $tempo = tempnam( wfTempDir(), 'download' );
1544                 $f = fopen( $tempo, 'wb' );
1545                 if( !$f ) {
1546                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1547                         return false;
1548                 }
1549
1550                 // @todo FIXME!
1551                 $src = $this->getSrc();
1552                 $data = Http::get( $src );
1553                 if( !$data ) {
1554                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1555                         fclose( $f );
1556                         unlink( $tempo );
1557                         return false;
1558                 }
1559
1560                 fwrite( $f, $data );
1561                 fclose( $f );
1562
1563                 return $tempo;
1564         }
1565
1566 }
1567
1568 /**
1569  * @todo document (e.g. one-sentence class description).
1570  * @ingroup SpecialPage
1571  */
1572 class ImportStringSource {
1573         function __construct( $string ) {
1574                 $this->mString = $string;
1575                 $this->mRead = false;
1576         }
1577
1578         /**
1579          * @return bool
1580          */
1581         function atEnd() {
1582                 return $this->mRead;
1583         }
1584
1585         /**
1586          * @return bool|string
1587          */
1588         function readChunk() {
1589                 if( $this->atEnd() ) {
1590                         return false;
1591                 }
1592                 $this->mRead = true;
1593                 return $this->mString;
1594         }
1595 }
1596
1597 /**
1598  * @todo document (e.g. one-sentence class description).
1599  * @ingroup SpecialPage
1600  */
1601 class ImportStreamSource {
1602         function __construct( $handle ) {
1603                 $this->mHandle = $handle;
1604         }
1605
1606         /**
1607          * @return bool
1608          */
1609         function atEnd() {
1610                 return feof( $this->mHandle );
1611         }
1612
1613         /**
1614          * @return string
1615          */
1616         function readChunk() {
1617                 return fread( $this->mHandle, 32768 );
1618         }
1619
1620         /**
1621          * @param $filename string
1622          * @return Status
1623          */
1624         static function newFromFile( $filename ) {
1625                 wfSuppressWarnings();
1626                 $file = fopen( $filename, 'rt' );
1627                 wfRestoreWarnings();
1628                 if( !$file ) {
1629                         return Status::newFatal( "importcantopen" );
1630                 }
1631                 return Status::newGood( new ImportStreamSource( $file ) );
1632         }
1633
1634         /**
1635          * @param $fieldname string
1636          * @return Status
1637          */
1638         static function newFromUpload( $fieldname = "xmlimport" ) {
1639                 $upload =& $_FILES[$fieldname];
1640
1641                 if( !isset( $upload ) || !$upload['name'] ) {
1642                         return Status::newFatal( 'importnofile' );
1643                 }
1644                 if( !empty( $upload['error'] ) ) {
1645                         switch($upload['error']){
1646                                 case 1: # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1647                                         return Status::newFatal( 'importuploaderrorsize' );
1648                                 case 2: # The uploaded file exceeds the MAX_FILE_SIZE directive that was specified in the HTML form.
1649                                         return Status::newFatal( 'importuploaderrorsize' );
1650                                 case 3: # The uploaded file was only partially uploaded
1651                                         return Status::newFatal( 'importuploaderrorpartial' );
1652                                 case 6: #Missing a temporary folder.
1653                                         return Status::newFatal( 'importuploaderrortemp' );
1654                                 # case else: # Currently impossible
1655                         }
1656
1657                 }
1658                 $fname = $upload['tmp_name'];
1659                 if( is_uploaded_file( $fname ) ) {
1660                         return ImportStreamSource::newFromFile( $fname );
1661                 } else {
1662                         return Status::newFatal( 'importnofile' );
1663                 }
1664         }
1665
1666         /**
1667          * @param $url
1668          * @param $method string
1669          * @return Status
1670          */
1671         static function newFromURL( $url, $method = 'GET' ) {
1672                 wfDebug( __METHOD__ . ": opening $url\n" );
1673                 # Use the standard HTTP fetch function; it times out
1674                 # quicker and sorts out user-agent problems which might
1675                 # otherwise prevent importing from large sites, such
1676                 # as the Wikimedia cluster, etc.
1677                 $data = Http::request( $method, $url, array( 'followRedirects' => true ) );
1678                 if( $data !== false ) {
1679                         $file = tmpfile();
1680                         fwrite( $file, $data );
1681                         fflush( $file );
1682                         fseek( $file, 0 );
1683                         return Status::newGood( new ImportStreamSource( $file ) );
1684                 } else {
1685                         return Status::newFatal( 'importcantopen' );
1686                 }
1687         }
1688
1689         /**
1690          * @param $interwiki
1691          * @param $page
1692          * @param $history bool
1693          * @param $templates bool
1694          * @param $pageLinkDepth int
1695          * @return Status
1696          */
1697         public static function newFromInterwiki( $interwiki, $page, $history = false, $templates = false, $pageLinkDepth = 0 ) {
1698                 if( $page == '' ) {
1699                         return Status::newFatal( 'import-noarticle' );
1700                 }
1701                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1702                 if( is_null( $link ) || $link->getInterwiki() == '' ) {
1703                         return Status::newFatal( 'importbadinterwiki' );
1704                 } else {
1705                         $params = array();
1706                         if ( $history ) $params['history'] = 1;
1707                         if ( $templates ) $params['templates'] = 1;
1708                         if ( $pageLinkDepth ) $params['pagelink-depth'] = $pageLinkDepth;
1709                         $url = $link->getFullUrl( $params );
1710                         # For interwikis, use POST to avoid redirects.
1711                         return ImportStreamSource::newFromURL( $url, "POST" );
1712                 }
1713         }
1714 }