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          * @return bool
 433          */
 434         public function doImport() {
 435                 $this->reader->read();
 436
 437                 if ( $this->reader->name != 'mediawiki' ) {
 438                         throw new MWException( "Expected <mediawiki> tag, got ".
 439                                 $this->reader->name );
 440                 }
 441                 $this->debug( "<mediawiki> tag is correct." );
 442
 443                 $this->debug( "Starting primary dump processing loop." );
 444
 445                 $keepReading = $this->reader->read();
 446                 $skip = false;
 447                 while ( $keepReading ) {
 448                         $tag = $this->reader->name;
 449                         $type = $this->reader->nodeType;
 450
 451                         if ( !wfRunHooks( 'ImportHandleToplevelXMLTag', $this ) ) {
 452                                 // Do nothing
 453                         } elseif ( $tag == 'mediawiki' && $type == XmlReader::END_ELEMENT ) {
 454                                 break;
 455                         } elseif ( $tag == 'siteinfo' ) {
 456                                 $this->handleSiteInfo();
 457                         } elseif ( $tag == 'page' ) {
 458                                 $this->handlePage();
 459                         } elseif ( $tag == 'logitem' ) {
 460                                 $this->handleLogItem();
 461                         } elseif ( $tag != '#text' ) {
 462                                 $this->warn( "Unhandled top-level XML tag $tag" );
 463
 464                                 $skip = true;
 465                         }
 466
 467                         if ($skip) {
 468                                 $keepReading = $this->reader->next();
 469                                 $skip = false;
 470                                 $this->debug( "Skip" );
 471                         } else {
 472                                 $keepReading = $this->reader->read();
 473                         }
 474                 }
 475
 476                 return true;
 477         }
 478
 479         /**
 480          * @return bool
 481          * @throws MWException
 482          */
 483         private function handleSiteInfo() {
 484                 // Site info is useful, but not actually used for dump imports.
 485                 // Includes a quick short-circuit to save performance.
 486                 if ( ! $this->mSiteInfoCallback ) {
 487                         $this->reader->next();
 488                         return true;
 489                 }
 490                 throw new MWException( "SiteInfo tag is not yet handled, do not set mSiteInfoCallback" );
 491         }
 492
 493         private function handleLogItem() {
 494                 $this->debug( "Enter log item handler." );
 495                 $logInfo = array();
 496
 497                 // Fields that can just be stuffed in the pageInfo object
 498                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 499                                         'logtitle', 'params' );
 500
 501                 while ( $this->reader->read() ) {
 502                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 503                                         $this->reader->name == 'logitem') {
 504                                 break;
 505                         }
 506
 507                         $tag = $this->reader->name;
 508
 509                         if ( !wfRunHooks( 'ImportHandleLogItemXMLTag',
 510                                                 $this, $logInfo ) ) {
 511                                 // Do nothing
 512                         } elseif ( in_array( $tag, $normalFields ) ) {
 513                                 $logInfo[$tag] = $this->nodeContents();
 514                         } elseif ( $tag == 'contributor' ) {
 515                                 $logInfo['contributor'] = $this->handleContributor();
 516                         } elseif ( $tag != '#text' ) {
 517                                 $this->warn( "Unhandled log-item XML tag $tag" );
 518                         }
 519                 }
 520
 521                 $this->processLogItem( $logInfo );
 522         }
 523
 524         /**
 525          * @param $logInfo
 526          * @return bool|mixed
 527          */
 528         private function processLogItem( $logInfo ) {
 529                 $revision = new WikiRevision;
 530
 531                 $revision->setID( $logInfo['id'] );
 532                 $revision->setType( $logInfo['type'] );
 533                 $revision->setAction( $logInfo['action'] );
 534                 $revision->setTimestamp( $logInfo['timestamp'] );
 535                 $revision->setParams( $logInfo['params'] );
 536                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 537                 $revision->setNoUpdates( $this->mNoUpdates );
 538
 539                 if ( isset( $logInfo['comment'] ) ) {
 540                         $revision->setComment( $logInfo['comment'] );
 541                 }
 542
 543                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 544                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 545                 }
 546                 if ( isset( $logInfo['contributor']['username'] ) ) {
 547                         $revision->setUserName( $logInfo['contributor']['username'] );
 548                 }
 549
 550                 return $this->logItemCallback( $revision );
 551         }
 552
 553         private function handlePage() {
 554                 // Handle page data.
 555                 $this->debug( "Enter page handler." );
 556                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 557
 558                 // Fields that can just be stuffed in the pageInfo object
 559                 $normalFields = array( 'title', 'id', 'redirect', 'restrictions' );
 560
 561                 $skip = false;
 562                 $badTitle = false;
 563
 564                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 565                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 566                                         $this->reader->name == 'page') {
 567                                 break;
 568                         }
 569
 570                         $tag = $this->reader->name;
 571
 572                         if ( $badTitle ) {
 573                                 // The title is invalid, bail out of this page
 574                                 $skip = true;
 575                         } elseif ( !wfRunHooks( 'ImportHandlePageXMLTag', array( $this,
 576                                                 &$pageInfo ) ) ) {
 577                                 // Do nothing
 578                         } elseif ( in_array( $tag, $normalFields ) ) {
 579                                 $pageInfo[$tag] = $this->nodeContents();
 580                                 if ( $tag == 'title' ) {
 581                                         $title = $this->processTitle( $pageInfo['title'] );
 582
 583                                         if ( !$title ) {
 584                                                 $badTitle = true;
 585                                                 $skip = true;
 586                                         }
 587
 588                                         $this->pageCallback( $title );
 589                                         list( $pageInfo['_title'], $origTitle ) = $title;
 590                                 }
 591                         } elseif ( $tag == 'revision' ) {
 592                                 $this->handleRevision( $pageInfo );
 593                         } elseif ( $tag == 'upload' ) {
 594                                 $this->handleUpload( $pageInfo );
 595                         } elseif ( $tag != '#text' ) {
 596                                 $this->warn( "Unhandled page XML tag $tag" );
 597                                 $skip = true;
 598                         }
 599                 }
 600
 601                 $this->pageOutCallback( $pageInfo['_title'], $origTitle,
 602                                         $pageInfo['revisionCount'],
 603                                         $pageInfo['successfulRevisionCount'],
 604                                         $pageInfo );
 605         }
 606
 607         /**
 608          * @param $pageInfo array
 609          */
 610         private function handleRevision( &$pageInfo ) {
 611                 $this->debug( "Enter revision handler" );
 612                 $revisionInfo = array();
 613
 614                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' );
 615
 616                 $skip = false;
 617
 618                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 619                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 620                                         $this->reader->name == 'revision') {
 621                                 break;
 622                         }
 623
 624                         $tag = $this->reader->name;
 625
 626                         if ( !wfRunHooks( 'ImportHandleRevisionXMLTag', $this,
 627                                                 $pageInfo, $revisionInfo ) ) {
 628                                 // Do nothing
 629                         } elseif ( in_array( $tag, $normalFields ) ) {
 630                                 $revisionInfo[$tag] = $this->nodeContents();
 631                         } elseif ( $tag == 'contributor' ) {
 632                                 $revisionInfo['contributor'] = $this->handleContributor();
 633                         } elseif ( $tag != '#text' ) {
 634                                 $this->warn( "Unhandled revision XML tag $tag" );
 635                                 $skip = true;
 636                         }
 637                 }
 638
 639                 $pageInfo['revisionCount']++;
 640                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 641                         $pageInfo['successfulRevisionCount']++;
 642                 }
 643         }
 644
 645         /**
 646          * @param $pageInfo
 647          * @param $revisionInfo
 648          * @return bool|mixed
 649          */
 650         private function processRevision( $pageInfo, $revisionInfo ) {
 651                 $revision = new WikiRevision;
 652
 653                 if( isset( $revisionInfo['id'] ) ) {
 654                         $revision->setID( $revisionInfo['id'] );
 655                 }
 656                 if ( isset( $revisionInfo['text'] ) ) {
 657                         $revision->setText( $revisionInfo['text'] );
 658                 }
 659                 if ( isset( $revisionInfo['model'] ) ) {
 660                         $revision->setModel( $revisionInfo['model'] );
 661                 }
 662                 if ( isset( $revisionInfo['text'] ) ) {
 663                         $revision->setFormat( $revisionInfo['format'] );
 664                 }
 665                 $revision->setTitle( $pageInfo['_title'] );
 666
 667                 if ( isset( $revisionInfo['timestamp'] ) ) {
 668                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 669                 } else {
 670                         $revision->setTimestamp( wfTimestampNow() );
 671                 }
 672
 673                 if ( isset( $revisionInfo['comment'] ) ) {
 674                         $revision->setComment( $revisionInfo['comment'] );
 675                 }
 676
 677                 if ( isset( $revisionInfo['minor'] ) ) {
 678                         $revision->setMinor( true );
 679                 }
 680                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 681                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 682                 }
 683                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 684                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 685                 }
 686                 $revision->setNoUpdates( $this->mNoUpdates );
 687
 688                 return $this->revisionCallback( $revision );
 689         }
 690
 691         /**
 692          * @param $pageInfo
 693          * @return mixed
 694          */
 695         private function handleUpload( &$pageInfo ) {
 696                 $this->debug( "Enter upload handler" );
 697                 $uploadInfo = array();
 698
 699                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 700                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 701
 702                 $skip = false;
 703
 704                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 705                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 706                                         $this->reader->name == 'upload') {
 707                                 break;
 708                         }
 709
 710                         $tag = $this->reader->name;
 711
 712                         if ( !wfRunHooks( 'ImportHandleUploadXMLTag', $this,
 713                                                 $pageInfo ) ) {
 714                                 // Do nothing
 715                         } elseif ( in_array( $tag, $normalFields ) ) {
 716                                 $uploadInfo[$tag] = $this->nodeContents();
 717                         } elseif ( $tag == 'contributor' ) {
 718                                 $uploadInfo['contributor'] = $this->handleContributor();
 719                         } elseif ( $tag == 'contents' ) {
 720                                 $contents = $this->nodeContents();
 721                                 $encoding = $this->reader->getAttribute( 'encoding' );
 722                                 if ( $encoding === 'base64' ) {
 723                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 724                                         $uploadInfo['isTempSrc'] = true;
 725                                 }
 726                         } elseif ( $tag != '#text' ) {
 727                                 $this->warn( "Unhandled upload XML tag $tag" );
 728                                 $skip = true;
 729                         }
 730                 }
 731
 732                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 733                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 734                         if ( file_exists( $path ) ) {
 735                                 $uploadInfo['fileSrc'] = $path;
 736                                 $uploadInfo['isTempSrc'] = false;
 737                         }
 738                 }
 739
 740                 if ( $this->mImportUploads ) {
 741                         return $this->processUpload( $pageInfo, $uploadInfo );
 742                 }
 743         }
 744
 745         /**
 746          * @param $contents
 747          * @return string
 748          */
 749         private function dumpTemp( $contents ) {
 750                 $filename = tempnam( wfTempDir(), 'importupload' );
 751                 file_put_contents( $filename, $contents );
 752                 return $filename;
 753         }
 754
 755         /**
 756          * @param $pageInfo
 757          * @param $uploadInfo
 758          * @return mixed
 759          */
 760         private function processUpload( $pageInfo, $uploadInfo ) {
 761                 $revision = new WikiRevision;
 762                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 763
 764                 $revision->setTitle( $pageInfo['_title'] );
 765                 $revision->setID( $pageInfo['id'] );
 766                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 767                 $revision->setText( $text );
 768                 $revision->setFilename( $uploadInfo['filename'] );
 769                 if ( isset( $uploadInfo['archivename'] ) ) {
 770                         $revision->setArchiveName( $uploadInfo['archivename'] );
 771                 }
 772                 $revision->setSrc( $uploadInfo['src'] );
 773                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 774                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 775                                 !empty( $uploadInfo['isTempSrc'] ) );
 776                 }
 777                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 778                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 779                 }
 780                 $revision->setSize( intval( $uploadInfo['size'] ) );
 781                 $revision->setComment( $uploadInfo['comment'] );
 782
 783                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 784                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 785                 }
 786                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 787                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 788                 }
 789                 $revision->setNoUpdates( $this->mNoUpdates );
 790
 791                 return call_user_func( $this->mUploadCallback, $revision );
 792         }
 793
 794         /**
 795          * @return array
 796          */
 797         private function handleContributor() {
 798                 $fields = array( 'id', 'ip', 'username' );
 799                 $info = array();
 800
 801                 while ( $this->reader->read() ) {
 802                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 803                                         $this->reader->name == 'contributor') {
 804                                 break;
 805                         }
 806
 807                         $tag = $this->reader->name;
 808
 809                         if ( in_array( $tag, $fields ) ) {
 810                                 $info[$tag] = $this->nodeContents();
 811                         }
 812                 }
 813
 814                 return $info;
 815         }
 816
 817         /**
 818          * @param $text string
 819          * @return Array or false
 820          */
 821         private function processTitle( $text ) {
 822                 global $wgCommandLineMode;
 823
 824                 $workTitle = $text;
 825                 $origTitle = Title::newFromText( $workTitle );
 826
 827                 if( !is_null( $this->mTargetNamespace ) && !is_null( $origTitle ) ) {
 828                         # makeTitleSafe, because $origTitle can have a interwiki (different setting of interwiki map)
 829                         # and than dbKey can begin with a lowercase char
 830                         $title = Title::makeTitleSafe( $this->mTargetNamespace,
 831                                 $origTitle->getDBkey() );
 832                 } else {
 833                         if( !is_null( $this->mTargetRootPage ) ) {
 834                                 $workTitle = $this->mTargetRootPage . '/' . $workTitle;
 835                         }
 836                         $title = Title::newFromText( $workTitle );
 837                 }
 838
 839                 if( is_null( $title ) ) {
 840                         # Invalid page title? Ignore the page
 841                         $this->notice( 'import-error-invalid', $workTitle );
 842                         return false;
 843                 } elseif( $title->isExternal() ) {
 844                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
 845                         return false;
 846                 } elseif( !$title->canExist() ) {
 847                         $this->notice( 'import-error-special', $title->getPrefixedText() );
 848                         return false;
 849                 } elseif( !$title->userCan( 'edit' ) && !$wgCommandLineMode ) {
 850                         # Do not import if the importing wiki user cannot edit this page
 851                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
 852                         return false;
 853                 } elseif( !$title->exists() && !$title->userCan( 'create' ) && !$wgCommandLineMode ) {
 854                         # Do not import if the importing wiki user cannot create this page
 855                         $this->notice( 'import-error-create', $title->getPrefixedText() );
 856                         return false;
 857                 }
 858
 859                 return array( $title, $origTitle );
 860         }
 861 }
 862
 863 /** This is a horrible hack used to keep source compatibility */
 864 class UploadSourceAdapter {
 865         static $sourceRegistrations = array();
 866
 867         private $mSource;
 868         private $mBuffer;
 869         private $mPosition;
 870
 871         /**
 872          * @param $source
 873          * @return string
 874          */
 875         static function registerSource( $source ) {
 876                 $id = wfRandomString();
 877
 878                 self::$sourceRegistrations[$id] = $source;
 879
 880                 return $id;
 881         }
 882
 883         /**
 884          * @param $path
 885          * @param $mode
 886          * @param $options
 887          * @param $opened_path
 888          * @return bool
 889          */
 890         function stream_open( $path, $mode, $options, &$opened_path ) {
 891                 $url = parse_url($path);
 892                 $id = $url['host'];
 893
 894                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
 895                         return false;
 896                 }
 897
 898                 $this->mSource = self::$sourceRegistrations[$id];
 899
 900                 return true;
 901         }
 902
 903         /**
 904          * @param $count
 905          * @return string
 906          */
 907         function stream_read( $count ) {
 908                 $return = '';
 909                 $leave = false;
 910
 911                 while ( !$leave && !$this->mSource->atEnd() &&
 912                                 strlen($this->mBuffer) < $count ) {
 913                         $read = $this->mSource->readChunk();
 914
 915                         if ( !strlen($read) ) {
 916                                 $leave = true;
 917                         }
 918
 919                         $this->mBuffer .= $read;
 920                 }
 921
 922                 if ( strlen($this->mBuffer) ) {
 923                         $return = substr( $this->mBuffer, 0, $count );
 924                         $this->mBuffer = substr( $this->mBuffer, $count );
 925                 }
 926
 927                 $this->mPosition += strlen($return);
 928
 929                 return $return;
 930         }
 931
 932         /**
 933          * @param $data
 934          * @return bool
 935          */
 936         function stream_write( $data ) {
 937                 return false;
 938         }
 939
 940         /**
 941          * @return mixed
 942          */
 943         function stream_tell() {
 944                 return $this->mPosition;
 945         }
 946
 947         /**
 948          * @return bool
 949          */
 950         function stream_eof() {
 951                 return $this->mSource->atEnd();
 952         }
 953
 954         /**
 955          * @return array
 956          */
 957         function url_stat() {
 958                 $result = array();
 959
 960                 $result['dev'] = $result[0] = 0;
 961                 $result['ino'] = $result[1] = 0;
 962                 $result['mode'] = $result[2] = 0;
 963                 $result['nlink'] = $result[3] = 0;
 964                 $result['uid'] = $result[4] = 0;
 965                 $result['gid'] = $result[5] = 0;
 966                 $result['rdev'] = $result[6] = 0;
 967                 $result['size'] = $result[7] = 0;
 968                 $result['atime'] = $result[8] = 0;
 969                 $result['mtime'] = $result[9] = 0;
 970                 $result['ctime'] = $result[10] = 0;
 971                 $result['blksize'] = $result[11] = 0;
 972                 $result['blocks'] = $result[12] = 0;
 973
 974                 return $result;
 975         }
 976 }
 977
 978 class XMLReader2 extends XMLReader {
 979
 980         /**
 981          * @return bool|string
 982          */
 983         function nodeContents() {
 984                 if( $this->isEmptyElement ) {
 985                         return "";
 986                 }
 987                 $buffer = "";
 988                 while( $this->read() ) {
 989                         switch( $this->nodeType ) {
 990                         case XmlReader::TEXT:
 991                         case XmlReader::SIGNIFICANT_WHITESPACE:
 992                                 $buffer .= $this->value;
 993                                 break;
 994                         case XmlReader::END_ELEMENT:
 995                                 return $buffer;
 996                         }
 997                 }
 998                 return $this->close();
 999         }
1000 }
1001
1002 /**
1003  * @todo document (e.g. one-sentence class description).
1004  * @ingroup SpecialPage
1005  */
1006 class WikiRevision {
1007         var $importer = null;
1008
1009         /**
1010          * @var Title
1011          */
1012         var $title = null;
1013         var $id = 0;
1014         var $timestamp = "20010115000000";
1015         var $user = 0;
1016         var $user_text = "";
1017         var $model = null;
1018         var $format = null;
1019         var $text = "";
1020         var $content = null;
1021         var $comment = "";
1022         var $minor = false;
1023         var $type = "";
1024         var $action = "";
1025         var $params = "";
1026         var $fileSrc = '';
1027         var $sha1base36 = false;
1028         var $isTemp = false;
1029         var $archiveName = '';
1030         var $fileIsTemp;
1031         private $mNoUpdates = false;
1032
1033         /**
1034          * @param $title
1035          * @throws MWException
1036          */
1037         function setTitle( $title ) {
1038                 if( is_object( $title ) ) {
1039                         $this->title = $title;
1040                 } elseif( is_null( $title ) ) {
1041                         throw new MWException( "WikiRevision given a null title in import. You may need to adjust \$wgLegalTitleChars." );
1042                 } else {
1043                         throw new MWException( "WikiRevision given non-object title in import." );
1044                 }
1045         }
1046
1047         /**
1048          * @param $id
1049          */
1050         function setID( $id ) {
1051                 $this->id = $id;
1052         }
1053
1054         /**
1055          * @param $ts
1056          */
1057         function setTimestamp( $ts ) {
1058                 # 2003-08-05T18:30:02Z
1059                 $this->timestamp = wfTimestamp( TS_MW, $ts );
1060         }
1061
1062         /**
1063          * @param $user
1064          */
1065         function setUsername( $user ) {
1066                 $this->user_text = $user;
1067         }
1068
1069         /**
1070          * @param $ip
1071          */
1072         function setUserIP( $ip ) {
1073                 $this->user_text = $ip;
1074         }
1075
1076         /**
1077          * @param $model
1078          */
1079         function setModel( $model ) {
1080                 $this->model = $model;
1081         }
1082
1083         /**
1084          * @param $format
1085          */
1086         function setFormat( $format ) {
1087                 $this->format = $format;
1088         }
1089
1090         /**
1091          * @param $text
1092          */
1093         function setText( $text ) {
1094                 $this->text = $text;
1095         }
1096
1097         /**
1098          * @param $text
1099          */
1100         function setComment( $text ) {
1101                 $this->comment = $text;
1102         }
1103
1104         /**
1105          * @param $minor
1106          */
1107         function setMinor( $minor ) {
1108                 $this->minor = (bool)$minor;
1109         }
1110
1111         /**
1112          * @param $src
1113          */
1114         function setSrc( $src ) {
1115                 $this->src = $src;
1116         }
1117
1118         /**
1119          * @param $src
1120          * @param $isTemp
1121          */
1122         function setFileSrc( $src, $isTemp ) {
1123                 $this->fileSrc = $src;
1124                 $this->fileIsTemp = $isTemp;
1125         }
1126
1127         /**
1128          * @param $sha1base36
1129          */
1130         function setSha1Base36( $sha1base36 ) {
1131                 $this->sha1base36 = $sha1base36;
1132         }
1133
1134         /**
1135          * @param $filename
1136          */
1137         function setFilename( $filename ) {
1138                 $this->filename = $filename;
1139         }
1140
1141         /**
1142          * @param $archiveName
1143          */
1144         function setArchiveName( $archiveName ) {
1145                 $this->archiveName = $archiveName;
1146         }
1147
1148         /**
1149          * @param $size
1150          */
1151         function setSize( $size ) {
1152                 $this->size = intval( $size );
1153         }
1154
1155         /**
1156          * @param $type
1157          */
1158         function setType( $type ) {
1159                 $this->type = $type;
1160         }
1161
1162         /**
1163          * @param $action
1164          */
1165         function setAction( $action ) {
1166                 $this->action = $action;
1167         }
1168
1169         /**
1170          * @param $params
1171          */
1172         function setParams( $params ) {
1173                 $this->params = $params;
1174         }
1175
1176         /**
1177          * @param $noupdates
1178          */
1179         public function setNoUpdates( $noupdates ) {
1180                 $this->mNoUpdates = $noupdates;
1181         }
1182
1183         /**
1184          * @return Title
1185          */
1186         function getTitle() {
1187                 return $this->title;
1188         }
1189
1190         /**
1191          * @return int
1192          */
1193         function getID() {
1194                 return $this->id;
1195         }
1196
1197         /**
1198          * @return string
1199          */
1200         function getTimestamp() {
1201                 return $this->timestamp;
1202         }
1203
1204         /**
1205          * @return string
1206          */
1207         function getUser() {
1208                 return $this->user_text;
1209         }
1210
1211         /**
1212          * @return string
1213          *
1214          * @deprecated Since 1.21, use getContent() instead.
1215          */
1216         function getText() {
1217                 wfDeprecated( "Use getContent() instead." );
1218
1219                 return $this->text;
1220         }
1221
1222         /**
1223          * @return Content
1224          */
1225         function getContent() {
1226                 if ( is_null( $this->content ) ) {
1227                         $this->content =
1228                                 ContentHandler::makeContent(
1229                                         $this->text,
1230                                         $this->getTitle(),
1231                                         $this->getModel(),
1232                                         $this->getFormat()
1233                                 );
1234                 }
1235
1236                 return $this->content;
1237         }
1238
1239         /**
1240          * @return String
1241          */
1242         function getModel() {
1243                 if ( is_null( $this->model ) ) {
1244                         $this->model = $this->getTitle()->getContentModel();
1245                 }
1246
1247                 return $this->model;
1248         }
1249
1250         /**
1251          * @return String
1252          */
1253         function getFormat() {
1254                 if ( is_null( $this->model ) ) {
1255                         $this->format = ContentHandler::getForTitle( $this->getTitle() )->getDefaultFormat();
1256                 }
1257
1258                 return $this->format;
1259         }
1260
1261         /**
1262          * @return string
1263          */
1264         function getComment() {
1265                 return $this->comment;
1266         }
1267
1268         /**
1269          * @return bool
1270          */
1271         function getMinor() {
1272                 return $this->minor;
1273         }
1274
1275         /**
1276          * @return mixed
1277          */
1278         function getSrc() {
1279                 return $this->src;
1280         }
1281
1282         /**
1283          * @return bool|String
1284          */
1285         function getSha1() {
1286                 if ( $this->sha1base36 ) {
1287                         return wfBaseConvert( $this->sha1base36, 36, 16 );
1288                 }
1289                 return false;
1290         }
1291
1292         /**
1293          * @return string
1294          */
1295         function getFileSrc() {
1296                 return $this->fileSrc;
1297         }
1298
1299         /**
1300          * @return bool
1301          */
1302         function isTempSrc() {
1303                 return $this->isTemp;
1304         }
1305
1306         /**
1307          * @return mixed
1308          */
1309         function getFilename() {
1310                 return $this->filename;
1311         }
1312
1313         /**
1314          * @return string
1315          */
1316         function getArchiveName() {
1317                 return $this->archiveName;
1318         }
1319
1320         /**
1321          * @return mixed
1322          */
1323         function getSize() {
1324                 return $this->size;
1325         }
1326
1327         /**
1328          * @return string
1329          */
1330         function getType() {
1331                 return $this->type;
1332         }
1333
1334         /**
1335          * @return string
1336          */
1337         function getAction() {
1338                 return $this->action;
1339         }
1340
1341         /**
1342          * @return string
1343          */
1344         function getParams() {
1345                 return $this->params;
1346         }
1347
1348         /**
1349          * @return bool
1350          */
1351         function importOldRevision() {
1352                 $dbw = wfGetDB( DB_MASTER );
1353
1354                 # Sneak a single revision into place
1355                 $user = User::newFromName( $this->getUser() );
1356                 if( $user ) {
1357                         $userId = intval( $user->getId() );
1358                         $userText = $user->getName();
1359                         $userObj = $user;
1360                 } else {
1361                         $userId = 0;
1362                         $userText = $this->getUser();
1363                         $userObj = new User;
1364                 }
1365
1366                 // avoid memory leak...?
1367                 $linkCache = LinkCache::singleton();
1368                 $linkCache->clear();
1369
1370                 $page = WikiPage::factory( $this->title );
1371                 if( !$page->exists() ) {
1372                         # must create the page...
1373                         $pageId = $page->insertOn( $dbw );
1374                         $created = true;
1375                         $oldcountable = null;
1376                 } else {
1377                         $pageId = $page->getId();
1378                         $created = false;
1379
1380                         $prior = $dbw->selectField( 'revision', '1',
1381                                 array( 'rev_page' => $pageId,
1382                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1383                                         'rev_user_text' => $userText,
1384                                         'rev_comment'   => $this->getComment() ),
1385                                 __METHOD__
1386                         );
1387                         if( $prior ) {
1388                                 // @todo FIXME: This could fail slightly for multiple matches :P
1389                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1390                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1391                                 return false;
1392                         }
1393                         $oldcountable = $page->isCountable();
1394                 }
1395
1396                 # @todo FIXME: Use original rev_id optionally (better for backups)
1397                 # Insert the row
1398                 $revision = new Revision( array(
1399                         'page'       => $pageId,
1400                         'content_model'  => $this->getModel(),
1401                         'content_format' => $this->getFormat(),
1402                         'text'       => $this->getContent()->serialize( $this->getFormat() ), //XXX: just set 'content' => $this->getContent()?
1403                         'comment'    => $this->getComment(),
1404                         'user'       => $userId,
1405                         'user_text'  => $userText,
1406                         'timestamp'  => $this->timestamp,
1407                         'minor_edit' => $this->minor,
1408                         ) );
1409                 $revision->insertOn( $dbw );
1410                 $changed = $page->updateIfNewerOn( $dbw, $revision );
1411
1412                 if ( $changed !== false && !$this->mNoUpdates ) {
1413                         wfDebug( __METHOD__ . ": running updates\n" );
1414                         $page->doEditUpdates( $revision, $userObj, array( 'created' => $created, 'oldcountable' => $oldcountable ) );
1415                 }
1416
1417                 return true;
1418         }
1419
1420         /**
1421          * @return mixed
1422          */
1423         function importLogItem() {
1424                 $dbw = wfGetDB( DB_MASTER );
1425                 # @todo FIXME: This will not record autoblocks
1426                 if( !$this->getTitle() ) {
1427                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1428                                 $this->timestamp . "\n" );
1429                         return;
1430                 }
1431                 # Check if it exists already
1432                 // @todo FIXME: Use original log ID (better for backups)
1433                 $prior = $dbw->selectField( 'logging', '1',
1434                         array( 'log_type' => $this->getType(),
1435                                 'log_action'    => $this->getAction(),
1436                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1437                                 'log_namespace' => $this->getTitle()->getNamespace(),
1438                                 'log_title'     => $this->getTitle()->getDBkey(),
1439                                 'log_comment'   => $this->getComment(),
1440                                 #'log_user_text' => $this->user_text,
1441                                 'log_params'    => $this->params ),
1442                         __METHOD__
1443                 );
1444                 // @todo FIXME: This could fail slightly for multiple matches :P
1445                 if( $prior ) {
1446                         wfDebug( __METHOD__ . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp " .
1447                                 $this->timestamp . "\n" );
1448                         return;
1449                 }
1450                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1451                 $data = array(
1452                         'log_id' => $log_id,
1453                         'log_type' => $this->type,
1454                         'log_action' => $this->action,
1455                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1456                         'log_user' => User::idFromName( $this->user_text ),
1457                         #'log_user_text' => $this->user_text,
1458                         'log_namespace' => $this->getTitle()->getNamespace(),
1459                         'log_title' => $this->getTitle()->getDBkey(),
1460                         'log_comment' => $this->getComment(),
1461                         'log_params' => $this->params
1462                 );
1463                 $dbw->insert( 'logging', $data, __METHOD__ );
1464         }
1465
1466         /**
1467          * @return bool
1468          */
1469         function importUpload() {
1470                 # Construct a file
1471                 $archiveName = $this->getArchiveName();
1472                 if ( $archiveName ) {
1473                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1474                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1475                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1476                 } else {
1477                         $file = wfLocalFile( $this->getTitle() );
1478                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1479                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1480                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1481                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1482                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1483                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1484                         }
1485                 }
1486                 if( !$file ) {
1487                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1488                         return false;
1489                 }
1490
1491                 # Get the file source or download if necessary
1492                 $source = $this->getFileSrc();
1493                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1494                 if ( !$source ) {
1495                         $source = $this->downloadSource();
1496                         $flags |= File::DELETE_SOURCE;
1497                 }
1498                 if( !$source ) {
1499                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1500                         return false;
1501                 }
1502                 $sha1 = $this->getSha1();
1503                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1504                         if ( $flags & File::DELETE_SOURCE ) {
1505                                 # Broken file; delete it if it is a temporary file
1506                                 unlink( $source );
1507                         }
1508                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1509                         return false;
1510                 }
1511
1512                 $user = User::newFromName( $this->user_text );
1513
1514                 # Do the actual upload
1515                 if ( $archiveName ) {
1516                         $status = $file->uploadOld( $source, $archiveName,
1517                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1518                 } else {
1519                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1520                                 $flags, false, $this->getTimestamp(), $user );
1521                 }
1522
1523                 if ( $status->isGood() ) {
1524                         wfDebug( __METHOD__ . ": Succesful\n" );
1525                         return true;
1526                 } else {
1527                         wfDebug( __METHOD__ . ': failed: ' . $status->getXml() . "\n" );
1528                         return false;
1529                 }
1530         }
1531
1532         /**
1533          * @return bool|string
1534          */
1535         function downloadSource() {
1536                 global $wgEnableUploads;
1537                 if( !$wgEnableUploads ) {
1538                         return false;
1539                 }
1540
1541                 $tempo = tempnam( wfTempDir(), 'download' );
1542                 $f = fopen( $tempo, 'wb' );
1543                 if( !$f ) {
1544                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1545                         return false;
1546                 }
1547
1548                 // @todo FIXME!
1549                 $src = $this->getSrc();
1550                 $data = Http::get( $src );
1551                 if( !$data ) {
1552                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1553                         fclose( $f );
1554                         unlink( $tempo );
1555                         return false;
1556                 }
1557
1558                 fwrite( $f, $data );
1559                 fclose( $f );
1560
1561                 return $tempo;
1562         }
1563
1564 }
1565
1566 /**
1567  * @todo document (e.g. one-sentence class description).
1568  * @ingroup SpecialPage
1569  */
1570 class ImportStringSource {
1571         function __construct( $string ) {
1572                 $this->mString = $string;
1573                 $this->mRead = false;
1574         }
1575
1576         /**
1577          * @return bool
1578          */
1579         function atEnd() {
1580                 return $this->mRead;
1581         }
1582
1583         /**
1584          * @return bool|string
1585          */
1586         function readChunk() {
1587                 if( $this->atEnd() ) {
1588                         return false;
1589                 }
1590                 $this->mRead = true;
1591                 return $this->mString;
1592         }
1593 }
1594
1595 /**
1596  * @todo document (e.g. one-sentence class description).
1597  * @ingroup SpecialPage
1598  */
1599 class ImportStreamSource {
1600         function __construct( $handle ) {
1601                 $this->mHandle = $handle;
1602         }
1603
1604         /**
1605          * @return bool
1606          */
1607         function atEnd() {
1608                 return feof( $this->mHandle );
1609         }
1610
1611         /**
1612          * @return string
1613          */
1614         function readChunk() {
1615                 return fread( $this->mHandle, 32768 );
1616         }
1617
1618         /**
1619          * @param $filename string
1620          * @return Status
1621          */
1622         static function newFromFile( $filename ) {
1623                 wfSuppressWarnings();
1624                 $file = fopen( $filename, 'rt' );
1625                 wfRestoreWarnings();
1626                 if( !$file ) {
1627                         return Status::newFatal( "importcantopen" );
1628                 }
1629                 return Status::newGood( new ImportStreamSource( $file ) );
1630         }
1631
1632         /**
1633          * @param $fieldname string
1634          * @return Status
1635          */
1636         static function newFromUpload( $fieldname = "xmlimport" ) {
1637                 $upload =& $_FILES[$fieldname];
1638
1639                 if( !isset( $upload ) || !$upload['name'] ) {
1640                         return Status::newFatal( 'importnofile' );
1641                 }
1642                 if( !empty( $upload['error'] ) ) {
1643                         switch($upload['error']){
1644                                 case 1: # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1645                                         return Status::newFatal( 'importuploaderrorsize' );
1646                                 case 2: # The uploaded file exceeds the MAX_FILE_SIZE directive that was specified in the HTML form.
1647                                         return Status::newFatal( 'importuploaderrorsize' );
1648                                 case 3: # The uploaded file was only partially uploaded
1649                                         return Status::newFatal( 'importuploaderrorpartial' );
1650                                 case 6: #Missing a temporary folder.
1651                                         return Status::newFatal( 'importuploaderrortemp' );
1652                                 # case else: # Currently impossible
1653                         }
1654
1655                 }
1656                 $fname = $upload['tmp_name'];
1657                 if( is_uploaded_file( $fname ) ) {
1658                         return ImportStreamSource::newFromFile( $fname );
1659                 } else {
1660                         return Status::newFatal( 'importnofile' );
1661                 }
1662         }
1663
1664         /**
1665          * @param $url
1666          * @param $method string
1667          * @return Status
1668          */
1669         static function newFromURL( $url, $method = 'GET' ) {
1670                 wfDebug( __METHOD__ . ": opening $url\n" );
1671                 # Use the standard HTTP fetch function; it times out
1672                 # quicker and sorts out user-agent problems which might
1673                 # otherwise prevent importing from large sites, such
1674                 # as the Wikimedia cluster, etc.
1675                 $data = Http::request( $method, $url, array( 'followRedirects' => true ) );
1676                 if( $data !== false ) {
1677                         $file = tmpfile();
1678                         fwrite( $file, $data );
1679                         fflush( $file );
1680                         fseek( $file, 0 );
1681                         return Status::newGood( new ImportStreamSource( $file ) );
1682                 } else {
1683                         return Status::newFatal( 'importcantopen' );
1684                 }
1685         }
1686
1687         /**
1688          * @param $interwiki
1689          * @param $page
1690          * @param $history bool
1691          * @param $templates bool
1692          * @param $pageLinkDepth int
1693          * @return Status
1694          */
1695         public static function newFromInterwiki( $interwiki, $page, $history = false, $templates = false, $pageLinkDepth = 0 ) {
1696                 if( $page == '' ) {
1697                         return Status::newFatal( 'import-noarticle' );
1698                 }
1699                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1700                 if( is_null( $link ) || $link->getInterwiki() == '' ) {
1701                         return Status::newFatal( 'importbadinterwiki' );
1702                 } else {
1703                         $params = array();
1704                         if ( $history ) $params['history'] = 1;
1705                         if ( $templates ) $params['templates'] = 1;
1706                         if ( $pageLinkDepth ) $params['pagelink-depth'] = $pageLinkDepth;
1707                         $url = $link->getFullUrl( $params );
1708                         # For interwikis, use POST to avoid redirects.
1709                         return ImportStreamSource::newFromURL( $url, "POST" );
1710                 }
1711         }
1712 }