includes/Import.php

   1 <?php
   2 /**
   3  * MediaWiki page data importer.
   4  *
   5  * Copyright © 2003,2005 Brion Vibber <brion@pobox.com>
   6  * https://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 $foreignNamespaces = null;
  36         private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
  37         private $mSiteInfoCallback, $mTargetNamespace, $mPageOutCallback;
  38         private $mNoticeCallback, $mDebug;
  39         private $mImportUploads, $mImageBasePath;
  40         private $mNoUpdates = false;
  41         /** @var Config */
  42         private $config;
  43         /** @var ImportTitleFactory */
  44         private $importTitleFactory;
  45         /** @var array */
  46         private $countableCache = array();
  47
  48         /**
  49          * Creates an ImportXMLReader drawing from the source provided
  50          * @param ImportSource $source
  51          * @param Config $config
  52          */
  53         function __construct( ImportSource $source, Config $config = null ) {
  54                 $this->reader = new XMLReader();
  55                 if ( !$config ) {
  56                         wfDeprecated( __METHOD__ . ' without a Config instance', '1.25' );
  57                         $config = ConfigFactory::getDefaultInstance()->makeConfig( 'main' );
  58                 }
  59                 $this->config = $config;
  60
  61                 if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) {
  62                         stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  63                 }
  64                 $id = UploadSourceAdapter::registerSource( $source );
  65                 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
  66                         $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  67                 } else {
  68                         $this->reader->open( "uploadsource://$id" );
  69                 }
  70
  71                 // Default callbacks
  72                 $this->setPageCallback( array( $this, 'beforeImportPage' ) );
  73                 $this->setRevisionCallback( array( $this, "importRevision" ) );
  74                 $this->setUploadCallback( array( $this, 'importUpload' ) );
  75                 $this->setLogItemCallback( array( $this, 'importLogItem' ) );
  76                 $this->setPageOutCallback( array( $this, 'finishImportPage' ) );
  77
  78                 $this->importTitleFactory = new NaiveImportTitleFactory();
  79         }
  80
  81         /**
  82          * @return null|XMLReader
  83          */
  84         public function getReader() {
  85                 return $this->reader;
  86         }
  87
  88         public function throwXmlError( $err ) {
  89                 $this->debug( "FAILURE: $err" );
  90                 wfDebug( "WikiImporter XML error: $err\n" );
  91         }
  92
  93         public function debug( $data ) {
  94                 if ( $this->mDebug ) {
  95                         wfDebug( "IMPORT: $data\n" );
  96                 }
  97         }
  98
  99         public function warn( $data ) {
 100                 wfDebug( "IMPORT: $data\n" );
 101         }
 102
 103         public function notice( $msg /*, $param, ...*/ ) {
 104                 $params = func_get_args();
 105                 array_shift( $params );
 106
 107                 if ( is_callable( $this->mNoticeCallback ) ) {
 108                         call_user_func( $this->mNoticeCallback, $msg, $params );
 109                 } else { # No ImportReporter -> CLI
 110                         echo wfMessage( $msg, $params )->text() . "\n";
 111                 }
 112         }
 113
 114         /**
 115          * Set debug mode...
 116          * @param bool $debug
 117          */
 118         function setDebug( $debug ) {
 119                 $this->mDebug = $debug;
 120         }
 121
 122         /**
 123          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
 124          * @param bool $noupdates
 125          */
 126         function setNoUpdates( $noupdates ) {
 127                 $this->mNoUpdates = $noupdates;
 128         }
 129
 130         /**
 131          * Set a callback that displays notice messages
 132          *
 133          * @param callable $callback
 134          * @return callable
 135          */
 136         public function setNoticeCallback( $callback ) {
 137                 return wfSetVar( $this->mNoticeCallback, $callback );
 138         }
 139
 140         /**
 141          * Sets the action to perform as each new page in the stream is reached.
 142          * @param callable $callback
 143          * @return callable
 144          */
 145         public function setPageCallback( $callback ) {
 146                 $previous = $this->mPageCallback;
 147                 $this->mPageCallback = $callback;
 148                 return $previous;
 149         }
 150
 151         /**
 152          * Sets the action to perform as each page in the stream is completed.
 153          * Callback accepts the page title (as a Title object), a second object
 154          * with the original title form (in case it's been overridden into a
 155          * local namespace), and a count of revisions.
 156          *
 157          * @param callable $callback
 158          * @return callable
 159          */
 160         public function setPageOutCallback( $callback ) {
 161                 $previous = $this->mPageOutCallback;
 162                 $this->mPageOutCallback = $callback;
 163                 return $previous;
 164         }
 165
 166         /**
 167          * Sets the action to perform as each page revision is reached.
 168          * @param callable $callback
 169          * @return callable
 170          */
 171         public function setRevisionCallback( $callback ) {
 172                 $previous = $this->mRevisionCallback;
 173                 $this->mRevisionCallback = $callback;
 174                 return $previous;
 175         }
 176
 177         /**
 178          * Sets the action to perform as each file upload version is reached.
 179          * @param callable $callback
 180          * @return callable
 181          */
 182         public function setUploadCallback( $callback ) {
 183                 $previous = $this->mUploadCallback;
 184                 $this->mUploadCallback = $callback;
 185                 return $previous;
 186         }
 187
 188         /**
 189          * Sets the action to perform as each log item reached.
 190          * @param callable $callback
 191          * @return callable
 192          */
 193         public function setLogItemCallback( $callback ) {
 194                 $previous = $this->mLogItemCallback;
 195                 $this->mLogItemCallback = $callback;
 196                 return $previous;
 197         }
 198
 199         /**
 200          * Sets the action to perform when site info is encountered
 201          * @param callable $callback
 202          * @return callable
 203          */
 204         public function setSiteInfoCallback( $callback ) {
 205                 $previous = $this->mSiteInfoCallback;
 206                 $this->mSiteInfoCallback = $callback;
 207                 return $previous;
 208         }
 209
 210         /**
 211          * Sets the factory object to use to convert ForeignTitle objects into local
 212          * Title objects
 213          * @param ImportTitleFactory $factory
 214          */
 215         public function setImportTitleFactory( $factory ) {
 216                 $this->importTitleFactory = $factory;
 217         }
 218
 219         /**
 220          * Set a target namespace to override the defaults
 221          * @param null|int $namespace
 222          * @return bool
 223          */
 224         public function setTargetNamespace( $namespace ) {
 225                 if ( is_null( $namespace ) ) {
 226                         // Don't override namespaces
 227                         $this->mTargetNamespace = null;
 228                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 229                         return true;
 230                 } elseif (
 231                         $namespace >= 0 &&
 232                         MWNamespace::exists( intval( $namespace ) )
 233                 ) {
 234                         $namespace = intval( $namespace );
 235                         $this->mTargetNamespace = $namespace;
 236                         $this->setImportTitleFactory( new NamespaceImportTitleFactory( $namespace ) );
 237                         return true;
 238                 } else {
 239                         return false;
 240                 }
 241         }
 242
 243         /**
 244          * Set a target root page under which all pages are imported
 245          * @param null|string $rootpage
 246          * @return Status
 247          */
 248         public function setTargetRootPage( $rootpage ) {
 249                 $status = Status::newGood();
 250                 if ( is_null( $rootpage ) ) {
 251                         // No rootpage
 252                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 253                 } elseif ( $rootpage !== '' ) {
 254                         $rootpage = rtrim( $rootpage, '/' ); //avoid double slashes
 255                         $title = Title::newFromText( $rootpage, !is_null( $this->mTargetNamespace )
 256                                 ? $this->mTargetNamespace
 257                                 : NS_MAIN
 258                         );
 259
 260                         if ( !$title || $title->isExternal() ) {
 261                                 $status->fatal( 'import-rootpage-invalid' );
 262                         } else {
 263                                 if ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 264                                         global $wgContLang;
 265
 266                                         $displayNSText = $title->getNamespace() == NS_MAIN
 267                                                 ? wfMessage( 'blanknamespace' )->text()
 268                                                 : $wgContLang->getNsText( $title->getNamespace() );
 269                                         $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 270                                 } else {
 271                                         // set namespace to 'all', so the namespace check in processTitle() can pass
 272                                         $this->setTargetNamespace( null );
 273                                         $this->setImportTitleFactory( new SubpageImportTitleFactory( $title ) );
 274                                 }
 275                         }
 276                 }
 277                 return $status;
 278         }
 279
 280         /**
 281          * @param string $dir
 282          */
 283         public function setImageBasePath( $dir ) {
 284                 $this->mImageBasePath = $dir;
 285         }
 286
 287         /**
 288          * @param bool $import
 289          */
 290         public function setImportUploads( $import ) {
 291                 $this->mImportUploads = $import;
 292         }
 293
 294         /**
 295          * Default per-page callback. Sets up some things related to site statistics
 296          * @param array $titleAndForeignTitle Two-element array, with Title object at
 297          * index 0 and ForeignTitle object at index 1
 298          * @return bool
 299          */
 300         public function beforeImportPage( $titleAndForeignTitle ) {
 301                 $title = $titleAndForeignTitle[0];
 302                 $page = WikiPage::factory( $title );
 303                 $this->countableCache['title_' . $title->getPrefixedText()] = $page->isCountable();
 304                 return true;
 305         }
 306
 307         /**
 308          * Default per-revision callback, performs the import.
 309          * @param WikiRevision $revision
 310          * @return bool
 311          */
 312         public function importRevision( $revision ) {
 313                 if ( !$revision->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
 314                         $this->notice( 'import-error-bad-location',
 315                                 $revision->getTitle()->getPrefixedText(),
 316                                 $revision->getID(),
 317                                 $revision->getModel(),
 318                                 $revision->getFormat() );
 319
 320                         return false;
 321                 }
 322
 323                 try {
 324                         $dbw = wfGetDB( DB_MASTER );
 325                         return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
 326                 } catch ( MWContentSerializationException $ex ) {
 327                         $this->notice( 'import-error-unserialize',
 328                                 $revision->getTitle()->getPrefixedText(),
 329                                 $revision->getID(),
 330                                 $revision->getModel(),
 331                                 $revision->getFormat() );
 332                 }
 333
 334                 return false;
 335         }
 336
 337         /**
 338          * Default per-revision callback, performs the import.
 339          * @param WikiRevision $revision
 340          * @return bool
 341          */
 342         public function importLogItem( $revision ) {
 343                 $dbw = wfGetDB( DB_MASTER );
 344                 return $dbw->deadlockLoop( array( $revision, 'importLogItem' ) );
 345         }
 346
 347         /**
 348          * Dummy for now...
 349          * @param WikiRevision $revision
 350          * @return bool
 351          */
 352         public function importUpload( $revision ) {
 353                 $dbw = wfGetDB( DB_MASTER );
 354                 return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
 355         }
 356
 357         /**
 358          * Mostly for hook use
 359          * @param Title $title
 360          * @param ForeignTitle $foreignTitle
 361          * @param int $revCount
 362          * @param int $sRevCount
 363          * @param array $pageInfo
 364          * @return bool
 365          */
 366         public function finishImportPage( $title, $foreignTitle, $revCount,
 367                         $sRevCount, $pageInfo ) {
 368
 369                 // Update article count statistics (T42009)
 370                 // The normal counting logic in WikiPage->doEditUpdates() is designed for
 371                 // one-revision-at-a-time editing, not bulk imports. In this situation it
 372                 // suffers from issues of slave lag. We let WikiPage handle the total page
 373                 // and revision count, and we implement our own custom logic for the
 374                 // article (content page) count.
 375                 $page = WikiPage::factory( $title );
 376                 $page->loadPageData( 'fromdbmaster' );
 377                 $content = $page->getContent();
 378                 $editInfo = $page->prepareContentForEdit( $content );
 379                 $countKey = 'title_' . $title->getPrefixedText();
 380                 $countable = $page->isCountable( $editInfo );
 381                 if ( array_key_exists( $countKey, $this->countableCache ) &&
 382                         $countable != $this->countableCache[ $countKey ] ) {
 383                         DeferredUpdates::addUpdate( SiteStatsUpdate::factory( array(
 384                                 'articles' => ( (int)$countable - (int)$this->countableCache[ $countKey ] )
 385                         ) ) );
 386                 }
 387
 388                 $args = func_get_args();
 389                 return Hooks::run( 'AfterImportPage', $args );
 390         }
 391
 392         /**
 393          * Alternate per-revision callback, for debugging.
 394          * @param WikiRevision $revision
 395          */
 396         public function debugRevisionHandler( &$revision ) {
 397                 $this->debug( "Got revision:" );
 398                 if ( is_object( $revision->title ) ) {
 399                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 400                 } else {
 401                         $this->debug( "-- Title: <invalid>" );
 402                 }
 403                 $this->debug( "-- User: " . $revision->user_text );
 404                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 405                 $this->debug( "-- Comment: " . $revision->comment );
 406                 $this->debug( "-- Text: " . $revision->text );
 407         }
 408
 409         /**
 410          * Notify the callback function of site info
 411          * @param array $siteInfo
 412          * @return bool|mixed
 413          */
 414         private function siteInfoCallback( $siteInfo ) {
 415                 if ( isset( $this->mSiteInfoCallback ) ) {
 416                         return call_user_func_array( $this->mSiteInfoCallback,
 417                                         array( $siteInfo, $this ) );
 418                 } else {
 419                         return false;
 420                 }
 421         }
 422
 423         /**
 424          * Notify the callback function when a new "<page>" is reached.
 425          * @param Title $title
 426          */
 427         function pageCallback( $title ) {
 428                 if ( isset( $this->mPageCallback ) ) {
 429                         call_user_func( $this->mPageCallback, $title );
 430                 }
 431         }
 432
 433         /**
 434          * Notify the callback function when a "</page>" is closed.
 435          * @param Title $title
 436          * @param ForeignTitle $foreignTitle
 437          * @param int $revCount
 438          * @param int $sucCount Number of revisions for which callback returned true
 439          * @param array $pageInfo Associative array of page information
 440          */
 441         private function pageOutCallback( $title, $foreignTitle, $revCount,
 442                         $sucCount, $pageInfo ) {
 443                 if ( isset( $this->mPageOutCallback ) ) {
 444                         $args = func_get_args();
 445                         call_user_func_array( $this->mPageOutCallback, $args );
 446                 }
 447         }
 448
 449         /**
 450          * Notify the callback function of a revision
 451          * @param WikiRevision $revision
 452          * @return bool|mixed
 453          */
 454         private function revisionCallback( $revision ) {
 455                 if ( isset( $this->mRevisionCallback ) ) {
 456                         return call_user_func_array( $this->mRevisionCallback,
 457                                         array( $revision, $this ) );
 458                 } else {
 459                         return false;
 460                 }
 461         }
 462
 463         /**
 464          * Notify the callback function of a new log item
 465          * @param WikiRevision $revision
 466          * @return bool|mixed
 467          */
 468         private function logItemCallback( $revision ) {
 469                 if ( isset( $this->mLogItemCallback ) ) {
 470                         return call_user_func_array( $this->mLogItemCallback,
 471                                         array( $revision, $this ) );
 472                 } else {
 473                         return false;
 474                 }
 475         }
 476
 477         /**
 478          * Retrieves the contents of the named attribute of the current element.
 479          * @param string $attr The name of the attribute
 480          * @return string The value of the attribute or an empty string if it is not set in the current
 481          * element.
 482          */
 483         public function nodeAttribute( $attr ) {
 484                 return $this->reader->getAttribute( $attr );
 485         }
 486
 487         /**
 488          * Shouldn't something like this be built-in to XMLReader?
 489          * Fetches text contents of the current element, assuming
 490          * no sub-elements or such scary things.
 491          * @return string
 492          * @access private
 493          */
 494         public function nodeContents() {
 495                 if ( $this->reader->isEmptyElement ) {
 496                         return "";
 497                 }
 498                 $buffer = "";
 499                 while ( $this->reader->read() ) {
 500                         switch ( $this->reader->nodeType ) {
 501                         case XMLReader::TEXT:
 502                         case XMLReader::SIGNIFICANT_WHITESPACE:
 503                                 $buffer .= $this->reader->value;
 504                                 break;
 505                         case XMLReader::END_ELEMENT:
 506                                 return $buffer;
 507                         }
 508                 }
 509
 510                 $this->reader->close();
 511                 return '';
 512         }
 513
 514         /**
 515          * Primary entry point
 516          * @throws MWException
 517          * @return bool
 518          */
 519         public function doImport() {
 520                 // Calls to reader->read need to be wrapped in calls to
 521                 // libxml_disable_entity_loader() to avoid local file
 522                 // inclusion attacks (bug 46932).
 523                 $oldDisable = libxml_disable_entity_loader( true );
 524                 $this->reader->read();
 525
 526                 if ( $this->reader->name != 'mediawiki' ) {
 527                         libxml_disable_entity_loader( $oldDisable );
 528                         throw new MWException( "Expected <mediawiki> tag, got " .
 529                                 $this->reader->name );
 530                 }
 531                 $this->debug( "<mediawiki> tag is correct." );
 532
 533                 $this->debug( "Starting primary dump processing loop." );
 534
 535                 $keepReading = $this->reader->read();
 536                 $skip = false;
 537                 $rethrow = null;
 538                 try {
 539                         while ( $keepReading ) {
 540                                 $tag = $this->reader->name;
 541                                 $type = $this->reader->nodeType;
 542
 543                                 if ( !Hooks::run( 'ImportHandleToplevelXMLTag', array( $this ) ) ) {
 544                                         // Do nothing
 545                                 } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) {
 546                                         break;
 547                                 } elseif ( $tag == 'siteinfo' ) {
 548                                         $this->handleSiteInfo();
 549                                 } elseif ( $tag == 'page' ) {
 550                                         $this->handlePage();
 551                                 } elseif ( $tag == 'logitem' ) {
 552                                         $this->handleLogItem();
 553                                 } elseif ( $tag != '#text' ) {
 554                                         $this->warn( "Unhandled top-level XML tag $tag" );
 555
 556                                         $skip = true;
 557                                 }
 558
 559                                 if ( $skip ) {
 560                                         $keepReading = $this->reader->next();
 561                                         $skip = false;
 562                                         $this->debug( "Skip" );
 563                                 } else {
 564                                         $keepReading = $this->reader->read();
 565                                 }
 566                         }
 567                 } catch ( Exception $ex ) {
 568                         $rethrow = $ex;
 569                 }
 570
 571                 // finally
 572                 libxml_disable_entity_loader( $oldDisable );
 573                 $this->reader->close();
 574
 575                 if ( $rethrow ) {
 576                         throw $rethrow;
 577                 }
 578
 579                 return true;
 580         }
 581
 582         private function handleSiteInfo() {
 583                 $this->debug( "Enter site info handler." );
 584                 $siteInfo = array();
 585
 586                 // Fields that can just be stuffed in the siteInfo object
 587                 $normalFields = array( 'sitename', 'base', 'generator', 'case' );
 588
 589                 while ( $this->reader->read() ) {
 590                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 591                                         $this->reader->name == 'siteinfo' ) {
 592                                 break;
 593                         }
 594
 595                         $tag = $this->reader->name;
 596
 597                         if ( $tag == 'namespace' ) {
 598                                 $this->foreignNamespaces[ $this->nodeAttribute( 'key' ) ] =
 599                                         $this->nodeContents();
 600                         } elseif ( in_array( $tag, $normalFields ) ) {
 601                                 $siteInfo[$tag] = $this->nodeContents();
 602                         }
 603                 }
 604
 605                 $siteInfo['_namespaces'] = $this->foreignNamespaces;
 606                 $this->siteInfoCallback( $siteInfo );
 607         }
 608
 609         private function handleLogItem() {
 610                 $this->debug( "Enter log item handler." );
 611                 $logInfo = array();
 612
 613                 // Fields that can just be stuffed in the pageInfo object
 614                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 615                                         'logtitle', 'params' );
 616
 617                 while ( $this->reader->read() ) {
 618                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 619                                         $this->reader->name == 'logitem' ) {
 620                                 break;
 621                         }
 622
 623                         $tag = $this->reader->name;
 624
 625                         if ( !Hooks::run( 'ImportHandleLogItemXMLTag', array(
 626                                 $this, $logInfo
 627                         ) ) ) {
 628                                 // Do nothing
 629                         } elseif ( in_array( $tag, $normalFields ) ) {
 630                                 $logInfo[$tag] = $this->nodeContents();
 631                         } elseif ( $tag == 'contributor' ) {
 632                                 $logInfo['contributor'] = $this->handleContributor();
 633                         } elseif ( $tag != '#text' ) {
 634                                 $this->warn( "Unhandled log-item XML tag $tag" );
 635                         }
 636                 }
 637
 638                 $this->processLogItem( $logInfo );
 639         }
 640
 641         /**
 642          * @param array $logInfo
 643          * @return bool|mixed
 644          */
 645         private function processLogItem( $logInfo ) {
 646                 $revision = new WikiRevision( $this->config );
 647
 648                 $revision->setID( $logInfo['id'] );
 649                 $revision->setType( $logInfo['type'] );
 650                 $revision->setAction( $logInfo['action'] );
 651                 $revision->setTimestamp( $logInfo['timestamp'] );
 652                 $revision->setParams( $logInfo['params'] );
 653                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 654                 $revision->setNoUpdates( $this->mNoUpdates );
 655
 656                 if ( isset( $logInfo['comment'] ) ) {
 657                         $revision->setComment( $logInfo['comment'] );
 658                 }
 659
 660                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 661                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 662                 }
 663                 if ( isset( $logInfo['contributor']['username'] ) ) {
 664                         $revision->setUserName( $logInfo['contributor']['username'] );
 665                 }
 666
 667                 return $this->logItemCallback( $revision );
 668         }
 669
 670         private function handlePage() {
 671                 // Handle page data.
 672                 $this->debug( "Enter page handler." );
 673                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 674
 675                 // Fields that can just be stuffed in the pageInfo object
 676                 $normalFields = array( 'title', 'ns', 'id', 'redirect', 'restrictions' );
 677
 678                 $skip = false;
 679                 $badTitle = false;
 680
 681                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 682                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 683                                         $this->reader->name == 'page' ) {
 684                                 break;
 685                         }
 686
 687                         $skip = false;
 688
 689                         $tag = $this->reader->name;
 690
 691                         if ( $badTitle ) {
 692                                 // The title is invalid, bail out of this page
 693                                 $skip = true;
 694                         } elseif ( !Hooks::run( 'ImportHandlePageXMLTag', array( $this,
 695                                                 &$pageInfo ) ) ) {
 696                                 // Do nothing
 697                         } elseif ( in_array( $tag, $normalFields ) ) {
 698                                 // An XML snippet:
 699                                 // <page>
 700                                 //     <id>123</id>
 701                                 //     <title>Page</title>
 702                                 //     <redirect title="NewTitle"/>
 703                                 //     ...
 704                                 // Because the redirect tag is built differently, we need special handling for that case.
 705                                 if ( $tag == 'redirect' ) {
 706                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 707                                 } else {
 708                                         $pageInfo[$tag] = $this->nodeContents();
 709                                 }
 710                         } elseif ( $tag == 'revision' || $tag == 'upload' ) {
 711                                 if ( !isset( $title ) ) {
 712                                         $title = $this->processTitle( $pageInfo['title'],
 713                                                 isset( $pageInfo['ns'] ) ? $pageInfo['ns'] : null );
 714
 715                                         if ( !$title ) {
 716                                                 $badTitle = true;
 717                                                 $skip = true;
 718                                         }
 719
 720                                         $this->pageCallback( $title );
 721                                         list( $pageInfo['_title'], $foreignTitle ) = $title;
 722                                 }
 723
 724                                 if ( $title ) {
 725                                         if ( $tag == 'revision' ) {
 726                                                 $this->handleRevision( $pageInfo );
 727                                         } else {
 728                                                 $this->handleUpload( $pageInfo );
 729                                         }
 730                                 }
 731                         } elseif ( $tag != '#text' ) {
 732                                 $this->warn( "Unhandled page XML tag $tag" );
 733                                 $skip = true;
 734                         }
 735                 }
 736
 737                 $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
 738                                         $pageInfo['revisionCount'],
 739                                         $pageInfo['successfulRevisionCount'],
 740                                         $pageInfo );
 741         }
 742
 743         /**
 744          * @param array $pageInfo
 745          */
 746         private function handleRevision( &$pageInfo ) {
 747                 $this->debug( "Enter revision handler" );
 748                 $revisionInfo = array();
 749
 750                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' );
 751
 752                 $skip = false;
 753
 754                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 755                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 756                                         $this->reader->name == 'revision' ) {
 757                                 break;
 758                         }
 759
 760                         $tag = $this->reader->name;
 761
 762                         if ( !Hooks::run( 'ImportHandleRevisionXMLTag', array(
 763                                 $this, $pageInfo, $revisionInfo
 764                         ) ) ) {
 765                                 // Do nothing
 766                         } elseif ( in_array( $tag, $normalFields ) ) {
 767                                 $revisionInfo[$tag] = $this->nodeContents();
 768                         } elseif ( $tag == 'contributor' ) {
 769                                 $revisionInfo['contributor'] = $this->handleContributor();
 770                         } elseif ( $tag != '#text' ) {
 771                                 $this->warn( "Unhandled revision XML tag $tag" );
 772                                 $skip = true;
 773                         }
 774                 }
 775
 776                 $pageInfo['revisionCount']++;
 777                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 778                         $pageInfo['successfulRevisionCount']++;
 779                 }
 780         }
 781
 782         /**
 783          * @param array $pageInfo
 784          * @param array $revisionInfo
 785          * @return bool|mixed
 786          */
 787         private function processRevision( $pageInfo, $revisionInfo ) {
 788                 $revision = new WikiRevision( $this->config );
 789
 790                 if ( isset( $revisionInfo['id'] ) ) {
 791                         $revision->setID( $revisionInfo['id'] );
 792                 }
 793                 if ( isset( $revisionInfo['model'] ) ) {
 794                         $revision->setModel( $revisionInfo['model'] );
 795                 }
 796                 if ( isset( $revisionInfo['format'] ) ) {
 797                         $revision->setFormat( $revisionInfo['format'] );
 798                 }
 799                 $revision->setTitle( $pageInfo['_title'] );
 800
 801                 if ( isset( $revisionInfo['text'] ) ) {
 802                         $handler = $revision->getContentHandler();
 803                         $text = $handler->importTransform(
 804                                 $revisionInfo['text'],
 805                                 $revision->getFormat() );
 806
 807                         $revision->setText( $text );
 808                 }
 809                 if ( isset( $revisionInfo['timestamp'] ) ) {
 810                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 811                 } else {
 812                         $revision->setTimestamp( wfTimestampNow() );
 813                 }
 814
 815                 if ( isset( $revisionInfo['comment'] ) ) {
 816                         $revision->setComment( $revisionInfo['comment'] );
 817                 }
 818
 819                 if ( isset( $revisionInfo['minor'] ) ) {
 820                         $revision->setMinor( true );
 821                 }
 822                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 823                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 824                 }
 825                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 826                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 827                 }
 828                 $revision->setNoUpdates( $this->mNoUpdates );
 829
 830                 return $this->revisionCallback( $revision );
 831         }
 832
 833         /**
 834          * @param array $pageInfo
 835          * @return mixed
 836          */
 837         private function handleUpload( &$pageInfo ) {
 838                 $this->debug( "Enter upload handler" );
 839                 $uploadInfo = array();
 840
 841                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 842                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 843
 844                 $skip = false;
 845
 846                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 847                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 848                                         $this->reader->name == 'upload' ) {
 849                                 break;
 850                         }
 851
 852                         $tag = $this->reader->name;
 853
 854                         if ( !Hooks::run( 'ImportHandleUploadXMLTag', array(
 855                                 $this, $pageInfo
 856                         ) ) ) {
 857                                 // Do nothing
 858                         } elseif ( in_array( $tag, $normalFields ) ) {
 859                                 $uploadInfo[$tag] = $this->nodeContents();
 860                         } elseif ( $tag == 'contributor' ) {
 861                                 $uploadInfo['contributor'] = $this->handleContributor();
 862                         } elseif ( $tag == 'contents' ) {
 863                                 $contents = $this->nodeContents();
 864                                 $encoding = $this->reader->getAttribute( 'encoding' );
 865                                 if ( $encoding === 'base64' ) {
 866                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 867                                         $uploadInfo['isTempSrc'] = true;
 868                                 }
 869                         } elseif ( $tag != '#text' ) {
 870                                 $this->warn( "Unhandled upload XML tag $tag" );
 871                                 $skip = true;
 872                         }
 873                 }
 874
 875                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 876                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 877                         if ( file_exists( $path ) ) {
 878                                 $uploadInfo['fileSrc'] = $path;
 879                                 $uploadInfo['isTempSrc'] = false;
 880                         }
 881                 }
 882
 883                 if ( $this->mImportUploads ) {
 884                         return $this->processUpload( $pageInfo, $uploadInfo );
 885                 }
 886         }
 887
 888         /**
 889          * @param string $contents
 890          * @return string
 891          */
 892         private function dumpTemp( $contents ) {
 893                 $filename = tempnam( wfTempDir(), 'importupload' );
 894                 file_put_contents( $filename, $contents );
 895                 return $filename;
 896         }
 897
 898         /**
 899          * @param array $pageInfo
 900          * @param array $uploadInfo
 901          * @return mixed
 902          */
 903         private function processUpload( $pageInfo, $uploadInfo ) {
 904                 $revision = new WikiRevision( $this->config );
 905                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 906
 907                 $revision->setTitle( $pageInfo['_title'] );
 908                 $revision->setID( $pageInfo['id'] );
 909                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 910                 $revision->setText( $text );
 911                 $revision->setFilename( $uploadInfo['filename'] );
 912                 if ( isset( $uploadInfo['archivename'] ) ) {
 913                         $revision->setArchiveName( $uploadInfo['archivename'] );
 914                 }
 915                 $revision->setSrc( $uploadInfo['src'] );
 916                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 917                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 918                                 !empty( $uploadInfo['isTempSrc'] ) );
 919                 }
 920                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 921                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 922                 }
 923                 $revision->setSize( intval( $uploadInfo['size'] ) );
 924                 $revision->setComment( $uploadInfo['comment'] );
 925
 926                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 927                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 928                 }
 929                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 930                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 931                 }
 932                 $revision->setNoUpdates( $this->mNoUpdates );
 933
 934                 return call_user_func( $this->mUploadCallback, $revision );
 935         }
 936
 937         /**
 938          * @return array
 939          */
 940         private function handleContributor() {
 941                 $fields = array( 'id', 'ip', 'username' );
 942                 $info = array();
 943
 944                 while ( $this->reader->read() ) {
 945                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 946                                         $this->reader->name == 'contributor' ) {
 947                                 break;
 948                         }
 949
 950                         $tag = $this->reader->name;
 951
 952                         if ( in_array( $tag, $fields ) ) {
 953                                 $info[$tag] = $this->nodeContents();
 954                         }
 955                 }
 956
 957                 return $info;
 958         }
 959
 960         /**
 961          * @param string $text
 962          * @param string|null $ns
 963          * @return array|bool
 964          */
 965         private function processTitle( $text, $ns = null ) {
 966                 if ( is_null( $this->foreignNamespaces ) ) {
 967                         $foreignTitleFactory = new NaiveForeignTitleFactory();
 968                 } else {
 969                         $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
 970                                 $this->foreignNamespaces );
 971                 }
 972
 973                 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
 974                         intval( $ns ) );
 975
 976                 $title = $this->importTitleFactory->createTitleFromForeignTitle(
 977                         $foreignTitle );
 978
 979                 $commandLineMode = $this->config->get( 'CommandLineMode' );
 980                 if ( is_null( $title ) ) {
 981                         # Invalid page title? Ignore the page
 982                         $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
 983                         return false;
 984                 } elseif ( $title->isExternal() ) {
 985                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
 986                         return false;
 987                 } elseif ( !$title->canExist() ) {
 988                         $this->notice( 'import-error-special', $title->getPrefixedText() );
 989                         return false;
 990                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
 991                         # Do not import if the importing wiki user cannot edit this page
 992                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
 993                         return false;
 994                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
 995                         # Do not import if the importing wiki user cannot create this page
 996                         $this->notice( 'import-error-create', $title->getPrefixedText() );
 997                         return false;
 998                 }
 999
1000                 return array( $title, $foreignTitle );
1001         }
1002 }
1003
1004 /** This is a horrible hack used to keep source compatibility */
1005 class UploadSourceAdapter {
1006         /** @var array */
1007         public static $sourceRegistrations = array();
1008
1009         /** @var string */
1010         private $mSource;
1011
1012         /** @var string */
1013         private $mBuffer;
1014
1015         /** @var int */
1016         private $mPosition;
1017
1018         /**
1019          * @param ImportSource $source
1020          * @return string
1021          */
1022         static function registerSource( ImportSource $source ) {
1023                 $id = wfRandomString();
1024
1025                 self::$sourceRegistrations[$id] = $source;
1026
1027                 return $id;
1028         }
1029
1030         /**
1031          * @param string $path
1032          * @param string $mode
1033          * @param array $options
1034          * @param string $opened_path
1035          * @return bool
1036          */
1037         function stream_open( $path, $mode, $options, &$opened_path ) {
1038                 $url = parse_url( $path );
1039                 $id = $url['host'];
1040
1041                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
1042                         return false;
1043                 }
1044
1045                 $this->mSource = self::$sourceRegistrations[$id];
1046
1047                 return true;
1048         }
1049
1050         /**
1051          * @param int $count
1052          * @return string
1053          */
1054         function stream_read( $count ) {
1055                 $return = '';
1056                 $leave = false;
1057
1058                 while ( !$leave && !$this->mSource->atEnd() &&
1059                                 strlen( $this->mBuffer ) < $count ) {
1060                         $read = $this->mSource->readChunk();
1061
1062                         if ( !strlen( $read ) ) {
1063                                 $leave = true;
1064                         }
1065
1066                         $this->mBuffer .= $read;
1067                 }
1068
1069                 if ( strlen( $this->mBuffer ) ) {
1070                         $return = substr( $this->mBuffer, 0, $count );
1071                         $this->mBuffer = substr( $this->mBuffer, $count );
1072                 }
1073
1074                 $this->mPosition += strlen( $return );
1075
1076                 return $return;
1077         }
1078
1079         /**
1080          * @param string $data
1081          * @return bool
1082          */
1083         function stream_write( $data ) {
1084                 return false;
1085         }
1086
1087         /**
1088          * @return mixed
1089          */
1090         function stream_tell() {
1091                 return $this->mPosition;
1092         }
1093
1094         /**
1095          * @return bool
1096          */
1097         function stream_eof() {
1098                 return $this->mSource->atEnd();
1099         }
1100
1101         /**
1102          * @return array
1103          */
1104         function url_stat() {
1105                 $result = array();
1106
1107                 $result['dev'] = $result[0] = 0;
1108                 $result['ino'] = $result[1] = 0;
1109                 $result['mode'] = $result[2] = 0;
1110                 $result['nlink'] = $result[3] = 0;
1111                 $result['uid'] = $result[4] = 0;
1112                 $result['gid'] = $result[5] = 0;
1113                 $result['rdev'] = $result[6] = 0;
1114                 $result['size'] = $result[7] = 0;
1115                 $result['atime'] = $result[8] = 0;
1116                 $result['mtime'] = $result[9] = 0;
1117                 $result['ctime'] = $result[10] = 0;
1118                 $result['blksize'] = $result[11] = 0;
1119                 $result['blocks'] = $result[12] = 0;
1120
1121                 return $result;
1122         }
1123 }
1124
1125 /**
1126  * @todo document (e.g. one-sentence class description).
1127  * @ingroup SpecialPage
1128  */
1129 class WikiRevision {
1130         /** @todo Unused? */
1131         public $importer = null;
1132
1133         /** @var Title */
1134         public $title = null;
1135
1136         /** @var int */
1137         public $id = 0;
1138
1139         /** @var string */
1140         public $timestamp = "20010115000000";
1141
1142         /**
1143          * @var int
1144          * @todo Can't find any uses. Public, because that's suspicious. Get clarity. */
1145         public $user = 0;
1146
1147         /** @var string */
1148         public $user_text = "";
1149
1150         /** @var string */
1151         public $model = null;
1152
1153         /** @var string */
1154         public $format = null;
1155
1156         /** @var string */
1157         public $text = "";
1158
1159         /** @var int */
1160         protected $size;
1161
1162         /** @var Content */
1163         public $content = null;
1164
1165         /** @var ContentHandler */
1166         protected $contentHandler = null;
1167
1168         /** @var string */
1169         public $comment = "";
1170
1171         /** @var bool */
1172         public $minor = false;
1173
1174         /** @var string */
1175         public $type = "";
1176
1177         /** @var string */
1178         public $action = "";
1179
1180         /** @var string */
1181         public $params = "";
1182
1183         /** @var string */
1184         public $fileSrc = '';
1185
1186         /** @var bool|string */
1187         public $sha1base36 = false;
1188
1189         /**
1190          * @var bool
1191          * @todo Unused?
1192          */
1193         public $isTemp = false;
1194
1195         /** @var string */
1196         public $archiveName = '';
1197
1198         protected $filename;
1199
1200         /** @var mixed */
1201         protected $src;
1202
1203         /** @todo Unused? */
1204         public $fileIsTemp;
1205
1206         /** @var bool */
1207         private $mNoUpdates = false;
1208
1209         /** @var Config $config */
1210         private $config;
1211
1212         public function __construct( Config $config ) {
1213                 $this->config = $config;
1214         }
1215
1216         /**
1217          * @param Title $title
1218          * @throws MWException
1219          */
1220         function setTitle( $title ) {
1221                 if ( is_object( $title ) ) {
1222                         $this->title = $title;
1223                 } elseif ( is_null( $title ) ) {
1224                         throw new MWException( "WikiRevision given a null title in import. "
1225                                 . "You may need to adjust \$wgLegalTitleChars." );
1226                 } else {
1227                         throw new MWException( "WikiRevision given non-object title in import." );
1228                 }
1229         }
1230
1231         /**
1232          * @param int $id
1233          */
1234         function setID( $id ) {
1235                 $this->id = $id;
1236         }
1237
1238         /**
1239          * @param string $ts
1240          */
1241         function setTimestamp( $ts ) {
1242                 # 2003-08-05T18:30:02Z
1243                 $this->timestamp = wfTimestamp( TS_MW, $ts );
1244         }
1245
1246         /**
1247          * @param string $user
1248          */
1249         function setUsername( $user ) {
1250                 $this->user_text = $user;
1251         }
1252
1253         /**
1254          * @param string $ip
1255          */
1256         function setUserIP( $ip ) {
1257                 $this->user_text = $ip;
1258         }
1259
1260         /**
1261          * @param string $model
1262          */
1263         function setModel( $model ) {
1264                 $this->model = $model;
1265         }
1266
1267         /**
1268          * @param string $format
1269          */
1270         function setFormat( $format ) {
1271                 $this->format = $format;
1272         }
1273
1274         /**
1275          * @param string $text
1276          */
1277         function setText( $text ) {
1278                 $this->text = $text;
1279         }
1280
1281         /**
1282          * @param string $text
1283          */
1284         function setComment( $text ) {
1285                 $this->comment = $text;
1286         }
1287
1288         /**
1289          * @param bool $minor
1290          */
1291         function setMinor( $minor ) {
1292                 $this->minor = (bool)$minor;
1293         }
1294
1295         /**
1296          * @param mixed $src
1297          */
1298         function setSrc( $src ) {
1299                 $this->src = $src;
1300         }
1301
1302         /**
1303          * @param string $src
1304          * @param bool $isTemp
1305          */
1306         function setFileSrc( $src, $isTemp ) {
1307                 $this->fileSrc = $src;
1308                 $this->fileIsTemp = $isTemp;
1309         }
1310
1311         /**
1312          * @param string $sha1base36
1313          */
1314         function setSha1Base36( $sha1base36 ) {
1315                 $this->sha1base36 = $sha1base36;
1316         }
1317
1318         /**
1319          * @param string $filename
1320          */
1321         function setFilename( $filename ) {
1322                 $this->filename = $filename;
1323         }
1324
1325         /**
1326          * @param string $archiveName
1327          */
1328         function setArchiveName( $archiveName ) {
1329                 $this->archiveName = $archiveName;
1330         }
1331
1332         /**
1333          * @param int $size
1334          */
1335         function setSize( $size ) {
1336                 $this->size = intval( $size );
1337         }
1338
1339         /**
1340          * @param string $type
1341          */
1342         function setType( $type ) {
1343                 $this->type = $type;
1344         }
1345
1346         /**
1347          * @param string $action
1348          */
1349         function setAction( $action ) {
1350                 $this->action = $action;
1351         }
1352
1353         /**
1354          * @param array $params
1355          */
1356         function setParams( $params ) {
1357                 $this->params = $params;
1358         }
1359
1360         /**
1361          * @param bool $noupdates
1362          */
1363         public function setNoUpdates( $noupdates ) {
1364                 $this->mNoUpdates = $noupdates;
1365         }
1366
1367         /**
1368          * @return Title
1369          */
1370         function getTitle() {
1371                 return $this->title;
1372         }
1373
1374         /**
1375          * @return int
1376          */
1377         function getID() {
1378                 return $this->id;
1379         }
1380
1381         /**
1382          * @return string
1383          */
1384         function getTimestamp() {
1385                 return $this->timestamp;
1386         }
1387
1388         /**
1389          * @return string
1390          */
1391         function getUser() {
1392                 return $this->user_text;
1393         }
1394
1395         /**
1396          * @return string
1397          *
1398          * @deprecated Since 1.21, use getContent() instead.
1399          */
1400         function getText() {
1401                 ContentHandler::deprecated( __METHOD__, '1.21' );
1402
1403                 return $this->text;
1404         }
1405
1406         /**
1407          * @return ContentHandler
1408          */
1409         function getContentHandler() {
1410                 if ( is_null( $this->contentHandler ) ) {
1411                         $this->contentHandler = ContentHandler::getForModelID( $this->getModel() );
1412                 }
1413
1414                 return $this->contentHandler;
1415         }
1416
1417         /**
1418          * @return Content
1419          */
1420         function getContent() {
1421                 if ( is_null( $this->content ) ) {
1422                         $handler = $this->getContentHandler();
1423                         $this->content = $handler->unserializeContent( $this->text, $this->getFormat() );
1424                 }
1425
1426                 return $this->content;
1427         }
1428
1429         /**
1430          * @return string
1431          */
1432         function getModel() {
1433                 if ( is_null( $this->model ) ) {
1434                         $this->model = $this->getTitle()->getContentModel();
1435                 }
1436
1437                 return $this->model;
1438         }
1439
1440         /**
1441          * @return string
1442          */
1443         function getFormat() {
1444                 if ( is_null( $this->format ) ) {
1445                         $this->format = $this->getContentHandler()->getDefaultFormat();
1446                 }
1447
1448                 return $this->format;
1449         }
1450
1451         /**
1452          * @return string
1453          */
1454         function getComment() {
1455                 return $this->comment;
1456         }
1457
1458         /**
1459          * @return bool
1460          */
1461         function getMinor() {
1462                 return $this->minor;
1463         }
1464
1465         /**
1466          * @return mixed
1467          */
1468         function getSrc() {
1469                 return $this->src;
1470         }
1471
1472         /**
1473          * @return bool|string
1474          */
1475         function getSha1() {
1476                 if ( $this->sha1base36 ) {
1477                         return wfBaseConvert( $this->sha1base36, 36, 16 );
1478                 }
1479                 return false;
1480         }
1481
1482         /**
1483          * @return string
1484          */
1485         function getFileSrc() {
1486                 return $this->fileSrc;
1487         }
1488
1489         /**
1490          * @return bool
1491          */
1492         function isTempSrc() {
1493                 return $this->isTemp;
1494         }
1495
1496         /**
1497          * @return mixed
1498          */
1499         function getFilename() {
1500                 return $this->filename;
1501         }
1502
1503         /**
1504          * @return string
1505          */
1506         function getArchiveName() {
1507                 return $this->archiveName;
1508         }
1509
1510         /**
1511          * @return mixed
1512          */
1513         function getSize() {
1514                 return $this->size;
1515         }
1516
1517         /**
1518          * @return string
1519          */
1520         function getType() {
1521                 return $this->type;
1522         }
1523
1524         /**
1525          * @return string
1526          */
1527         function getAction() {
1528                 return $this->action;
1529         }
1530
1531         /**
1532          * @return string
1533          */
1534         function getParams() {
1535                 return $this->params;
1536         }
1537
1538         /**
1539          * @return bool
1540          */
1541         function importOldRevision() {
1542                 $dbw = wfGetDB( DB_MASTER );
1543
1544                 # Sneak a single revision into place
1545                 $user = User::newFromName( $this->getUser() );
1546                 if ( $user ) {
1547                         $userId = intval( $user->getId() );
1548                         $userText = $user->getName();
1549                         $userObj = $user;
1550                 } else {
1551                         $userId = 0;
1552                         $userText = $this->getUser();
1553                         $userObj = new User;
1554                 }
1555
1556                 // avoid memory leak...?
1557                 $linkCache = LinkCache::singleton();
1558                 $linkCache->clear();
1559
1560                 $page = WikiPage::factory( $this->title );
1561                 $page->loadPageData( 'fromdbmaster' );
1562                 if ( !$page->exists() ) {
1563                         # must create the page...
1564                         $pageId = $page->insertOn( $dbw );
1565                         $created = true;
1566                         $oldcountable = null;
1567                 } else {
1568                         $pageId = $page->getId();
1569                         $created = false;
1570
1571                         $prior = $dbw->selectField( 'revision', '1',
1572                                 array( 'rev_page' => $pageId,
1573                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1574                                         'rev_user_text' => $userText,
1575                                         'rev_comment' => $this->getComment() ),
1576                                 __METHOD__
1577                         );
1578                         if ( $prior ) {
1579                                 // @todo FIXME: This could fail slightly for multiple matches :P
1580                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1581                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1582                                 return false;
1583                         }
1584                 }
1585
1586                 # @todo FIXME: Use original rev_id optionally (better for backups)
1587                 # Insert the row
1588                 $revision = new Revision( array(
1589                         'title' => $this->title,
1590                         'page' => $pageId,
1591                         'content_model' => $this->getModel(),
1592                         'content_format' => $this->getFormat(),
1593                         //XXX: just set 'content' => $this->getContent()?
1594                         'text' => $this->getContent()->serialize( $this->getFormat() ),
1595                         'comment' => $this->getComment(),
1596                         'user' => $userId,
1597                         'user_text' => $userText,
1598                         'timestamp' => $this->timestamp,
1599                         'minor_edit' => $this->minor,
1600                         ) );
1601                 $revision->insertOn( $dbw );
1602                 $changed = $page->updateIfNewerOn( $dbw, $revision );
1603
1604                 if ( $changed !== false && !$this->mNoUpdates ) {
1605                         wfDebug( __METHOD__ . ": running updates\n" );
1606                         // countable/oldcountable stuff is handled in WikiImporter::finishImportPage
1607                         $page->doEditUpdates(
1608                                 $revision,
1609                                 $userObj,
1610                                 array( 'created' => $created, 'oldcountable' => 'no-change' )
1611                         );
1612                 }
1613
1614                 return true;
1615         }
1616
1617         function importLogItem() {
1618                 $dbw = wfGetDB( DB_MASTER );
1619                 # @todo FIXME: This will not record autoblocks
1620                 if ( !$this->getTitle() ) {
1621                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1622                                 $this->timestamp . "\n" );
1623                         return;
1624                 }
1625                 # Check if it exists already
1626                 // @todo FIXME: Use original log ID (better for backups)
1627                 $prior = $dbw->selectField( 'logging', '1',
1628                         array( 'log_type' => $this->getType(),
1629                                 'log_action' => $this->getAction(),
1630                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1631                                 'log_namespace' => $this->getTitle()->getNamespace(),
1632                                 'log_title' => $this->getTitle()->getDBkey(),
1633                                 'log_comment' => $this->getComment(),
1634                                 #'log_user_text' => $this->user_text,
1635                                 'log_params' => $this->params ),
1636                         __METHOD__
1637                 );
1638                 // @todo FIXME: This could fail slightly for multiple matches :P
1639                 if ( $prior ) {
1640                         wfDebug( __METHOD__
1641                                 . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp "
1642                                 . $this->timestamp . "\n" );
1643                         return;
1644                 }
1645                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1646                 $data = array(
1647                         'log_id' => $log_id,
1648                         'log_type' => $this->type,
1649                         'log_action' => $this->action,
1650                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1651                         'log_user' => User::idFromName( $this->user_text ),
1652                         #'log_user_text' => $this->user_text,
1653                         'log_namespace' => $this->getTitle()->getNamespace(),
1654                         'log_title' => $this->getTitle()->getDBkey(),
1655                         'log_comment' => $this->getComment(),
1656                         'log_params' => $this->params
1657                 );
1658                 $dbw->insert( 'logging', $data, __METHOD__ );
1659         }
1660
1661         /**
1662          * @return bool
1663          */
1664         function importUpload() {
1665                 # Construct a file
1666                 $archiveName = $this->getArchiveName();
1667                 if ( $archiveName ) {
1668                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1669                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1670                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1671                 } else {
1672                         $file = wfLocalFile( $this->getTitle() );
1673                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1674                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1675                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1676                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1677                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1678                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1679                         }
1680                 }
1681                 if ( !$file ) {
1682                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1683                         return false;
1684                 }
1685
1686                 # Get the file source or download if necessary
1687                 $source = $this->getFileSrc();
1688                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1689                 if ( !$source ) {
1690                         $source = $this->downloadSource();
1691                         $flags |= File::DELETE_SOURCE;
1692                 }
1693                 if ( !$source ) {
1694                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1695                         return false;
1696                 }
1697                 $sha1 = $this->getSha1();
1698                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1699                         if ( $flags & File::DELETE_SOURCE ) {
1700                                 # Broken file; delete it if it is a temporary file
1701                                 unlink( $source );
1702                         }
1703                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1704                         return false;
1705                 }
1706
1707                 $user = User::newFromName( $this->user_text );
1708
1709                 # Do the actual upload
1710                 if ( $archiveName ) {
1711                         $status = $file->uploadOld( $source, $archiveName,
1712                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1713                 } else {
1714                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1715                                 $flags, false, $this->getTimestamp(), $user );
1716                 }
1717
1718                 if ( $status->isGood() ) {
1719                         wfDebug( __METHOD__ . ": Successful\n" );
1720                         return true;
1721                 } else {
1722                         wfDebug( __METHOD__ . ': failed: ' . $status->getHTML() . "\n" );
1723                         return false;
1724                 }
1725         }
1726
1727         /**
1728          * @return bool|string
1729          */
1730         function downloadSource() {
1731                 if ( !$this->config->get( 'EnableUploads' ) ) {
1732                         return false;
1733                 }
1734
1735                 $tempo = tempnam( wfTempDir(), 'download' );
1736                 $f = fopen( $tempo, 'wb' );
1737                 if ( !$f ) {
1738                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1739                         return false;
1740                 }
1741
1742                 // @todo FIXME!
1743                 $src = $this->getSrc();
1744                 $data = Http::get( $src, array(), __METHOD__ );
1745                 if ( !$data ) {
1746                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1747                         fclose( $f );
1748                         unlink( $tempo );
1749                         return false;
1750                 }
1751
1752                 fwrite( $f, $data );
1753                 fclose( $f );
1754
1755                 return $tempo;
1756         }
1757
1758 }
1759
1760 /**
1761  * Source interface for XML import.
1762  */
1763 interface ImportSource {
1764
1765         /**
1766          * Indicates whether the end of the input has been reached.
1767          * Will return true after a finite number of calls to readChunk.
1768          *
1769          * @return bool true if there is no more input, false otherwise.
1770          */
1771         function atEnd();
1772
1773         /**
1774          * Return a chunk of the input, as a (possibly empty) string.
1775          * When the end of input is reached, readChunk() returns false.
1776          * If atEnd() returns false, readChunk() will return a string.
1777          * If atEnd() returns true, readChunk() will return false.
1778          *
1779          * @return bool|string
1780          */
1781         function readChunk();
1782 }
1783
1784 /**
1785  * Used for importing XML dumps where the content of the dump is in a string.
1786  * This class is ineffecient, and should only be used for small dumps.
1787  * For larger dumps, ImportStreamSource should be used instead.
1788  *
1789  * @ingroup SpecialPage
1790  */
1791 class ImportStringSource implements ImportSource {
1792         function __construct( $string ) {
1793                 $this->mString = $string;
1794                 $this->mRead = false;
1795         }
1796
1797         /**
1798          * @return bool
1799          */
1800         function atEnd() {
1801                 return $this->mRead;
1802         }
1803
1804         /**
1805          * @return bool|string
1806          */
1807         function readChunk() {
1808                 if ( $this->atEnd() ) {
1809                         return false;
1810                 }
1811                 $this->mRead = true;
1812                 return $this->mString;
1813         }
1814 }
1815
1816 /**
1817  * Imports a XML dump from a file (either from file upload, files on disk, or HTTP)
1818  * @ingroup SpecialPage
1819  */
1820 class ImportStreamSource implements ImportSource {
1821         function __construct( $handle ) {
1822                 $this->mHandle = $handle;
1823         }
1824
1825         /**
1826          * @return bool
1827          */
1828         function atEnd() {
1829                 return feof( $this->mHandle );
1830         }
1831
1832         /**
1833          * @return string
1834          */
1835         function readChunk() {
1836                 return fread( $this->mHandle, 32768 );
1837         }
1838
1839         /**
1840          * @param string $filename
1841          * @return Status
1842          */
1843         static function newFromFile( $filename ) {
1844                 wfSuppressWarnings();
1845                 $file = fopen( $filename, 'rt' );
1846                 wfRestoreWarnings();
1847                 if ( !$file ) {
1848                         return Status::newFatal( "importcantopen" );
1849                 }
1850                 return Status::newGood( new ImportStreamSource( $file ) );
1851         }
1852
1853         /**
1854          * @param string $fieldname
1855          * @return Status
1856          */
1857         static function newFromUpload( $fieldname = "xmlimport" ) {
1858                 $upload =& $_FILES[$fieldname];
1859
1860                 if ( $upload === null || !$upload['name'] ) {
1861                         return Status::newFatal( 'importnofile' );
1862                 }
1863                 if ( !empty( $upload['error'] ) ) {
1864                         switch ( $upload['error'] ) {
1865                                 case 1:
1866                                         # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1867                                         return Status::newFatal( 'importuploaderrorsize' );
1868                                 case 2:
1869                                         # The uploaded file exceeds the MAX_FILE_SIZE directive that
1870                                         # was specified in the HTML form.
1871                                         return Status::newFatal( 'importuploaderrorsize' );
1872                                 case 3:
1873                                         # The uploaded file was only partially uploaded
1874                                         return Status::newFatal( 'importuploaderrorpartial' );
1875                                 case 6:
1876                                         # Missing a temporary folder.
1877                                         return Status::newFatal( 'importuploaderrortemp' );
1878                                 # case else: # Currently impossible
1879                         }
1880
1881                 }
1882                 $fname = $upload['tmp_name'];
1883                 if ( is_uploaded_file( $fname ) ) {
1884                         return ImportStreamSource::newFromFile( $fname );
1885                 } else {
1886                         return Status::newFatal( 'importnofile' );
1887                 }
1888         }
1889
1890         /**
1891          * @param string $url
1892          * @param string $method
1893          * @return Status
1894          */
1895         static function newFromURL( $url, $method = 'GET' ) {
1896                 wfDebug( __METHOD__ . ": opening $url\n" );
1897                 # Use the standard HTTP fetch function; it times out
1898                 # quicker and sorts out user-agent problems which might
1899                 # otherwise prevent importing from large sites, such
1900                 # as the Wikimedia cluster, etc.
1901                 $data = Http::request( $method, $url, array( 'followRedirects' => true ), __METHOD__ );
1902                 if ( $data !== false ) {
1903                         $file = tmpfile();
1904                         fwrite( $file, $data );
1905                         fflush( $file );
1906                         fseek( $file, 0 );
1907                         return Status::newGood( new ImportStreamSource( $file ) );
1908                 } else {
1909                         return Status::newFatal( 'importcantopen' );
1910                 }
1911         }
1912
1913         /**
1914          * @param string $interwiki
1915          * @param string $page
1916          * @param bool $history
1917          * @param bool $templates
1918          * @param int $pageLinkDepth
1919          * @return Status
1920          */
1921         public static function newFromInterwiki( $interwiki, $page, $history = false,
1922                 $templates = false, $pageLinkDepth = 0
1923         ) {
1924                 if ( $page == '' ) {
1925                         return Status::newFatal( 'import-noarticle' );
1926                 }
1927                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1928                 if ( is_null( $link ) || !$link->isExternal() ) {
1929                         return Status::newFatal( 'importbadinterwiki' );
1930                 } else {
1931                         $params = array();
1932                         if ( $history ) {
1933                                 $params['history'] = 1;
1934                         }
1935                         if ( $templates ) {
1936                                 $params['templates'] = 1;
1937                         }
1938                         if ( $pageLinkDepth ) {
1939                                 $params['pagelink-depth'] = $pageLinkDepth;
1940                         }
1941                         $url = $link->getFullURL( $params );
1942                         # For interwikis, use POST to avoid redirects.
1943                         return ImportStreamSource::newFromURL( $url, "POST" );
1944                 }
1945         }
1946 }