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                 if ( $content === null ) {
 379                         wfDebug( __METHOD__ . ': Skipping article count adjustment for ' . $title .
 380                                 ' because WikiPage::getContent() returned null' );
 381                 } else {
 382                         $editInfo = $page->prepareContentForEdit( $content );
 383                         $countKey = 'title_' . $title->getPrefixedText();
 384                         $countable = $page->isCountable( $editInfo );
 385                         if ( array_key_exists( $countKey, $this->countableCache ) &&
 386                                 $countable != $this->countableCache[ $countKey ] ) {
 387                                 DeferredUpdates::addUpdate( SiteStatsUpdate::factory( array(
 388                                         'articles' => ( (int)$countable - (int)$this->countableCache[ $countKey ] )
 389                                 ) ) );
 390                         }
 391                 }
 392
 393                 $args = func_get_args();
 394                 return Hooks::run( 'AfterImportPage', $args );
 395         }
 396
 397         /**
 398          * Alternate per-revision callback, for debugging.
 399          * @param WikiRevision $revision
 400          */
 401         public function debugRevisionHandler( &$revision ) {
 402                 $this->debug( "Got revision:" );
 403                 if ( is_object( $revision->title ) ) {
 404                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 405                 } else {
 406                         $this->debug( "-- Title: <invalid>" );
 407                 }
 408                 $this->debug( "-- User: " . $revision->user_text );
 409                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 410                 $this->debug( "-- Comment: " . $revision->comment );
 411                 $this->debug( "-- Text: " . $revision->text );
 412         }
 413
 414         /**
 415          * Notify the callback function of site info
 416          * @param array $siteInfo
 417          * @return bool|mixed
 418          */
 419         private function siteInfoCallback( $siteInfo ) {
 420                 if ( isset( $this->mSiteInfoCallback ) ) {
 421                         return call_user_func_array( $this->mSiteInfoCallback,
 422                                         array( $siteInfo, $this ) );
 423                 } else {
 424                         return false;
 425                 }
 426         }
 427
 428         /**
 429          * Notify the callback function when a new "<page>" is reached.
 430          * @param Title $title
 431          */
 432         function pageCallback( $title ) {
 433                 if ( isset( $this->mPageCallback ) ) {
 434                         call_user_func( $this->mPageCallback, $title );
 435                 }
 436         }
 437
 438         /**
 439          * Notify the callback function when a "</page>" is closed.
 440          * @param Title $title
 441          * @param ForeignTitle $foreignTitle
 442          * @param int $revCount
 443          * @param int $sucCount Number of revisions for which callback returned true
 444          * @param array $pageInfo Associative array of page information
 445          */
 446         private function pageOutCallback( $title, $foreignTitle, $revCount,
 447                         $sucCount, $pageInfo ) {
 448                 if ( isset( $this->mPageOutCallback ) ) {
 449                         $args = func_get_args();
 450                         call_user_func_array( $this->mPageOutCallback, $args );
 451                 }
 452         }
 453
 454         /**
 455          * Notify the callback function of a revision
 456          * @param WikiRevision $revision
 457          * @return bool|mixed
 458          */
 459         private function revisionCallback( $revision ) {
 460                 if ( isset( $this->mRevisionCallback ) ) {
 461                         return call_user_func_array( $this->mRevisionCallback,
 462                                         array( $revision, $this ) );
 463                 } else {
 464                         return false;
 465                 }
 466         }
 467
 468         /**
 469          * Notify the callback function of a new log item
 470          * @param WikiRevision $revision
 471          * @return bool|mixed
 472          */
 473         private function logItemCallback( $revision ) {
 474                 if ( isset( $this->mLogItemCallback ) ) {
 475                         return call_user_func_array( $this->mLogItemCallback,
 476                                         array( $revision, $this ) );
 477                 } else {
 478                         return false;
 479                 }
 480         }
 481
 482         /**
 483          * Retrieves the contents of the named attribute of the current element.
 484          * @param string $attr The name of the attribute
 485          * @return string The value of the attribute or an empty string if it is not set in the current
 486          * element.
 487          */
 488         public function nodeAttribute( $attr ) {
 489                 return $this->reader->getAttribute( $attr );
 490         }
 491
 492         /**
 493          * Shouldn't something like this be built-in to XMLReader?
 494          * Fetches text contents of the current element, assuming
 495          * no sub-elements or such scary things.
 496          * @return string
 497          * @access private
 498          */
 499         public function nodeContents() {
 500                 if ( $this->reader->isEmptyElement ) {
 501                         return "";
 502                 }
 503                 $buffer = "";
 504                 while ( $this->reader->read() ) {
 505                         switch ( $this->reader->nodeType ) {
 506                         case XMLReader::TEXT:
 507                         case XMLReader::SIGNIFICANT_WHITESPACE:
 508                                 $buffer .= $this->reader->value;
 509                                 break;
 510                         case XMLReader::END_ELEMENT:
 511                                 return $buffer;
 512                         }
 513                 }
 514
 515                 $this->reader->close();
 516                 return '';
 517         }
 518
 519         /**
 520          * Primary entry point
 521          * @throws MWException
 522          * @return bool
 523          */
 524         public function doImport() {
 525                 // Calls to reader->read need to be wrapped in calls to
 526                 // libxml_disable_entity_loader() to avoid local file
 527                 // inclusion attacks (bug 46932).
 528                 $oldDisable = libxml_disable_entity_loader( true );
 529                 $this->reader->read();
 530
 531                 if ( $this->reader->name != 'mediawiki' ) {
 532                         libxml_disable_entity_loader( $oldDisable );
 533                         throw new MWException( "Expected <mediawiki> tag, got " .
 534                                 $this->reader->name );
 535                 }
 536                 $this->debug( "<mediawiki> tag is correct." );
 537
 538                 $this->debug( "Starting primary dump processing loop." );
 539
 540                 $keepReading = $this->reader->read();
 541                 $skip = false;
 542                 $rethrow = null;
 543                 try {
 544                         while ( $keepReading ) {
 545                                 $tag = $this->reader->name;
 546                                 $type = $this->reader->nodeType;
 547
 548                                 if ( !Hooks::run( 'ImportHandleToplevelXMLTag', array( $this ) ) ) {
 549                                         // Do nothing
 550                                 } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) {
 551                                         break;
 552                                 } elseif ( $tag == 'siteinfo' ) {
 553                                         $this->handleSiteInfo();
 554                                 } elseif ( $tag == 'page' ) {
 555                                         $this->handlePage();
 556                                 } elseif ( $tag == 'logitem' ) {
 557                                         $this->handleLogItem();
 558                                 } elseif ( $tag != '#text' ) {
 559                                         $this->warn( "Unhandled top-level XML tag $tag" );
 560
 561                                         $skip = true;
 562                                 }
 563
 564                                 if ( $skip ) {
 565                                         $keepReading = $this->reader->next();
 566                                         $skip = false;
 567                                         $this->debug( "Skip" );
 568                                 } else {
 569                                         $keepReading = $this->reader->read();
 570                                 }
 571                         }
 572                 } catch ( Exception $ex ) {
 573                         $rethrow = $ex;
 574                 }
 575
 576                 // finally
 577                 libxml_disable_entity_loader( $oldDisable );
 578                 $this->reader->close();
 579
 580                 if ( $rethrow ) {
 581                         throw $rethrow;
 582                 }
 583
 584                 return true;
 585         }
 586
 587         private function handleSiteInfo() {
 588                 $this->debug( "Enter site info handler." );
 589                 $siteInfo = array();
 590
 591                 // Fields that can just be stuffed in the siteInfo object
 592                 $normalFields = array( 'sitename', 'base', 'generator', 'case' );
 593
 594                 while ( $this->reader->read() ) {
 595                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 596                                         $this->reader->name == 'siteinfo' ) {
 597                                 break;
 598                         }
 599
 600                         $tag = $this->reader->name;
 601
 602                         if ( $tag == 'namespace' ) {
 603                                 $this->foreignNamespaces[ $this->nodeAttribute( 'key' ) ] =
 604                                         $this->nodeContents();
 605                         } elseif ( in_array( $tag, $normalFields ) ) {
 606                                 $siteInfo[$tag] = $this->nodeContents();
 607                         }
 608                 }
 609
 610                 $siteInfo['_namespaces'] = $this->foreignNamespaces;
 611                 $this->siteInfoCallback( $siteInfo );
 612         }
 613
 614         private function handleLogItem() {
 615                 $this->debug( "Enter log item handler." );
 616                 $logInfo = array();
 617
 618                 // Fields that can just be stuffed in the pageInfo object
 619                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 620                                         'logtitle', 'params' );
 621
 622                 while ( $this->reader->read() ) {
 623                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 624                                         $this->reader->name == 'logitem' ) {
 625                                 break;
 626                         }
 627
 628                         $tag = $this->reader->name;
 629
 630                         if ( !Hooks::run( 'ImportHandleLogItemXMLTag', array(
 631                                 $this, $logInfo
 632                         ) ) ) {
 633                                 // Do nothing
 634                         } elseif ( in_array( $tag, $normalFields ) ) {
 635                                 $logInfo[$tag] = $this->nodeContents();
 636                         } elseif ( $tag == 'contributor' ) {
 637                                 $logInfo['contributor'] = $this->handleContributor();
 638                         } elseif ( $tag != '#text' ) {
 639                                 $this->warn( "Unhandled log-item XML tag $tag" );
 640                         }
 641                 }
 642
 643                 $this->processLogItem( $logInfo );
 644         }
 645
 646         /**
 647          * @param array $logInfo
 648          * @return bool|mixed
 649          */
 650         private function processLogItem( $logInfo ) {
 651                 $revision = new WikiRevision( $this->config );
 652
 653                 $revision->setID( $logInfo['id'] );
 654                 $revision->setType( $logInfo['type'] );
 655                 $revision->setAction( $logInfo['action'] );
 656                 $revision->setTimestamp( $logInfo['timestamp'] );
 657                 $revision->setParams( $logInfo['params'] );
 658                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 659                 $revision->setNoUpdates( $this->mNoUpdates );
 660
 661                 if ( isset( $logInfo['comment'] ) ) {
 662                         $revision->setComment( $logInfo['comment'] );
 663                 }
 664
 665                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 666                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 667                 }
 668                 if ( isset( $logInfo['contributor']['username'] ) ) {
 669                         $revision->setUserName( $logInfo['contributor']['username'] );
 670                 }
 671
 672                 return $this->logItemCallback( $revision );
 673         }
 674
 675         private function handlePage() {
 676                 // Handle page data.
 677                 $this->debug( "Enter page handler." );
 678                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 679
 680                 // Fields that can just be stuffed in the pageInfo object
 681                 $normalFields = array( 'title', 'ns', 'id', 'redirect', 'restrictions' );
 682
 683                 $skip = false;
 684                 $badTitle = false;
 685
 686                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 687                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 688                                         $this->reader->name == 'page' ) {
 689                                 break;
 690                         }
 691
 692                         $skip = false;
 693
 694                         $tag = $this->reader->name;
 695
 696                         if ( $badTitle ) {
 697                                 // The title is invalid, bail out of this page
 698                                 $skip = true;
 699                         } elseif ( !Hooks::run( 'ImportHandlePageXMLTag', array( $this,
 700                                                 &$pageInfo ) ) ) {
 701                                 // Do nothing
 702                         } elseif ( in_array( $tag, $normalFields ) ) {
 703                                 // An XML snippet:
 704                                 // <page>
 705                                 //     <id>123</id>
 706                                 //     <title>Page</title>
 707                                 //     <redirect title="NewTitle"/>
 708                                 //     ...
 709                                 // Because the redirect tag is built differently, we need special handling for that case.
 710                                 if ( $tag == 'redirect' ) {
 711                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 712                                 } else {
 713                                         $pageInfo[$tag] = $this->nodeContents();
 714                                 }
 715                         } elseif ( $tag == 'revision' || $tag == 'upload' ) {
 716                                 if ( !isset( $title ) ) {
 717                                         $title = $this->processTitle( $pageInfo['title'],
 718                                                 isset( $pageInfo['ns'] ) ? $pageInfo['ns'] : null );
 719
 720                                         if ( !$title ) {
 721                                                 $badTitle = true;
 722                                                 $skip = true;
 723                                         }
 724
 725                                         $this->pageCallback( $title );
 726                                         list( $pageInfo['_title'], $foreignTitle ) = $title;
 727                                 }
 728
 729                                 if ( $title ) {
 730                                         if ( $tag == 'revision' ) {
 731                                                 $this->handleRevision( $pageInfo );
 732                                         } else {
 733                                                 $this->handleUpload( $pageInfo );
 734                                         }
 735                                 }
 736                         } elseif ( $tag != '#text' ) {
 737                                 $this->warn( "Unhandled page XML tag $tag" );
 738                                 $skip = true;
 739                         }
 740                 }
 741
 742                 $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
 743                                         $pageInfo['revisionCount'],
 744                                         $pageInfo['successfulRevisionCount'],
 745                                         $pageInfo );
 746         }
 747
 748         /**
 749          * @param array $pageInfo
 750          */
 751         private function handleRevision( &$pageInfo ) {
 752                 $this->debug( "Enter revision handler" );
 753                 $revisionInfo = array();
 754
 755                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' );
 756
 757                 $skip = false;
 758
 759                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 760                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 761                                         $this->reader->name == 'revision' ) {
 762                                 break;
 763                         }
 764
 765                         $tag = $this->reader->name;
 766
 767                         if ( !Hooks::run( 'ImportHandleRevisionXMLTag', array(
 768                                 $this, $pageInfo, $revisionInfo
 769                         ) ) ) {
 770                                 // Do nothing
 771                         } elseif ( in_array( $tag, $normalFields ) ) {
 772                                 $revisionInfo[$tag] = $this->nodeContents();
 773                         } elseif ( $tag == 'contributor' ) {
 774                                 $revisionInfo['contributor'] = $this->handleContributor();
 775                         } elseif ( $tag != '#text' ) {
 776                                 $this->warn( "Unhandled revision XML tag $tag" );
 777                                 $skip = true;
 778                         }
 779                 }
 780
 781                 $pageInfo['revisionCount']++;
 782                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 783                         $pageInfo['successfulRevisionCount']++;
 784                 }
 785         }
 786
 787         /**
 788          * @param array $pageInfo
 789          * @param array $revisionInfo
 790          * @return bool|mixed
 791          */
 792         private function processRevision( $pageInfo, $revisionInfo ) {
 793                 $revision = new WikiRevision( $this->config );
 794
 795                 if ( isset( $revisionInfo['id'] ) ) {
 796                         $revision->setID( $revisionInfo['id'] );
 797                 }
 798                 if ( isset( $revisionInfo['model'] ) ) {
 799                         $revision->setModel( $revisionInfo['model'] );
 800                 }
 801                 if ( isset( $revisionInfo['format'] ) ) {
 802                         $revision->setFormat( $revisionInfo['format'] );
 803                 }
 804                 $revision->setTitle( $pageInfo['_title'] );
 805
 806                 if ( isset( $revisionInfo['text'] ) ) {
 807                         $handler = $revision->getContentHandler();
 808                         $text = $handler->importTransform(
 809                                 $revisionInfo['text'],
 810                                 $revision->getFormat() );
 811
 812                         $revision->setText( $text );
 813                 }
 814                 if ( isset( $revisionInfo['timestamp'] ) ) {
 815                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 816                 } else {
 817                         $revision->setTimestamp( wfTimestampNow() );
 818                 }
 819
 820                 if ( isset( $revisionInfo['comment'] ) ) {
 821                         $revision->setComment( $revisionInfo['comment'] );
 822                 }
 823
 824                 if ( isset( $revisionInfo['minor'] ) ) {
 825                         $revision->setMinor( true );
 826                 }
 827                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 828                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 829                 }
 830                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 831                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 832                 }
 833                 $revision->setNoUpdates( $this->mNoUpdates );
 834
 835                 return $this->revisionCallback( $revision );
 836         }
 837
 838         /**
 839          * @param array $pageInfo
 840          * @return mixed
 841          */
 842         private function handleUpload( &$pageInfo ) {
 843                 $this->debug( "Enter upload handler" );
 844                 $uploadInfo = array();
 845
 846                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 847                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 848
 849                 $skip = false;
 850
 851                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 852                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 853                                         $this->reader->name == 'upload' ) {
 854                                 break;
 855                         }
 856
 857                         $tag = $this->reader->name;
 858
 859                         if ( !Hooks::run( 'ImportHandleUploadXMLTag', array(
 860                                 $this, $pageInfo
 861                         ) ) ) {
 862                                 // Do nothing
 863                         } elseif ( in_array( $tag, $normalFields ) ) {
 864                                 $uploadInfo[$tag] = $this->nodeContents();
 865                         } elseif ( $tag == 'contributor' ) {
 866                                 $uploadInfo['contributor'] = $this->handleContributor();
 867                         } elseif ( $tag == 'contents' ) {
 868                                 $contents = $this->nodeContents();
 869                                 $encoding = $this->reader->getAttribute( 'encoding' );
 870                                 if ( $encoding === 'base64' ) {
 871                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 872                                         $uploadInfo['isTempSrc'] = true;
 873                                 }
 874                         } elseif ( $tag != '#text' ) {
 875                                 $this->warn( "Unhandled upload XML tag $tag" );
 876                                 $skip = true;
 877                         }
 878                 }
 879
 880                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 881                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 882                         if ( file_exists( $path ) ) {
 883                                 $uploadInfo['fileSrc'] = $path;
 884                                 $uploadInfo['isTempSrc'] = false;
 885                         }
 886                 }
 887
 888                 if ( $this->mImportUploads ) {
 889                         return $this->processUpload( $pageInfo, $uploadInfo );
 890                 }
 891         }
 892
 893         /**
 894          * @param string $contents
 895          * @return string
 896          */
 897         private function dumpTemp( $contents ) {
 898                 $filename = tempnam( wfTempDir(), 'importupload' );
 899                 file_put_contents( $filename, $contents );
 900                 return $filename;
 901         }
 902
 903         /**
 904          * @param array $pageInfo
 905          * @param array $uploadInfo
 906          * @return mixed
 907          */
 908         private function processUpload( $pageInfo, $uploadInfo ) {
 909                 $revision = new WikiRevision( $this->config );
 910                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 911
 912                 $revision->setTitle( $pageInfo['_title'] );
 913                 $revision->setID( $pageInfo['id'] );
 914                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 915                 $revision->setText( $text );
 916                 $revision->setFilename( $uploadInfo['filename'] );
 917                 if ( isset( $uploadInfo['archivename'] ) ) {
 918                         $revision->setArchiveName( $uploadInfo['archivename'] );
 919                 }
 920                 $revision->setSrc( $uploadInfo['src'] );
 921                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 922                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 923                                 !empty( $uploadInfo['isTempSrc'] ) );
 924                 }
 925                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 926                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 927                 }
 928                 $revision->setSize( intval( $uploadInfo['size'] ) );
 929                 $revision->setComment( $uploadInfo['comment'] );
 930
 931                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 932                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 933                 }
 934                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 935                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 936                 }
 937                 $revision->setNoUpdates( $this->mNoUpdates );
 938
 939                 return call_user_func( $this->mUploadCallback, $revision );
 940         }
 941
 942         /**
 943          * @return array
 944          */
 945         private function handleContributor() {
 946                 $fields = array( 'id', 'ip', 'username' );
 947                 $info = array();
 948
 949                 while ( $this->reader->read() ) {
 950                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 951                                         $this->reader->name == 'contributor' ) {
 952                                 break;
 953                         }
 954
 955                         $tag = $this->reader->name;
 956
 957                         if ( in_array( $tag, $fields ) ) {
 958                                 $info[$tag] = $this->nodeContents();
 959                         }
 960                 }
 961
 962                 return $info;
 963         }
 964
 965         /**
 966          * @param string $text
 967          * @param string|null $ns
 968          * @return array|bool
 969          */
 970         private function processTitle( $text, $ns = null ) {
 971                 if ( is_null( $this->foreignNamespaces ) ) {
 972                         $foreignTitleFactory = new NaiveForeignTitleFactory();
 973                 } else {
 974                         $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
 975                                 $this->foreignNamespaces );
 976                 }
 977
 978                 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
 979                         intval( $ns ) );
 980
 981                 $title = $this->importTitleFactory->createTitleFromForeignTitle(
 982                         $foreignTitle );
 983
 984                 $commandLineMode = $this->config->get( 'CommandLineMode' );
 985                 if ( is_null( $title ) ) {
 986                         # Invalid page title? Ignore the page
 987                         $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
 988                         return false;
 989                 } elseif ( $title->isExternal() ) {
 990                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
 991                         return false;
 992                 } elseif ( !$title->canExist() ) {
 993                         $this->notice( 'import-error-special', $title->getPrefixedText() );
 994                         return false;
 995                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
 996                         # Do not import if the importing wiki user cannot edit this page
 997                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
 998                         return false;
 999                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
1000                         # Do not import if the importing wiki user cannot create this page
1001                         $this->notice( 'import-error-create', $title->getPrefixedText() );
1002                         return false;
1003                 }
1004
1005                 return array( $title, $foreignTitle );
1006         }
1007 }
1008
1009 /** This is a horrible hack used to keep source compatibility */
1010 class UploadSourceAdapter {
1011         /** @var array */
1012         public static $sourceRegistrations = array();
1013
1014         /** @var string */
1015         private $mSource;
1016
1017         /** @var string */
1018         private $mBuffer;
1019
1020         /** @var int */
1021         private $mPosition;
1022
1023         /**
1024          * @param ImportSource $source
1025          * @return string
1026          */
1027         static function registerSource( ImportSource $source ) {
1028                 $id = wfRandomString();
1029
1030                 self::$sourceRegistrations[$id] = $source;
1031
1032                 return $id;
1033         }
1034
1035         /**
1036          * @param string $path
1037          * @param string $mode
1038          * @param array $options
1039          * @param string $opened_path
1040          * @return bool
1041          */
1042         function stream_open( $path, $mode, $options, &$opened_path ) {
1043                 $url = parse_url( $path );
1044                 $id = $url['host'];
1045
1046                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
1047                         return false;
1048                 }
1049
1050                 $this->mSource = self::$sourceRegistrations[$id];
1051
1052                 return true;
1053         }
1054
1055         /**
1056          * @param int $count
1057          * @return string
1058          */
1059         function stream_read( $count ) {
1060                 $return = '';
1061                 $leave = false;
1062
1063                 while ( !$leave && !$this->mSource->atEnd() &&
1064                                 strlen( $this->mBuffer ) < $count ) {
1065                         $read = $this->mSource->readChunk();
1066
1067                         if ( !strlen( $read ) ) {
1068                                 $leave = true;
1069                         }
1070
1071                         $this->mBuffer .= $read;
1072                 }
1073
1074                 if ( strlen( $this->mBuffer ) ) {
1075                         $return = substr( $this->mBuffer, 0, $count );
1076                         $this->mBuffer = substr( $this->mBuffer, $count );
1077                 }
1078
1079                 $this->mPosition += strlen( $return );
1080
1081                 return $return;
1082         }
1083
1084         /**
1085          * @param string $data
1086          * @return bool
1087          */
1088         function stream_write( $data ) {
1089                 return false;
1090         }
1091
1092         /**
1093          * @return mixed
1094          */
1095         function stream_tell() {
1096                 return $this->mPosition;
1097         }
1098
1099         /**
1100          * @return bool
1101          */
1102         function stream_eof() {
1103                 return $this->mSource->atEnd();
1104         }
1105
1106         /**
1107          * @return array
1108          */
1109         function url_stat() {
1110                 $result = array();
1111
1112                 $result['dev'] = $result[0] = 0;
1113                 $result['ino'] = $result[1] = 0;
1114                 $result['mode'] = $result[2] = 0;
1115                 $result['nlink'] = $result[3] = 0;
1116                 $result['uid'] = $result[4] = 0;
1117                 $result['gid'] = $result[5] = 0;
1118                 $result['rdev'] = $result[6] = 0;
1119                 $result['size'] = $result[7] = 0;
1120                 $result['atime'] = $result[8] = 0;
1121                 $result['mtime'] = $result[9] = 0;
1122                 $result['ctime'] = $result[10] = 0;
1123                 $result['blksize'] = $result[11] = 0;
1124                 $result['blocks'] = $result[12] = 0;
1125
1126                 return $result;
1127         }
1128 }
1129
1130 /**
1131  * @todo document (e.g. one-sentence class description).
1132  * @ingroup SpecialPage
1133  */
1134 class WikiRevision {
1135         /** @todo Unused? */
1136         public $importer = null;
1137
1138         /** @var Title */
1139         public $title = null;
1140
1141         /** @var int */
1142         public $id = 0;
1143
1144         /** @var string */
1145         public $timestamp = "20010115000000";
1146
1147         /**
1148          * @var int
1149          * @todo Can't find any uses. Public, because that's suspicious. Get clarity. */
1150         public $user = 0;
1151
1152         /** @var string */
1153         public $user_text = "";
1154
1155         /** @var string */
1156         public $model = null;
1157
1158         /** @var string */
1159         public $format = null;
1160
1161         /** @var string */
1162         public $text = "";
1163
1164         /** @var int */
1165         protected $size;
1166
1167         /** @var Content */
1168         public $content = null;
1169
1170         /** @var ContentHandler */
1171         protected $contentHandler = null;
1172
1173         /** @var string */
1174         public $comment = "";
1175
1176         /** @var bool */
1177         public $minor = false;
1178
1179         /** @var string */
1180         public $type = "";
1181
1182         /** @var string */
1183         public $action = "";
1184
1185         /** @var string */
1186         public $params = "";
1187
1188         /** @var string */
1189         public $fileSrc = '';
1190
1191         /** @var bool|string */
1192         public $sha1base36 = false;
1193
1194         /**
1195          * @var bool
1196          * @todo Unused?
1197          */
1198         public $isTemp = false;
1199
1200         /** @var string */
1201         public $archiveName = '';
1202
1203         protected $filename;
1204
1205         /** @var mixed */
1206         protected $src;
1207
1208         /** @todo Unused? */
1209         public $fileIsTemp;
1210
1211         /** @var bool */
1212         private $mNoUpdates = false;
1213
1214         /** @var Config $config */
1215         private $config;
1216
1217         public function __construct( Config $config ) {
1218                 $this->config = $config;
1219         }
1220
1221         /**
1222          * @param Title $title
1223          * @throws MWException
1224          */
1225         function setTitle( $title ) {
1226                 if ( is_object( $title ) ) {
1227                         $this->title = $title;
1228                 } elseif ( is_null( $title ) ) {
1229                         throw new MWException( "WikiRevision given a null title in import. "
1230                                 . "You may need to adjust \$wgLegalTitleChars." );
1231                 } else {
1232                         throw new MWException( "WikiRevision given non-object title in import." );
1233                 }
1234         }
1235
1236         /**
1237          * @param int $id
1238          */
1239         function setID( $id ) {
1240                 $this->id = $id;
1241         }
1242
1243         /**
1244          * @param string $ts
1245          */
1246         function setTimestamp( $ts ) {
1247                 # 2003-08-05T18:30:02Z
1248                 $this->timestamp = wfTimestamp( TS_MW, $ts );
1249         }
1250
1251         /**
1252          * @param string $user
1253          */
1254         function setUsername( $user ) {
1255                 $this->user_text = $user;
1256         }
1257
1258         /**
1259          * @param string $ip
1260          */
1261         function setUserIP( $ip ) {
1262                 $this->user_text = $ip;
1263         }
1264
1265         /**
1266          * @param string $model
1267          */
1268         function setModel( $model ) {
1269                 $this->model = $model;
1270         }
1271
1272         /**
1273          * @param string $format
1274          */
1275         function setFormat( $format ) {
1276                 $this->format = $format;
1277         }
1278
1279         /**
1280          * @param string $text
1281          */
1282         function setText( $text ) {
1283                 $this->text = $text;
1284         }
1285
1286         /**
1287          * @param string $text
1288          */
1289         function setComment( $text ) {
1290                 $this->comment = $text;
1291         }
1292
1293         /**
1294          * @param bool $minor
1295          */
1296         function setMinor( $minor ) {
1297                 $this->minor = (bool)$minor;
1298         }
1299
1300         /**
1301          * @param mixed $src
1302          */
1303         function setSrc( $src ) {
1304                 $this->src = $src;
1305         }
1306
1307         /**
1308          * @param string $src
1309          * @param bool $isTemp
1310          */
1311         function setFileSrc( $src, $isTemp ) {
1312                 $this->fileSrc = $src;
1313                 $this->fileIsTemp = $isTemp;
1314         }
1315
1316         /**
1317          * @param string $sha1base36
1318          */
1319         function setSha1Base36( $sha1base36 ) {
1320                 $this->sha1base36 = $sha1base36;
1321         }
1322
1323         /**
1324          * @param string $filename
1325          */
1326         function setFilename( $filename ) {
1327                 $this->filename = $filename;
1328         }
1329
1330         /**
1331          * @param string $archiveName
1332          */
1333         function setArchiveName( $archiveName ) {
1334                 $this->archiveName = $archiveName;
1335         }
1336
1337         /**
1338          * @param int $size
1339          */
1340         function setSize( $size ) {
1341                 $this->size = intval( $size );
1342         }
1343
1344         /**
1345          * @param string $type
1346          */
1347         function setType( $type ) {
1348                 $this->type = $type;
1349         }
1350
1351         /**
1352          * @param string $action
1353          */
1354         function setAction( $action ) {
1355                 $this->action = $action;
1356         }
1357
1358         /**
1359          * @param array $params
1360          */
1361         function setParams( $params ) {
1362                 $this->params = $params;
1363         }
1364
1365         /**
1366          * @param bool $noupdates
1367          */
1368         public function setNoUpdates( $noupdates ) {
1369                 $this->mNoUpdates = $noupdates;
1370         }
1371
1372         /**
1373          * @return Title
1374          */
1375         function getTitle() {
1376                 return $this->title;
1377         }
1378
1379         /**
1380          * @return int
1381          */
1382         function getID() {
1383                 return $this->id;
1384         }
1385
1386         /**
1387          * @return string
1388          */
1389         function getTimestamp() {
1390                 return $this->timestamp;
1391         }
1392
1393         /**
1394          * @return string
1395          */
1396         function getUser() {
1397                 return $this->user_text;
1398         }
1399
1400         /**
1401          * @return string
1402          *
1403          * @deprecated Since 1.21, use getContent() instead.
1404          */
1405         function getText() {
1406                 ContentHandler::deprecated( __METHOD__, '1.21' );
1407
1408                 return $this->text;
1409         }
1410
1411         /**
1412          * @return ContentHandler
1413          */
1414         function getContentHandler() {
1415                 if ( is_null( $this->contentHandler ) ) {
1416                         $this->contentHandler = ContentHandler::getForModelID( $this->getModel() );
1417                 }
1418
1419                 return $this->contentHandler;
1420         }
1421
1422         /**
1423          * @return Content
1424          */
1425         function getContent() {
1426                 if ( is_null( $this->content ) ) {
1427                         $handler = $this->getContentHandler();
1428                         $this->content = $handler->unserializeContent( $this->text, $this->getFormat() );
1429                 }
1430
1431                 return $this->content;
1432         }
1433
1434         /**
1435          * @return string
1436          */
1437         function getModel() {
1438                 if ( is_null( $this->model ) ) {
1439                         $this->model = $this->getTitle()->getContentModel();
1440                 }
1441
1442                 return $this->model;
1443         }
1444
1445         /**
1446          * @return string
1447          */
1448         function getFormat() {
1449                 if ( is_null( $this->format ) ) {
1450                         $this->format = $this->getContentHandler()->getDefaultFormat();
1451                 }
1452
1453                 return $this->format;
1454         }
1455
1456         /**
1457          * @return string
1458          */
1459         function getComment() {
1460                 return $this->comment;
1461         }
1462
1463         /**
1464          * @return bool
1465          */
1466         function getMinor() {
1467                 return $this->minor;
1468         }
1469
1470         /**
1471          * @return mixed
1472          */
1473         function getSrc() {
1474                 return $this->src;
1475         }
1476
1477         /**
1478          * @return bool|string
1479          */
1480         function getSha1() {
1481                 if ( $this->sha1base36 ) {
1482                         return wfBaseConvert( $this->sha1base36, 36, 16 );
1483                 }
1484                 return false;
1485         }
1486
1487         /**
1488          * @return string
1489          */
1490         function getFileSrc() {
1491                 return $this->fileSrc;
1492         }
1493
1494         /**
1495          * @return bool
1496          */
1497         function isTempSrc() {
1498                 return $this->isTemp;
1499         }
1500
1501         /**
1502          * @return mixed
1503          */
1504         function getFilename() {
1505                 return $this->filename;
1506         }
1507
1508         /**
1509          * @return string
1510          */
1511         function getArchiveName() {
1512                 return $this->archiveName;
1513         }
1514
1515         /**
1516          * @return mixed
1517          */
1518         function getSize() {
1519                 return $this->size;
1520         }
1521
1522         /**
1523          * @return string
1524          */
1525         function getType() {
1526                 return $this->type;
1527         }
1528
1529         /**
1530          * @return string
1531          */
1532         function getAction() {
1533                 return $this->action;
1534         }
1535
1536         /**
1537          * @return string
1538          */
1539         function getParams() {
1540                 return $this->params;
1541         }
1542
1543         /**
1544          * @return bool
1545          */
1546         function importOldRevision() {
1547                 $dbw = wfGetDB( DB_MASTER );
1548
1549                 # Sneak a single revision into place
1550                 $user = User::newFromName( $this->getUser() );
1551                 if ( $user ) {
1552                         $userId = intval( $user->getId() );
1553                         $userText = $user->getName();
1554                         $userObj = $user;
1555                 } else {
1556                         $userId = 0;
1557                         $userText = $this->getUser();
1558                         $userObj = new User;
1559                 }
1560
1561                 // avoid memory leak...?
1562                 $linkCache = LinkCache::singleton();
1563                 $linkCache->clear();
1564
1565                 $page = WikiPage::factory( $this->title );
1566                 $page->loadPageData( 'fromdbmaster' );
1567                 if ( !$page->exists() ) {
1568                         # must create the page...
1569                         $pageId = $page->insertOn( $dbw );
1570                         $created = true;
1571                         $oldcountable = null;
1572                 } else {
1573                         $pageId = $page->getId();
1574                         $created = false;
1575
1576                         $prior = $dbw->selectField( 'revision', '1',
1577                                 array( 'rev_page' => $pageId,
1578                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1579                                         'rev_user_text' => $userText,
1580                                         'rev_comment' => $this->getComment() ),
1581                                 __METHOD__
1582                         );
1583                         if ( $prior ) {
1584                                 // @todo FIXME: This could fail slightly for multiple matches :P
1585                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1586                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1587                                 return false;
1588                         }
1589                 }
1590
1591                 # @todo FIXME: Use original rev_id optionally (better for backups)
1592                 # Insert the row
1593                 $revision = new Revision( array(
1594                         'title' => $this->title,
1595                         'page' => $pageId,
1596                         'content_model' => $this->getModel(),
1597                         'content_format' => $this->getFormat(),
1598                         //XXX: just set 'content' => $this->getContent()?
1599                         'text' => $this->getContent()->serialize( $this->getFormat() ),
1600                         'comment' => $this->getComment(),
1601                         'user' => $userId,
1602                         'user_text' => $userText,
1603                         'timestamp' => $this->timestamp,
1604                         'minor_edit' => $this->minor,
1605                         ) );
1606                 $revision->insertOn( $dbw );
1607                 $changed = $page->updateIfNewerOn( $dbw, $revision );
1608
1609                 if ( $changed !== false && !$this->mNoUpdates ) {
1610                         wfDebug( __METHOD__ . ": running updates\n" );
1611                         // countable/oldcountable stuff is handled in WikiImporter::finishImportPage
1612                         $page->doEditUpdates(
1613                                 $revision,
1614                                 $userObj,
1615                                 array( 'created' => $created, 'oldcountable' => 'no-change' )
1616                         );
1617                 }
1618
1619                 return true;
1620         }
1621
1622         function importLogItem() {
1623                 $dbw = wfGetDB( DB_MASTER );
1624                 # @todo FIXME: This will not record autoblocks
1625                 if ( !$this->getTitle() ) {
1626                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1627                                 $this->timestamp . "\n" );
1628                         return;
1629                 }
1630                 # Check if it exists already
1631                 // @todo FIXME: Use original log ID (better for backups)
1632                 $prior = $dbw->selectField( 'logging', '1',
1633                         array( 'log_type' => $this->getType(),
1634                                 'log_action' => $this->getAction(),
1635                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1636                                 'log_namespace' => $this->getTitle()->getNamespace(),
1637                                 'log_title' => $this->getTitle()->getDBkey(),
1638                                 'log_comment' => $this->getComment(),
1639                                 #'log_user_text' => $this->user_text,
1640                                 'log_params' => $this->params ),
1641                         __METHOD__
1642                 );
1643                 // @todo FIXME: This could fail slightly for multiple matches :P
1644                 if ( $prior ) {
1645                         wfDebug( __METHOD__
1646                                 . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp "
1647                                 . $this->timestamp . "\n" );
1648                         return;
1649                 }
1650                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1651                 $data = array(
1652                         'log_id' => $log_id,
1653                         'log_type' => $this->type,
1654                         'log_action' => $this->action,
1655                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1656                         'log_user' => User::idFromName( $this->user_text ),
1657                         #'log_user_text' => $this->user_text,
1658                         'log_namespace' => $this->getTitle()->getNamespace(),
1659                         'log_title' => $this->getTitle()->getDBkey(),
1660                         'log_comment' => $this->getComment(),
1661                         'log_params' => $this->params
1662                 );
1663                 $dbw->insert( 'logging', $data, __METHOD__ );
1664         }
1665
1666         /**
1667          * @return bool
1668          */
1669         function importUpload() {
1670                 # Construct a file
1671                 $archiveName = $this->getArchiveName();
1672                 if ( $archiveName ) {
1673                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1674                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1675                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1676                 } else {
1677                         $file = wfLocalFile( $this->getTitle() );
1678                         $file->load( File::READ_LATEST );
1679                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1680                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1681                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1682                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1683                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1684                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1685                         }
1686                 }
1687                 if ( !$file ) {
1688                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1689                         return false;
1690                 }
1691
1692                 # Get the file source or download if necessary
1693                 $source = $this->getFileSrc();
1694                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1695                 if ( !$source ) {
1696                         $source = $this->downloadSource();
1697                         $flags |= File::DELETE_SOURCE;
1698                 }
1699                 if ( !$source ) {
1700                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1701                         return false;
1702                 }
1703                 $sha1 = $this->getSha1();
1704                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1705                         if ( $flags & File::DELETE_SOURCE ) {
1706                                 # Broken file; delete it if it is a temporary file
1707                                 unlink( $source );
1708                         }
1709                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1710                         return false;
1711                 }
1712
1713                 $user = User::newFromName( $this->user_text );
1714
1715                 # Do the actual upload
1716                 if ( $archiveName ) {
1717                         $status = $file->uploadOld( $source, $archiveName,
1718                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1719                 } else {
1720                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1721                                 $flags, false, $this->getTimestamp(), $user );
1722                 }
1723
1724                 if ( $status->isGood() ) {
1725                         wfDebug( __METHOD__ . ": Successful\n" );
1726                         return true;
1727                 } else {
1728                         wfDebug( __METHOD__ . ': failed: ' . $status->getHTML() . "\n" );
1729                         return false;
1730                 }
1731         }
1732
1733         /**
1734          * @return bool|string
1735          */
1736         function downloadSource() {
1737                 if ( !$this->config->get( 'EnableUploads' ) ) {
1738                         return false;
1739                 }
1740
1741                 $tempo = tempnam( wfTempDir(), 'download' );
1742                 $f = fopen( $tempo, 'wb' );
1743                 if ( !$f ) {
1744                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1745                         return false;
1746                 }
1747
1748                 // @todo FIXME!
1749                 $src = $this->getSrc();
1750                 $data = Http::get( $src, array(), __METHOD__ );
1751                 if ( !$data ) {
1752                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1753                         fclose( $f );
1754                         unlink( $tempo );
1755                         return false;
1756                 }
1757
1758                 fwrite( $f, $data );
1759                 fclose( $f );
1760
1761                 return $tempo;
1762         }
1763
1764 }
1765
1766 /**
1767  * Source interface for XML import.
1768  */
1769 interface ImportSource {
1770
1771         /**
1772          * Indicates whether the end of the input has been reached.
1773          * Will return true after a finite number of calls to readChunk.
1774          *
1775          * @return bool true if there is no more input, false otherwise.
1776          */
1777         function atEnd();
1778
1779         /**
1780          * Return a chunk of the input, as a (possibly empty) string.
1781          * When the end of input is reached, readChunk() returns false.
1782          * If atEnd() returns false, readChunk() will return a string.
1783          * If atEnd() returns true, readChunk() will return false.
1784          *
1785          * @return bool|string
1786          */
1787         function readChunk();
1788 }
1789
1790 /**
1791  * Used for importing XML dumps where the content of the dump is in a string.
1792  * This class is ineffecient, and should only be used for small dumps.
1793  * For larger dumps, ImportStreamSource should be used instead.
1794  *
1795  * @ingroup SpecialPage
1796  */
1797 class ImportStringSource implements ImportSource {
1798         function __construct( $string ) {
1799                 $this->mString = $string;
1800                 $this->mRead = false;
1801         }
1802
1803         /**
1804          * @return bool
1805          */
1806         function atEnd() {
1807                 return $this->mRead;
1808         }
1809
1810         /**
1811          * @return bool|string
1812          */
1813         function readChunk() {
1814                 if ( $this->atEnd() ) {
1815                         return false;
1816                 }
1817                 $this->mRead = true;
1818                 return $this->mString;
1819         }
1820 }
1821
1822 /**
1823  * Imports a XML dump from a file (either from file upload, files on disk, or HTTP)
1824  * @ingroup SpecialPage
1825  */
1826 class ImportStreamSource implements ImportSource {
1827         function __construct( $handle ) {
1828                 $this->mHandle = $handle;
1829         }
1830
1831         /**
1832          * @return bool
1833          */
1834         function atEnd() {
1835                 return feof( $this->mHandle );
1836         }
1837
1838         /**
1839          * @return string
1840          */
1841         function readChunk() {
1842                 return fread( $this->mHandle, 32768 );
1843         }
1844
1845         /**
1846          * @param string $filename
1847          * @return Status
1848          */
1849         static function newFromFile( $filename ) {
1850                 wfSuppressWarnings();
1851                 $file = fopen( $filename, 'rt' );
1852                 wfRestoreWarnings();
1853                 if ( !$file ) {
1854                         return Status::newFatal( "importcantopen" );
1855                 }
1856                 return Status::newGood( new ImportStreamSource( $file ) );
1857         }
1858
1859         /**
1860          * @param string $fieldname
1861          * @return Status
1862          */
1863         static function newFromUpload( $fieldname = "xmlimport" ) {
1864                 $upload =& $_FILES[$fieldname];
1865
1866                 if ( $upload === null || !$upload['name'] ) {
1867                         return Status::newFatal( 'importnofile' );
1868                 }
1869                 if ( !empty( $upload['error'] ) ) {
1870                         switch ( $upload['error'] ) {
1871                                 case 1:
1872                                         # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1873                                         return Status::newFatal( 'importuploaderrorsize' );
1874                                 case 2:
1875                                         # The uploaded file exceeds the MAX_FILE_SIZE directive that
1876                                         # was specified in the HTML form.
1877                                         return Status::newFatal( 'importuploaderrorsize' );
1878                                 case 3:
1879                                         # The uploaded file was only partially uploaded
1880                                         return Status::newFatal( 'importuploaderrorpartial' );
1881                                 case 6:
1882                                         # Missing a temporary folder.
1883                                         return Status::newFatal( 'importuploaderrortemp' );
1884                                 # case else: # Currently impossible
1885                         }
1886
1887                 }
1888                 $fname = $upload['tmp_name'];
1889                 if ( is_uploaded_file( $fname ) ) {
1890                         return ImportStreamSource::newFromFile( $fname );
1891                 } else {
1892                         return Status::newFatal( 'importnofile' );
1893                 }
1894         }
1895
1896         /**
1897          * @param string $url
1898          * @param string $method
1899          * @return Status
1900          */
1901         static function newFromURL( $url, $method = 'GET' ) {
1902                 wfDebug( __METHOD__ . ": opening $url\n" );
1903                 # Use the standard HTTP fetch function; it times out
1904                 # quicker and sorts out user-agent problems which might
1905                 # otherwise prevent importing from large sites, such
1906                 # as the Wikimedia cluster, etc.
1907                 $data = Http::request( $method, $url, array( 'followRedirects' => true ), __METHOD__ );
1908                 if ( $data !== false ) {
1909                         $file = tmpfile();
1910                         fwrite( $file, $data );
1911                         fflush( $file );
1912                         fseek( $file, 0 );
1913                         return Status::newGood( new ImportStreamSource( $file ) );
1914                 } else {
1915                         return Status::newFatal( 'importcantopen' );
1916                 }
1917         }
1918
1919         /**
1920          * @param string $interwiki
1921          * @param string $page
1922          * @param bool $history
1923          * @param bool $templates
1924          * @param int $pageLinkDepth
1925          * @return Status
1926          */
1927         public static function newFromInterwiki( $interwiki, $page, $history = false,
1928                 $templates = false, $pageLinkDepth = 0
1929         ) {
1930                 if ( $page == '' ) {
1931                         return Status::newFatal( 'import-noarticle' );
1932                 }
1933                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1934                 if ( is_null( $link ) || !$link->isExternal() ) {
1935                         return Status::newFatal( 'importbadinterwiki' );
1936                 } else {
1937                         $params = array();
1938                         if ( $history ) {
1939                                 $params['history'] = 1;
1940                         }
1941                         if ( $templates ) {
1942                                 $params['templates'] = 1;
1943                         }
1944                         if ( $pageLinkDepth ) {
1945                                 $params['pagelink-depth'] = $pageLinkDepth;
1946                         }
1947                         $url = $link->getFullURL( $params );
1948                         # For interwikis, use POST to avoid redirects.
1949                         return ImportStreamSource::newFromURL( $url, "POST" );
1950                 }
1951         }
1952 }