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