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 ImportStreamSource $source
  51          * @param Config $config
  52          */
  53         function __construct( ImportStreamSource $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                 while ( $keepReading ) {
 537                         $tag = $this->reader->name;
 538                         $type = $this->reader->nodeType;
 539
 540                         if ( !Hooks::run( 'ImportHandleToplevelXMLTag', array( $this ) ) ) {
 541                                 // Do nothing
 542                         } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) {
 543                                 break;
 544                         } elseif ( $tag == 'siteinfo' ) {
 545                                 $this->handleSiteInfo();
 546                         } elseif ( $tag == 'page' ) {
 547                                 $this->handlePage();
 548                         } elseif ( $tag == 'logitem' ) {
 549                                 $this->handleLogItem();
 550                         } elseif ( $tag != '#text' ) {
 551                                 $this->warn( "Unhandled top-level XML tag $tag" );
 552
 553                                 $skip = true;
 554                         }
 555
 556                         if ( $skip ) {
 557                                 $keepReading = $this->reader->next();
 558                                 $skip = false;
 559                                 $this->debug( "Skip" );
 560                         } else {
 561                                 $keepReading = $this->reader->read();
 562                         }
 563                 }
 564
 565                 libxml_disable_entity_loader( $oldDisable );
 566                 return true;
 567         }
 568
 569         private function handleSiteInfo() {
 570                 $this->debug( "Enter site info handler." );
 571                 $siteInfo = array();
 572
 573                 // Fields that can just be stuffed in the siteInfo object
 574                 $normalFields = array( 'sitename', 'base', 'generator', 'case' );
 575
 576                 while ( $this->reader->read() ) {
 577                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 578                                         $this->reader->name == 'siteinfo' ) {
 579                                 break;
 580                         }
 581
 582                         $tag = $this->reader->name;
 583
 584                         if ( $tag == 'namespace' ) {
 585                                 $this->foreignNamespaces[ $this->nodeAttribute( 'key' ) ] =
 586                                         $this->nodeContents();
 587                         } elseif ( in_array( $tag, $normalFields ) ) {
 588                                 $siteInfo[$tag] = $this->nodeContents();
 589                         }
 590                 }
 591
 592                 $siteInfo['_namespaces'] = $this->foreignNamespaces;
 593                 $this->siteInfoCallback( $siteInfo );
 594         }
 595
 596         private function handleLogItem() {
 597                 $this->debug( "Enter log item handler." );
 598                 $logInfo = array();
 599
 600                 // Fields that can just be stuffed in the pageInfo object
 601                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 602                                         'logtitle', 'params' );
 603
 604                 while ( $this->reader->read() ) {
 605                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 606                                         $this->reader->name == 'logitem' ) {
 607                                 break;
 608                         }
 609
 610                         $tag = $this->reader->name;
 611
 612                         if ( !Hooks::run( 'ImportHandleLogItemXMLTag', array(
 613                                 $this, $logInfo
 614                         ) ) ) {
 615                                 // Do nothing
 616                         } elseif ( in_array( $tag, $normalFields ) ) {
 617                                 $logInfo[$tag] = $this->nodeContents();
 618                         } elseif ( $tag == 'contributor' ) {
 619                                 $logInfo['contributor'] = $this->handleContributor();
 620                         } elseif ( $tag != '#text' ) {
 621                                 $this->warn( "Unhandled log-item XML tag $tag" );
 622                         }
 623                 }
 624
 625                 $this->processLogItem( $logInfo );
 626         }
 627
 628         /**
 629          * @param array $logInfo
 630          * @return bool|mixed
 631          */
 632         private function processLogItem( $logInfo ) {
 633                 $revision = new WikiRevision( $this->config );
 634
 635                 $revision->setID( $logInfo['id'] );
 636                 $revision->setType( $logInfo['type'] );
 637                 $revision->setAction( $logInfo['action'] );
 638                 $revision->setTimestamp( $logInfo['timestamp'] );
 639                 $revision->setParams( $logInfo['params'] );
 640                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 641                 $revision->setNoUpdates( $this->mNoUpdates );
 642
 643                 if ( isset( $logInfo['comment'] ) ) {
 644                         $revision->setComment( $logInfo['comment'] );
 645                 }
 646
 647                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 648                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 649                 }
 650                 if ( isset( $logInfo['contributor']['username'] ) ) {
 651                         $revision->setUserName( $logInfo['contributor']['username'] );
 652                 }
 653
 654                 return $this->logItemCallback( $revision );
 655         }
 656
 657         private function handlePage() {
 658                 // Handle page data.
 659                 $this->debug( "Enter page handler." );
 660                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 661
 662                 // Fields that can just be stuffed in the pageInfo object
 663                 $normalFields = array( 'title', 'ns', 'id', 'redirect', 'restrictions' );
 664
 665                 $skip = false;
 666                 $badTitle = false;
 667
 668                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 669                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 670                                         $this->reader->name == 'page' ) {
 671                                 break;
 672                         }
 673
 674                         $skip = false;
 675
 676                         $tag = $this->reader->name;
 677
 678                         if ( $badTitle ) {
 679                                 // The title is invalid, bail out of this page
 680                                 $skip = true;
 681                         } elseif ( !Hooks::run( 'ImportHandlePageXMLTag', array( $this,
 682                                                 &$pageInfo ) ) ) {
 683                                 // Do nothing
 684                         } elseif ( in_array( $tag, $normalFields ) ) {
 685                                 // An XML snippet:
 686                                 // <page>
 687                                 //     <id>123</id>
 688                                 //     <title>Page</title>
 689                                 //     <redirect title="NewTitle"/>
 690                                 //     ...
 691                                 // Because the redirect tag is built differently, we need special handling for that case.
 692                                 if ( $tag == 'redirect' ) {
 693                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 694                                 } else {
 695                                         $pageInfo[$tag] = $this->nodeContents();
 696                                 }
 697                         } elseif ( $tag == 'revision' || $tag == 'upload' ) {
 698                                 if ( !isset( $title ) ) {
 699                                         $title = $this->processTitle( $pageInfo['title'],
 700                                                 isset( $pageInfo['ns'] ) ? $pageInfo['ns'] : null );
 701
 702                                         if ( !$title ) {
 703                                                 $badTitle = true;
 704                                                 $skip = true;
 705                                         }
 706
 707                                         $this->pageCallback( $title );
 708                                         list( $pageInfo['_title'], $foreignTitle ) = $title;
 709                                 }
 710
 711                                 if ( $title ) {
 712                                         if ( $tag == 'revision' ) {
 713                                                 $this->handleRevision( $pageInfo );
 714                                         } else {
 715                                                 $this->handleUpload( $pageInfo );
 716                                         }
 717                                 }
 718                         } elseif ( $tag != '#text' ) {
 719                                 $this->warn( "Unhandled page XML tag $tag" );
 720                                 $skip = true;
 721                         }
 722                 }
 723
 724                 $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
 725                                         $pageInfo['revisionCount'],
 726                                         $pageInfo['successfulRevisionCount'],
 727                                         $pageInfo );
 728         }
 729
 730         /**
 731          * @param array $pageInfo
 732          */
 733         private function handleRevision( &$pageInfo ) {
 734                 $this->debug( "Enter revision handler" );
 735                 $revisionInfo = array();
 736
 737                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' );
 738
 739                 $skip = false;
 740
 741                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 742                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 743                                         $this->reader->name == 'revision' ) {
 744                                 break;
 745                         }
 746
 747                         $tag = $this->reader->name;
 748
 749                         if ( !Hooks::run( 'ImportHandleRevisionXMLTag', array(
 750                                 $this, $pageInfo, $revisionInfo
 751                         ) ) ) {
 752                                 // Do nothing
 753                         } elseif ( in_array( $tag, $normalFields ) ) {
 754                                 $revisionInfo[$tag] = $this->nodeContents();
 755                         } elseif ( $tag == 'contributor' ) {
 756                                 $revisionInfo['contributor'] = $this->handleContributor();
 757                         } elseif ( $tag != '#text' ) {
 758                                 $this->warn( "Unhandled revision XML tag $tag" );
 759                                 $skip = true;
 760                         }
 761                 }
 762
 763                 $pageInfo['revisionCount']++;
 764                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 765                         $pageInfo['successfulRevisionCount']++;
 766                 }
 767         }
 768
 769         /**
 770          * @param array $pageInfo
 771          * @param array $revisionInfo
 772          * @return bool|mixed
 773          */
 774         private function processRevision( $pageInfo, $revisionInfo ) {
 775                 $revision = new WikiRevision( $this->config );
 776
 777                 if ( isset( $revisionInfo['id'] ) ) {
 778                         $revision->setID( $revisionInfo['id'] );
 779                 }
 780                 if ( isset( $revisionInfo['model'] ) ) {
 781                         $revision->setModel( $revisionInfo['model'] );
 782                 }
 783                 if ( isset( $revisionInfo['format'] ) ) {
 784                         $revision->setFormat( $revisionInfo['format'] );
 785                 }
 786                 $revision->setTitle( $pageInfo['_title'] );
 787
 788                 if ( isset( $revisionInfo['text'] ) ) {
 789                         $handler = $revision->getContentHandler();
 790                         $text = $handler->importTransform(
 791                                 $revisionInfo['text'],
 792                                 $revision->getFormat() );
 793
 794                         $revision->setText( $text );
 795                 }
 796                 if ( isset( $revisionInfo['timestamp'] ) ) {
 797                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 798                 } else {
 799                         $revision->setTimestamp( wfTimestampNow() );
 800                 }
 801
 802                 if ( isset( $revisionInfo['comment'] ) ) {
 803                         $revision->setComment( $revisionInfo['comment'] );
 804                 }
 805
 806                 if ( isset( $revisionInfo['minor'] ) ) {
 807                         $revision->setMinor( true );
 808                 }
 809                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 810                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 811                 }
 812                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 813                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 814                 }
 815                 $revision->setNoUpdates( $this->mNoUpdates );
 816
 817                 return $this->revisionCallback( $revision );
 818         }
 819
 820         /**
 821          * @param array $pageInfo
 822          * @return mixed
 823          */
 824         private function handleUpload( &$pageInfo ) {
 825                 $this->debug( "Enter upload handler" );
 826                 $uploadInfo = array();
 827
 828                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 829                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 830
 831                 $skip = false;
 832
 833                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 834                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 835                                         $this->reader->name == 'upload' ) {
 836                                 break;
 837                         }
 838
 839                         $tag = $this->reader->name;
 840
 841                         if ( !Hooks::run( 'ImportHandleUploadXMLTag', array(
 842                                 $this, $pageInfo
 843                         ) ) ) {
 844                                 // Do nothing
 845                         } elseif ( in_array( $tag, $normalFields ) ) {
 846                                 $uploadInfo[$tag] = $this->nodeContents();
 847                         } elseif ( $tag == 'contributor' ) {
 848                                 $uploadInfo['contributor'] = $this->handleContributor();
 849                         } elseif ( $tag == 'contents' ) {
 850                                 $contents = $this->nodeContents();
 851                                 $encoding = $this->reader->getAttribute( 'encoding' );
 852                                 if ( $encoding === 'base64' ) {
 853                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 854                                         $uploadInfo['isTempSrc'] = true;
 855                                 }
 856                         } elseif ( $tag != '#text' ) {
 857                                 $this->warn( "Unhandled upload XML tag $tag" );
 858                                 $skip = true;
 859                         }
 860                 }
 861
 862                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 863                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 864                         if ( file_exists( $path ) ) {
 865                                 $uploadInfo['fileSrc'] = $path;
 866                                 $uploadInfo['isTempSrc'] = false;
 867                         }
 868                 }
 869
 870                 if ( $this->mImportUploads ) {
 871                         return $this->processUpload( $pageInfo, $uploadInfo );
 872                 }
 873         }
 874
 875         /**
 876          * @param string $contents
 877          * @return string
 878          */
 879         private function dumpTemp( $contents ) {
 880                 $filename = tempnam( wfTempDir(), 'importupload' );
 881                 file_put_contents( $filename, $contents );
 882                 return $filename;
 883         }
 884
 885         /**
 886          * @param array $pageInfo
 887          * @param array $uploadInfo
 888          * @return mixed
 889          */
 890         private function processUpload( $pageInfo, $uploadInfo ) {
 891                 $revision = new WikiRevision( $this->config );
 892                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 893
 894                 $revision->setTitle( $pageInfo['_title'] );
 895                 $revision->setID( $pageInfo['id'] );
 896                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 897                 $revision->setText( $text );
 898                 $revision->setFilename( $uploadInfo['filename'] );
 899                 if ( isset( $uploadInfo['archivename'] ) ) {
 900                         $revision->setArchiveName( $uploadInfo['archivename'] );
 901                 }
 902                 $revision->setSrc( $uploadInfo['src'] );
 903                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 904                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 905                                 !empty( $uploadInfo['isTempSrc'] ) );
 906                 }
 907                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 908                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 909                 }
 910                 $revision->setSize( intval( $uploadInfo['size'] ) );
 911                 $revision->setComment( $uploadInfo['comment'] );
 912
 913                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 914                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 915                 }
 916                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 917                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 918                 }
 919                 $revision->setNoUpdates( $this->mNoUpdates );
 920
 921                 return call_user_func( $this->mUploadCallback, $revision );
 922         }
 923
 924         /**
 925          * @return array
 926          */
 927         private function handleContributor() {
 928                 $fields = array( 'id', 'ip', 'username' );
 929                 $info = array();
 930
 931                 while ( $this->reader->read() ) {
 932                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 933                                         $this->reader->name == 'contributor' ) {
 934                                 break;
 935                         }
 936
 937                         $tag = $this->reader->name;
 938
 939                         if ( in_array( $tag, $fields ) ) {
 940                                 $info[$tag] = $this->nodeContents();
 941                         }
 942                 }
 943
 944                 return $info;
 945         }
 946
 947         /**
 948          * @param string $text
 949          * @param string|null $ns
 950          * @return array|bool
 951          */
 952         private function processTitle( $text, $ns = null ) {
 953                 if ( is_null( $this->foreignNamespaces ) ) {
 954                         $foreignTitleFactory = new NaiveForeignTitleFactory();
 955                 } else {
 956                         $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
 957                                 $this->foreignNamespaces );
 958                 }
 959
 960                 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
 961                         intval( $ns ) );
 962
 963                 $title = $this->importTitleFactory->createTitleFromForeignTitle(
 964                         $foreignTitle );
 965
 966                 $commandLineMode = $this->config->get( 'CommandLineMode' );
 967                 if ( is_null( $title ) ) {
 968                         # Invalid page title? Ignore the page
 969                         $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
 970                         return false;
 971                 } elseif ( $title->isExternal() ) {
 972                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
 973                         return false;
 974                 } elseif ( !$title->canExist() ) {
 975                         $this->notice( 'import-error-special', $title->getPrefixedText() );
 976                         return false;
 977                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
 978                         # Do not import if the importing wiki user cannot edit this page
 979                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
 980                         return false;
 981                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
 982                         # Do not import if the importing wiki user cannot create this page
 983                         $this->notice( 'import-error-create', $title->getPrefixedText() );
 984                         return false;
 985                 }
 986
 987                 return array( $title, $foreignTitle );
 988         }
 989 }
 990
 991 /** This is a horrible hack used to keep source compatibility */
 992 class UploadSourceAdapter {
 993         /** @var array */
 994         public static $sourceRegistrations = array();
 995
 996         /** @var string */
 997         private $mSource;
 998
 999         /** @var string */
1000         private $mBuffer;
1001
1002         /** @var int */
1003         private $mPosition;
1004
1005         /**
1006          * @param ImportStreamSource $source
1007          * @return string
1008          */
1009         static function registerSource( ImportStreamSource $source ) {
1010                 $id = wfRandomString();
1011
1012                 self::$sourceRegistrations[$id] = $source;
1013
1014                 return $id;
1015         }
1016
1017         /**
1018          * @param string $path
1019          * @param string $mode
1020          * @param array $options
1021          * @param string $opened_path
1022          * @return bool
1023          */
1024         function stream_open( $path, $mode, $options, &$opened_path ) {
1025                 $url = parse_url( $path );
1026                 $id = $url['host'];
1027
1028                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
1029                         return false;
1030                 }
1031
1032                 $this->mSource = self::$sourceRegistrations[$id];
1033
1034                 return true;
1035         }
1036
1037         /**
1038          * @param int $count
1039          * @return string
1040          */
1041         function stream_read( $count ) {
1042                 $return = '';
1043                 $leave = false;
1044
1045                 while ( !$leave && !$this->mSource->atEnd() &&
1046                                 strlen( $this->mBuffer ) < $count ) {
1047                         $read = $this->mSource->readChunk();
1048
1049                         if ( !strlen( $read ) ) {
1050                                 $leave = true;
1051                         }
1052
1053                         $this->mBuffer .= $read;
1054                 }
1055
1056                 if ( strlen( $this->mBuffer ) ) {
1057                         $return = substr( $this->mBuffer, 0, $count );
1058                         $this->mBuffer = substr( $this->mBuffer, $count );
1059                 }
1060
1061                 $this->mPosition += strlen( $return );
1062
1063                 return $return;
1064         }
1065
1066         /**
1067          * @param string $data
1068          * @return bool
1069          */
1070         function stream_write( $data ) {
1071                 return false;
1072         }
1073
1074         /**
1075          * @return mixed
1076          */
1077         function stream_tell() {
1078                 return $this->mPosition;
1079         }
1080
1081         /**
1082          * @return bool
1083          */
1084         function stream_eof() {
1085                 return $this->mSource->atEnd();
1086         }
1087
1088         /**
1089          * @return array
1090          */
1091         function url_stat() {
1092                 $result = array();
1093
1094                 $result['dev'] = $result[0] = 0;
1095                 $result['ino'] = $result[1] = 0;
1096                 $result['mode'] = $result[2] = 0;
1097                 $result['nlink'] = $result[3] = 0;
1098                 $result['uid'] = $result[4] = 0;
1099                 $result['gid'] = $result[5] = 0;
1100                 $result['rdev'] = $result[6] = 0;
1101                 $result['size'] = $result[7] = 0;
1102                 $result['atime'] = $result[8] = 0;
1103                 $result['mtime'] = $result[9] = 0;
1104                 $result['ctime'] = $result[10] = 0;
1105                 $result['blksize'] = $result[11] = 0;
1106                 $result['blocks'] = $result[12] = 0;
1107
1108                 return $result;
1109         }
1110 }
1111
1112 /**
1113  * @todo document (e.g. one-sentence class description).
1114  * @ingroup SpecialPage
1115  */
1116 class WikiRevision {
1117         /** @todo Unused? */
1118         public $importer = null;
1119
1120         /** @var Title */
1121         public $title = null;
1122
1123         /** @var int */
1124         public $id = 0;
1125
1126         /** @var string */
1127         public $timestamp = "20010115000000";
1128
1129         /**
1130          * @var int
1131          * @todo Can't find any uses. Public, because that's suspicious. Get clarity. */
1132         public $user = 0;
1133
1134         /** @var string */
1135         public $user_text = "";
1136
1137         /** @var string */
1138         public $model = null;
1139
1140         /** @var string */
1141         public $format = null;
1142
1143         /** @var string */
1144         public $text = "";
1145
1146         /** @var int */
1147         protected $size;
1148
1149         /** @var Content */
1150         public $content = null;
1151
1152         /** @var ContentHandler */
1153         protected $contentHandler = null;
1154
1155         /** @var string */
1156         public $comment = "";
1157
1158         /** @var bool */
1159         public $minor = false;
1160
1161         /** @var string */
1162         public $type = "";
1163
1164         /** @var string */
1165         public $action = "";
1166
1167         /** @var string */
1168         public $params = "";
1169
1170         /** @var string */
1171         public $fileSrc = '';
1172
1173         /** @var bool|string */
1174         public $sha1base36 = false;
1175
1176         /**
1177          * @var bool
1178          * @todo Unused?
1179          */
1180         public $isTemp = false;
1181
1182         /** @var string */
1183         public $archiveName = '';
1184
1185         protected $filename;
1186
1187         /** @var mixed */
1188         protected $src;
1189
1190         /** @todo Unused? */
1191         public $fileIsTemp;
1192
1193         /** @var bool */
1194         private $mNoUpdates = false;
1195
1196         /** @var Config $config */
1197         private $config;
1198
1199         public function __construct( Config $config ) {
1200                 $this->config = $config;
1201         }
1202
1203         /**
1204          * @param Title $title
1205          * @throws MWException
1206          */
1207         function setTitle( $title ) {
1208                 if ( is_object( $title ) ) {
1209                         $this->title = $title;
1210                 } elseif ( is_null( $title ) ) {
1211                         throw new MWException( "WikiRevision given a null title in import. "
1212                                 . "You may need to adjust \$wgLegalTitleChars." );
1213                 } else {
1214                         throw new MWException( "WikiRevision given non-object title in import." );
1215                 }
1216         }
1217
1218         /**
1219          * @param int $id
1220          */
1221         function setID( $id ) {
1222                 $this->id = $id;
1223         }
1224
1225         /**
1226          * @param string $ts
1227          */
1228         function setTimestamp( $ts ) {
1229                 # 2003-08-05T18:30:02Z
1230                 $this->timestamp = wfTimestamp( TS_MW, $ts );
1231         }
1232
1233         /**
1234          * @param string $user
1235          */
1236         function setUsername( $user ) {
1237                 $this->user_text = $user;
1238         }
1239
1240         /**
1241          * @param string $ip
1242          */
1243         function setUserIP( $ip ) {
1244                 $this->user_text = $ip;
1245         }
1246
1247         /**
1248          * @param string $model
1249          */
1250         function setModel( $model ) {
1251                 $this->model = $model;
1252         }
1253
1254         /**
1255          * @param string $format
1256          */
1257         function setFormat( $format ) {
1258                 $this->format = $format;
1259         }
1260
1261         /**
1262          * @param string $text
1263          */
1264         function setText( $text ) {
1265                 $this->text = $text;
1266         }
1267
1268         /**
1269          * @param string $text
1270          */
1271         function setComment( $text ) {
1272                 $this->comment = $text;
1273         }
1274
1275         /**
1276          * @param bool $minor
1277          */
1278         function setMinor( $minor ) {
1279                 $this->minor = (bool)$minor;
1280         }
1281
1282         /**
1283          * @param mixed $src
1284          */
1285         function setSrc( $src ) {
1286                 $this->src = $src;
1287         }
1288
1289         /**
1290          * @param string $src
1291          * @param bool $isTemp
1292          */
1293         function setFileSrc( $src, $isTemp ) {
1294                 $this->fileSrc = $src;
1295                 $this->fileIsTemp = $isTemp;
1296         }
1297
1298         /**
1299          * @param string $sha1base36
1300          */
1301         function setSha1Base36( $sha1base36 ) {
1302                 $this->sha1base36 = $sha1base36;
1303         }
1304
1305         /**
1306          * @param string $filename
1307          */
1308         function setFilename( $filename ) {
1309                 $this->filename = $filename;
1310         }
1311
1312         /**
1313          * @param string $archiveName
1314          */
1315         function setArchiveName( $archiveName ) {
1316                 $this->archiveName = $archiveName;
1317         }
1318
1319         /**
1320          * @param int $size
1321          */
1322         function setSize( $size ) {
1323                 $this->size = intval( $size );
1324         }
1325
1326         /**
1327          * @param string $type
1328          */
1329         function setType( $type ) {
1330                 $this->type = $type;
1331         }
1332
1333         /**
1334          * @param string $action
1335          */
1336         function setAction( $action ) {
1337                 $this->action = $action;
1338         }
1339
1340         /**
1341          * @param array $params
1342          */
1343         function setParams( $params ) {
1344                 $this->params = $params;
1345         }
1346
1347         /**
1348          * @param bool $noupdates
1349          */
1350         public function setNoUpdates( $noupdates ) {
1351                 $this->mNoUpdates = $noupdates;
1352         }
1353
1354         /**
1355          * @return Title
1356          */
1357         function getTitle() {
1358                 return $this->title;
1359         }
1360
1361         /**
1362          * @return int
1363          */
1364         function getID() {
1365                 return $this->id;
1366         }
1367
1368         /**
1369          * @return string
1370          */
1371         function getTimestamp() {
1372                 return $this->timestamp;
1373         }
1374
1375         /**
1376          * @return string
1377          */
1378         function getUser() {
1379                 return $this->user_text;
1380         }
1381
1382         /**
1383          * @return string
1384          *
1385          * @deprecated Since 1.21, use getContent() instead.
1386          */
1387         function getText() {
1388                 ContentHandler::deprecated( __METHOD__, '1.21' );
1389
1390                 return $this->text;
1391         }
1392
1393         /**
1394          * @return ContentHandler
1395          */
1396         function getContentHandler() {
1397                 if ( is_null( $this->contentHandler ) ) {
1398                         $this->contentHandler = ContentHandler::getForModelID( $this->getModel() );
1399                 }
1400
1401                 return $this->contentHandler;
1402         }
1403
1404         /**
1405          * @return Content
1406          */
1407         function getContent() {
1408                 if ( is_null( $this->content ) ) {
1409                         $handler = $this->getContentHandler();
1410                         $this->content = $handler->unserializeContent( $this->text, $this->getFormat() );
1411                 }
1412
1413                 return $this->content;
1414         }
1415
1416         /**
1417          * @return string
1418          */
1419         function getModel() {
1420                 if ( is_null( $this->model ) ) {
1421                         $this->model = $this->getTitle()->getContentModel();
1422                 }
1423
1424                 return $this->model;
1425         }
1426
1427         /**
1428          * @return string
1429          */
1430         function getFormat() {
1431                 if ( is_null( $this->format ) ) {
1432                         $this->format = $this->getContentHandler()->getDefaultFormat();
1433                 }
1434
1435                 return $this->format;
1436         }
1437
1438         /**
1439          * @return string
1440          */
1441         function getComment() {
1442                 return $this->comment;
1443         }
1444
1445         /**
1446          * @return bool
1447          */
1448         function getMinor() {
1449                 return $this->minor;
1450         }
1451
1452         /**
1453          * @return mixed
1454          */
1455         function getSrc() {
1456                 return $this->src;
1457         }
1458
1459         /**
1460          * @return bool|string
1461          */
1462         function getSha1() {
1463                 if ( $this->sha1base36 ) {
1464                         return wfBaseConvert( $this->sha1base36, 36, 16 );
1465                 }
1466                 return false;
1467         }
1468
1469         /**
1470          * @return string
1471          */
1472         function getFileSrc() {
1473                 return $this->fileSrc;
1474         }
1475
1476         /**
1477          * @return bool
1478          */
1479         function isTempSrc() {
1480                 return $this->isTemp;
1481         }
1482
1483         /**
1484          * @return mixed
1485          */
1486         function getFilename() {
1487                 return $this->filename;
1488         }
1489
1490         /**
1491          * @return string
1492          */
1493         function getArchiveName() {
1494                 return $this->archiveName;
1495         }
1496
1497         /**
1498          * @return mixed
1499          */
1500         function getSize() {
1501                 return $this->size;
1502         }
1503
1504         /**
1505          * @return string
1506          */
1507         function getType() {
1508                 return $this->type;
1509         }
1510
1511         /**
1512          * @return string
1513          */
1514         function getAction() {
1515                 return $this->action;
1516         }
1517
1518         /**
1519          * @return string
1520          */
1521         function getParams() {
1522                 return $this->params;
1523         }
1524
1525         /**
1526          * @return bool
1527          */
1528         function importOldRevision() {
1529                 $dbw = wfGetDB( DB_MASTER );
1530
1531                 # Sneak a single revision into place
1532                 $user = User::newFromName( $this->getUser() );
1533                 if ( $user ) {
1534                         $userId = intval( $user->getId() );
1535                         $userText = $user->getName();
1536                         $userObj = $user;
1537                 } else {
1538                         $userId = 0;
1539                         $userText = $this->getUser();
1540                         $userObj = new User;
1541                 }
1542
1543                 // avoid memory leak...?
1544                 $linkCache = LinkCache::singleton();
1545                 $linkCache->clear();
1546
1547                 $page = WikiPage::factory( $this->title );
1548                 $page->loadPageData( 'fromdbmaster' );
1549                 if ( !$page->exists() ) {
1550                         # must create the page...
1551                         $pageId = $page->insertOn( $dbw );
1552                         $created = true;
1553                         $oldcountable = null;
1554                 } else {
1555                         $pageId = $page->getId();
1556                         $created = false;
1557
1558                         $prior = $dbw->selectField( 'revision', '1',
1559                                 array( 'rev_page' => $pageId,
1560                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1561                                         'rev_user_text' => $userText,
1562                                         'rev_comment' => $this->getComment() ),
1563                                 __METHOD__
1564                         );
1565                         if ( $prior ) {
1566                                 // @todo FIXME: This could fail slightly for multiple matches :P
1567                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1568                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1569                                 return false;
1570                         }
1571                 }
1572
1573                 # @todo FIXME: Use original rev_id optionally (better for backups)
1574                 # Insert the row
1575                 $revision = new Revision( array(
1576                         'title' => $this->title,
1577                         'page' => $pageId,
1578                         'content_model' => $this->getModel(),
1579                         'content_format' => $this->getFormat(),
1580                         //XXX: just set 'content' => $this->getContent()?
1581                         'text' => $this->getContent()->serialize( $this->getFormat() ),
1582                         'comment' => $this->getComment(),
1583                         'user' => $userId,
1584                         'user_text' => $userText,
1585                         'timestamp' => $this->timestamp,
1586                         'minor_edit' => $this->minor,
1587                         ) );
1588                 $revision->insertOn( $dbw );
1589                 $changed = $page->updateIfNewerOn( $dbw, $revision );
1590
1591                 if ( $changed !== false && !$this->mNoUpdates ) {
1592                         wfDebug( __METHOD__ . ": running updates\n" );
1593                         // countable/oldcountable stuff is handled in WikiImporter::finishImportPage
1594                         $page->doEditUpdates(
1595                                 $revision,
1596                                 $userObj,
1597                                 array( 'created' => $created, 'oldcountable' => 'no-change' )
1598                         );
1599                 }
1600
1601                 return true;
1602         }
1603
1604         function importLogItem() {
1605                 $dbw = wfGetDB( DB_MASTER );
1606                 # @todo FIXME: This will not record autoblocks
1607                 if ( !$this->getTitle() ) {
1608                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1609                                 $this->timestamp . "\n" );
1610                         return;
1611                 }
1612                 # Check if it exists already
1613                 // @todo FIXME: Use original log ID (better for backups)
1614                 $prior = $dbw->selectField( 'logging', '1',
1615                         array( 'log_type' => $this->getType(),
1616                                 'log_action' => $this->getAction(),
1617                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1618                                 'log_namespace' => $this->getTitle()->getNamespace(),
1619                                 'log_title' => $this->getTitle()->getDBkey(),
1620                                 'log_comment' => $this->getComment(),
1621                                 #'log_user_text' => $this->user_text,
1622                                 'log_params' => $this->params ),
1623                         __METHOD__
1624                 );
1625                 // @todo FIXME: This could fail slightly for multiple matches :P
1626                 if ( $prior ) {
1627                         wfDebug( __METHOD__
1628                                 . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp "
1629                                 . $this->timestamp . "\n" );
1630                         return;
1631                 }
1632                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1633                 $data = array(
1634                         'log_id' => $log_id,
1635                         'log_type' => $this->type,
1636                         'log_action' => $this->action,
1637                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1638                         'log_user' => User::idFromName( $this->user_text ),
1639                         #'log_user_text' => $this->user_text,
1640                         'log_namespace' => $this->getTitle()->getNamespace(),
1641                         'log_title' => $this->getTitle()->getDBkey(),
1642                         'log_comment' => $this->getComment(),
1643                         'log_params' => $this->params
1644                 );
1645                 $dbw->insert( 'logging', $data, __METHOD__ );
1646         }
1647
1648         /**
1649          * @return bool
1650          */
1651         function importUpload() {
1652                 # Construct a file
1653                 $archiveName = $this->getArchiveName();
1654                 if ( $archiveName ) {
1655                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1656                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1657                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1658                 } else {
1659                         $file = wfLocalFile( $this->getTitle() );
1660                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1661                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1662                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1663                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1664                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1665                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1666                         }
1667                 }
1668                 if ( !$file ) {
1669                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1670                         return false;
1671                 }
1672
1673                 # Get the file source or download if necessary
1674                 $source = $this->getFileSrc();
1675                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1676                 if ( !$source ) {
1677                         $source = $this->downloadSource();
1678                         $flags |= File::DELETE_SOURCE;
1679                 }
1680                 if ( !$source ) {
1681                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1682                         return false;
1683                 }
1684                 $sha1 = $this->getSha1();
1685                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1686                         if ( $flags & File::DELETE_SOURCE ) {
1687                                 # Broken file; delete it if it is a temporary file
1688                                 unlink( $source );
1689                         }
1690                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1691                         return false;
1692                 }
1693
1694                 $user = User::newFromName( $this->user_text );
1695
1696                 # Do the actual upload
1697                 if ( $archiveName ) {
1698                         $status = $file->uploadOld( $source, $archiveName,
1699                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1700                 } else {
1701                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1702                                 $flags, false, $this->getTimestamp(), $user );
1703                 }
1704
1705                 if ( $status->isGood() ) {
1706                         wfDebug( __METHOD__ . ": Successful\n" );
1707                         return true;
1708                 } else {
1709                         wfDebug( __METHOD__ . ': failed: ' . $status->getXml() . "\n" );
1710                         return false;
1711                 }
1712         }
1713
1714         /**
1715          * @return bool|string
1716          */
1717         function downloadSource() {
1718                 if ( !$this->config->get( 'EnableUploads' ) ) {
1719                         return false;
1720                 }
1721
1722                 $tempo = tempnam( wfTempDir(), 'download' );
1723                 $f = fopen( $tempo, 'wb' );
1724                 if ( !$f ) {
1725                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1726                         return false;
1727                 }
1728
1729                 // @todo FIXME!
1730                 $src = $this->getSrc();
1731                 $data = Http::get( $src );
1732                 if ( !$data ) {
1733                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1734                         fclose( $f );
1735                         unlink( $tempo );
1736                         return false;
1737                 }
1738
1739                 fwrite( $f, $data );
1740                 fclose( $f );
1741
1742                 return $tempo;
1743         }
1744
1745 }
1746
1747 /**
1748  * Used for importing XML dumps where the content of the dump is in a string.
1749  * This class is ineffecient, and should only be used for small dumps.
1750  * For larger dumps, ImportStreamSource should be used instead.
1751  *
1752  * @ingroup SpecialPage
1753  */
1754 class ImportStringSource {
1755         function __construct( $string ) {
1756                 $this->mString = $string;
1757                 $this->mRead = false;
1758         }
1759
1760         /**
1761          * @return bool
1762          */
1763         function atEnd() {
1764                 return $this->mRead;
1765         }
1766
1767         /**
1768          * @return bool|string
1769          */
1770         function readChunk() {
1771                 if ( $this->atEnd() ) {
1772                         return false;
1773                 }
1774                 $this->mRead = true;
1775                 return $this->mString;
1776         }
1777 }
1778
1779 /**
1780  * Imports a XML dump from a file (either from file upload, files on disk, or HTTP)
1781  * @ingroup SpecialPage
1782  */
1783 class ImportStreamSource {
1784         function __construct( $handle ) {
1785                 $this->mHandle = $handle;
1786         }
1787
1788         /**
1789          * @return bool
1790          */
1791         function atEnd() {
1792                 return feof( $this->mHandle );
1793         }
1794
1795         /**
1796          * @return string
1797          */
1798         function readChunk() {
1799                 return fread( $this->mHandle, 32768 );
1800         }
1801
1802         /**
1803          * @param string $filename
1804          * @return Status
1805          */
1806         static function newFromFile( $filename ) {
1807                 wfSuppressWarnings();
1808                 $file = fopen( $filename, 'rt' );
1809                 wfRestoreWarnings();
1810                 if ( !$file ) {
1811                         return Status::newFatal( "importcantopen" );
1812                 }
1813                 return Status::newGood( new ImportStreamSource( $file ) );
1814         }
1815
1816         /**
1817          * @param string $fieldname
1818          * @return Status
1819          */
1820         static function newFromUpload( $fieldname = "xmlimport" ) {
1821                 $upload =& $_FILES[$fieldname];
1822
1823                 if ( $upload === null || !$upload['name'] ) {
1824                         return Status::newFatal( 'importnofile' );
1825                 }
1826                 if ( !empty( $upload['error'] ) ) {
1827                         switch ( $upload['error'] ) {
1828                                 case 1:
1829                                         # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1830                                         return Status::newFatal( 'importuploaderrorsize' );
1831                                 case 2:
1832                                         # The uploaded file exceeds the MAX_FILE_SIZE directive that
1833                                         # was specified in the HTML form.
1834                                         return Status::newFatal( 'importuploaderrorsize' );
1835                                 case 3:
1836                                         # The uploaded file was only partially uploaded
1837                                         return Status::newFatal( 'importuploaderrorpartial' );
1838                                 case 6:
1839                                         # Missing a temporary folder.
1840                                         return Status::newFatal( 'importuploaderrortemp' );
1841                                 # case else: # Currently impossible
1842                         }
1843
1844                 }
1845                 $fname = $upload['tmp_name'];
1846                 if ( is_uploaded_file( $fname ) ) {
1847                         return ImportStreamSource::newFromFile( $fname );
1848                 } else {
1849                         return Status::newFatal( 'importnofile' );
1850                 }
1851         }
1852
1853         /**
1854          * @param string $url
1855          * @param string $method
1856          * @return Status
1857          */
1858         static function newFromURL( $url, $method = 'GET' ) {
1859                 wfDebug( __METHOD__ . ": opening $url\n" );
1860                 # Use the standard HTTP fetch function; it times out
1861                 # quicker and sorts out user-agent problems which might
1862                 # otherwise prevent importing from large sites, such
1863                 # as the Wikimedia cluster, etc.
1864                 $data = Http::request( $method, $url, array( 'followRedirects' => true ) );
1865                 if ( $data !== false ) {
1866                         $file = tmpfile();
1867                         fwrite( $file, $data );
1868                         fflush( $file );
1869                         fseek( $file, 0 );
1870                         return Status::newGood( new ImportStreamSource( $file ) );
1871                 } else {
1872                         return Status::newFatal( 'importcantopen' );
1873                 }
1874         }
1875
1876         /**
1877          * @param string $interwiki
1878          * @param string $page
1879          * @param bool $history
1880          * @param bool $templates
1881          * @param int $pageLinkDepth
1882          * @return Status
1883          */
1884         public static function newFromInterwiki( $interwiki, $page, $history = false,
1885                 $templates = false, $pageLinkDepth = 0
1886         ) {
1887                 if ( $page == '' ) {
1888                         return Status::newFatal( 'import-noarticle' );
1889                 }
1890                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1891                 if ( is_null( $link ) || !$link->isExternal() ) {
1892                         return Status::newFatal( 'importbadinterwiki' );
1893                 } else {
1894                         $params = array();
1895                         if ( $history ) {
1896                                 $params['history'] = 1;
1897                         }
1898                         if ( $templates ) {
1899                                 $params['templates'] = 1;
1900                         }
1901                         if ( $pageLinkDepth ) {
1902                                 $params['pagelink-depth'] = $pageLinkDepth;
1903                         }
1904                         $url = $link->getFullURL( $params );
1905                         # For interwikis, use POST to avoid redirects.
1906                         return ImportStreamSource::newFromURL( $url, "POST" );
1907                 }
1908         }
1909 }