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, $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          * @throws Exception
  53          */
  54         function __construct( ImportSource $source, Config $config = null ) {
  55                 if ( !class_exists( 'XMLReader' ) ) {
  56                         throw new Exception( 'Import requires PHP to have been compiled with libxml support' );
  57                 }
  58
  59                 $this->reader = new XMLReader();
  60                 if ( !$config ) {
  61                         wfDeprecated( __METHOD__ . ' without a Config instance', '1.25' );
  62                         $config = ConfigFactory::getDefaultInstance()->makeConfig( 'main' );
  63                 }
  64                 $this->config = $config;
  65
  66                 if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) {
  67                         stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  68                 }
  69                 $id = UploadSourceAdapter::registerSource( $source );
  70
  71                 // Enable the entity loader, as it is needed for loading external URLs via
  72                 // XMLReader::open (T86036)
  73                 $oldDisable = libxml_disable_entity_loader( false );
  74                 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
  75                         $status = $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  76                 } else {
  77                         $status = $this->reader->open( "uploadsource://$id" );
  78                 }
  79                 if ( !$status ) {
  80                         $error = libxml_get_last_error();
  81                         libxml_disable_entity_loader( $oldDisable );
  82                         throw new MWException( 'Encountered an internal error while initializing WikiImporter object: ' .
  83                                 $error->message );
  84                 }
  85                 libxml_disable_entity_loader( $oldDisable );
  86
  87                 // Default callbacks
  88                 $this->setPageCallback( array( $this, 'beforeImportPage' ) );
  89                 $this->setRevisionCallback( array( $this, "importRevision" ) );
  90                 $this->setUploadCallback( array( $this, 'importUpload' ) );
  91                 $this->setLogItemCallback( array( $this, 'importLogItem' ) );
  92                 $this->setPageOutCallback( array( $this, 'finishImportPage' ) );
  93
  94                 $this->importTitleFactory = new NaiveImportTitleFactory();
  95         }
  96
  97         /**
  98          * @return null|XMLReader
  99          */
 100         public function getReader() {
 101                 return $this->reader;
 102         }
 103
 104         public function throwXmlError( $err ) {
 105                 $this->debug( "FAILURE: $err" );
 106                 wfDebug( "WikiImporter XML error: $err\n" );
 107         }
 108
 109         public function debug( $data ) {
 110                 if ( $this->mDebug ) {
 111                         wfDebug( "IMPORT: $data\n" );
 112                 }
 113         }
 114
 115         public function warn( $data ) {
 116                 wfDebug( "IMPORT: $data\n" );
 117         }
 118
 119         public function notice( $msg /*, $param, ...*/ ) {
 120                 $params = func_get_args();
 121                 array_shift( $params );
 122
 123                 if ( is_callable( $this->mNoticeCallback ) ) {
 124                         call_user_func( $this->mNoticeCallback, $msg, $params );
 125                 } else { # No ImportReporter -> CLI
 126                         echo wfMessage( $msg, $params )->text() . "\n";
 127                 }
 128         }
 129
 130         /**
 131          * Set debug mode...
 132          * @param bool $debug
 133          */
 134         function setDebug( $debug ) {
 135                 $this->mDebug = $debug;
 136         }
 137
 138         /**
 139          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
 140          * @param bool $noupdates
 141          */
 142         function setNoUpdates( $noupdates ) {
 143                 $this->mNoUpdates = $noupdates;
 144         }
 145
 146         /**
 147          * Set a callback that displays notice messages
 148          *
 149          * @param callable $callback
 150          * @return callable
 151          */
 152         public function setNoticeCallback( $callback ) {
 153                 return wfSetVar( $this->mNoticeCallback, $callback );
 154         }
 155
 156         /**
 157          * Sets the action to perform as each new page in the stream is reached.
 158          * @param callable $callback
 159          * @return callable
 160          */
 161         public function setPageCallback( $callback ) {
 162                 $previous = $this->mPageCallback;
 163                 $this->mPageCallback = $callback;
 164                 return $previous;
 165         }
 166
 167         /**
 168          * Sets the action to perform as each page in the stream is completed.
 169          * Callback accepts the page title (as a Title object), a second object
 170          * with the original title form (in case it's been overridden into a
 171          * local namespace), and a count of revisions.
 172          *
 173          * @param callable $callback
 174          * @return callable
 175          */
 176         public function setPageOutCallback( $callback ) {
 177                 $previous = $this->mPageOutCallback;
 178                 $this->mPageOutCallback = $callback;
 179                 return $previous;
 180         }
 181
 182         /**
 183          * Sets the action to perform as each page revision is reached.
 184          * @param callable $callback
 185          * @return callable
 186          */
 187         public function setRevisionCallback( $callback ) {
 188                 $previous = $this->mRevisionCallback;
 189                 $this->mRevisionCallback = $callback;
 190                 return $previous;
 191         }
 192
 193         /**
 194          * Sets the action to perform as each file upload version is reached.
 195          * @param callable $callback
 196          * @return callable
 197          */
 198         public function setUploadCallback( $callback ) {
 199                 $previous = $this->mUploadCallback;
 200                 $this->mUploadCallback = $callback;
 201                 return $previous;
 202         }
 203
 204         /**
 205          * Sets the action to perform as each log item reached.
 206          * @param callable $callback
 207          * @return callable
 208          */
 209         public function setLogItemCallback( $callback ) {
 210                 $previous = $this->mLogItemCallback;
 211                 $this->mLogItemCallback = $callback;
 212                 return $previous;
 213         }
 214
 215         /**
 216          * Sets the action to perform when site info is encountered
 217          * @param callable $callback
 218          * @return callable
 219          */
 220         public function setSiteInfoCallback( $callback ) {
 221                 $previous = $this->mSiteInfoCallback;
 222                 $this->mSiteInfoCallback = $callback;
 223                 return $previous;
 224         }
 225
 226         /**
 227          * Sets the factory object to use to convert ForeignTitle objects into local
 228          * Title objects
 229          * @param ImportTitleFactory $factory
 230          */
 231         public function setImportTitleFactory( $factory ) {
 232                 $this->importTitleFactory = $factory;
 233         }
 234
 235         /**
 236          * Set a target namespace to override the defaults
 237          * @param null|int $namespace
 238          * @return bool
 239          */
 240         public function setTargetNamespace( $namespace ) {
 241                 if ( is_null( $namespace ) ) {
 242                         // Don't override namespaces
 243                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 244                         return true;
 245                 } elseif (
 246                         $namespace >= 0 &&
 247                         MWNamespace::exists( intval( $namespace ) )
 248                 ) {
 249                         $namespace = intval( $namespace );
 250                         $this->setImportTitleFactory( new NamespaceImportTitleFactory( $namespace ) );
 251                         return true;
 252                 } else {
 253                         return false;
 254                 }
 255         }
 256
 257         /**
 258          * Set a target root page under which all pages are imported
 259          * @param null|string $rootpage
 260          * @return Status
 261          */
 262         public function setTargetRootPage( $rootpage ) {
 263                 $status = Status::newGood();
 264                 if ( is_null( $rootpage ) ) {
 265                         // No rootpage
 266                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 267                 } elseif ( $rootpage !== '' ) {
 268                         $rootpage = rtrim( $rootpage, '/' ); // avoid double slashes
 269                         $title = Title::newFromText( $rootpage );
 270
 271                         if ( !$title || $title->isExternal() ) {
 272                                 $status->fatal( 'import-rootpage-invalid' );
 273                         } else {
 274                                 if ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 275                                         global $wgContLang;
 276
 277                                         $displayNSText = $title->getNamespace() == NS_MAIN
 278                                                 ? wfMessage( 'blanknamespace' )->text()
 279                                                 : $wgContLang->getNsText( $title->getNamespace() );
 280                                         $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 281                                 } else {
 282                                         // set namespace to 'all', so the namespace check in processTitle() can pass
 283                                         $this->setTargetNamespace( null );
 284                                         $this->setImportTitleFactory( new SubpageImportTitleFactory( $title ) );
 285                                 }
 286                         }
 287                 }
 288                 return $status;
 289         }
 290
 291         /**
 292          * @param string $dir
 293          */
 294         public function setImageBasePath( $dir ) {
 295                 $this->mImageBasePath = $dir;
 296         }
 297
 298         /**
 299          * @param bool $import
 300          */
 301         public function setImportUploads( $import ) {
 302                 $this->mImportUploads = $import;
 303         }
 304
 305         /**
 306          * Default per-page callback. Sets up some things related to site statistics
 307          * @param array $titleAndForeignTitle Two-element array, with Title object at
 308          * index 0 and ForeignTitle object at index 1
 309          * @return bool
 310          */
 311         public function beforeImportPage( $titleAndForeignTitle ) {
 312                 $title = $titleAndForeignTitle[0];
 313                 $page = WikiPage::factory( $title );
 314                 $this->countableCache['title_' . $title->getPrefixedText()] = $page->isCountable();
 315                 return true;
 316         }
 317
 318         /**
 319          * Default per-revision callback, performs the import.
 320          * @param WikiRevision $revision
 321          * @return bool
 322          */
 323         public function importRevision( $revision ) {
 324                 if ( !$revision->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
 325                         $this->notice( 'import-error-bad-location',
 326                                 $revision->getTitle()->getPrefixedText(),
 327                                 $revision->getID(),
 328                                 $revision->getModel(),
 329                                 $revision->getFormat() );
 330
 331                         return false;
 332                 }
 333
 334                 try {
 335                         $dbw = wfGetDB( DB_MASTER );
 336                         return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
 337                 } catch ( MWContentSerializationException $ex ) {
 338                         $this->notice( 'import-error-unserialize',
 339                                 $revision->getTitle()->getPrefixedText(),
 340                                 $revision->getID(),
 341                                 $revision->getModel(),
 342                                 $revision->getFormat() );
 343                 }
 344
 345                 return false;
 346         }
 347
 348         /**
 349          * Default per-revision callback, performs the import.
 350          * @param WikiRevision $revision
 351          * @return bool
 352          */
 353         public function importLogItem( $revision ) {
 354                 $dbw = wfGetDB( DB_MASTER );
 355                 return $dbw->deadlockLoop( array( $revision, 'importLogItem' ) );
 356         }
 357
 358         /**
 359          * Dummy for now...
 360          * @param WikiRevision $revision
 361          * @return bool
 362          */
 363         public function importUpload( $revision ) {
 364                 $dbw = wfGetDB( DB_MASTER );
 365                 return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
 366         }
 367
 368         /**
 369          * Mostly for hook use
 370          * @param Title $title
 371          * @param ForeignTitle $foreignTitle
 372          * @param int $revCount
 373          * @param int $sRevCount
 374          * @param array $pageInfo
 375          * @return bool
 376          */
 377         public function finishImportPage( $title, $foreignTitle, $revCount,
 378                         $sRevCount, $pageInfo ) {
 379
 380                 // Update article count statistics (T42009)
 381                 // The normal counting logic in WikiPage->doEditUpdates() is designed for
 382                 // one-revision-at-a-time editing, not bulk imports. In this situation it
 383                 // suffers from issues of slave lag. We let WikiPage handle the total page
 384                 // and revision count, and we implement our own custom logic for the
 385                 // article (content page) count.
 386                 $page = WikiPage::factory( $title );
 387                 $page->loadPageData( 'fromdbmaster' );
 388                 $content = $page->getContent();
 389                 if ( $content === null ) {
 390                         wfDebug( __METHOD__ . ': Skipping article count adjustment for ' . $title .
 391                                 ' because WikiPage::getContent() returned null' );
 392                 } else {
 393                         $editInfo = $page->prepareContentForEdit( $content );
 394                         $countKey = 'title_' . $title->getPrefixedText();
 395                         $countable = $page->isCountable( $editInfo );
 396                         if ( array_key_exists( $countKey, $this->countableCache ) &&
 397                                 $countable != $this->countableCache[$countKey] ) {
 398                                 DeferredUpdates::addUpdate( SiteStatsUpdate::factory( array(
 399                                         'articles' => ( (int)$countable - (int)$this->countableCache[$countKey] )
 400                                 ) ) );
 401                         }
 402                 }
 403
 404                 $args = func_get_args();
 405                 return Hooks::run( 'AfterImportPage', $args );
 406         }
 407
 408         /**
 409          * Alternate per-revision callback, for debugging.
 410          * @param WikiRevision $revision
 411          */
 412         public function debugRevisionHandler( &$revision ) {
 413                 $this->debug( "Got revision:" );
 414                 if ( is_object( $revision->title ) ) {
 415                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 416                 } else {
 417                         $this->debug( "-- Title: <invalid>" );
 418                 }
 419                 $this->debug( "-- User: " . $revision->user_text );
 420                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 421                 $this->debug( "-- Comment: " . $revision->comment );
 422                 $this->debug( "-- Text: " . $revision->text );
 423         }
 424
 425         /**
 426          * Notify the callback function of site info
 427          * @param array $siteInfo
 428          * @return bool|mixed
 429          */
 430         private function siteInfoCallback( $siteInfo ) {
 431                 if ( isset( $this->mSiteInfoCallback ) ) {
 432                         return call_user_func_array( $this->mSiteInfoCallback,
 433                                         array( $siteInfo, $this ) );
 434                 } else {
 435                         return false;
 436                 }
 437         }
 438
 439         /**
 440          * Notify the callback function when a new "<page>" is reached.
 441          * @param Title $title
 442          */
 443         function pageCallback( $title ) {
 444                 if ( isset( $this->mPageCallback ) ) {
 445                         call_user_func( $this->mPageCallback, $title );
 446                 }
 447         }
 448
 449         /**
 450          * Notify the callback function when a "</page>" is closed.
 451          * @param Title $title
 452          * @param ForeignTitle $foreignTitle
 453          * @param int $revCount
 454          * @param int $sucCount Number of revisions for which callback returned true
 455          * @param array $pageInfo Associative array of page information
 456          */
 457         private function pageOutCallback( $title, $foreignTitle, $revCount,
 458                         $sucCount, $pageInfo ) {
 459                 if ( isset( $this->mPageOutCallback ) ) {
 460                         $args = func_get_args();
 461                         call_user_func_array( $this->mPageOutCallback, $args );
 462                 }
 463         }
 464
 465         /**
 466          * Notify the callback function of a revision
 467          * @param WikiRevision $revision
 468          * @return bool|mixed
 469          */
 470         private function revisionCallback( $revision ) {
 471                 if ( isset( $this->mRevisionCallback ) ) {
 472                         return call_user_func_array( $this->mRevisionCallback,
 473                                         array( $revision, $this ) );
 474                 } else {
 475                         return false;
 476                 }
 477         }
 478
 479         /**
 480          * Notify the callback function of a new log item
 481          * @param WikiRevision $revision
 482          * @return bool|mixed
 483          */
 484         private function logItemCallback( $revision ) {
 485                 if ( isset( $this->mLogItemCallback ) ) {
 486                         return call_user_func_array( $this->mLogItemCallback,
 487                                         array( $revision, $this ) );
 488                 } else {
 489                         return false;
 490                 }
 491         }
 492
 493         /**
 494          * Retrieves the contents of the named attribute of the current element.
 495          * @param string $attr The name of the attribute
 496          * @return string The value of the attribute or an empty string if it is not set in the current
 497          * element.
 498          */
 499         public function nodeAttribute( $attr ) {
 500                 return $this->reader->getAttribute( $attr );
 501         }
 502
 503         /**
 504          * Shouldn't something like this be built-in to XMLReader?
 505          * Fetches text contents of the current element, assuming
 506          * no sub-elements or such scary things.
 507          * @return string
 508          * @access private
 509          */
 510         public function nodeContents() {
 511                 if ( $this->reader->isEmptyElement ) {
 512                         return "";
 513                 }
 514                 $buffer = "";
 515                 while ( $this->reader->read() ) {
 516                         switch ( $this->reader->nodeType ) {
 517                         case XMLReader::TEXT:
 518                         case XMLReader::CDATA:
 519                         case XMLReader::SIGNIFICANT_WHITESPACE:
 520                                 $buffer .= $this->reader->value;
 521                                 break;
 522                         case XMLReader::END_ELEMENT:
 523                                 return $buffer;
 524                         }
 525                 }
 526
 527                 $this->reader->close();
 528                 return '';
 529         }
 530
 531         /**
 532          * Primary entry point
 533          * @throws MWException
 534          * @return bool
 535          */
 536         public function doImport() {
 537                 // Calls to reader->read need to be wrapped in calls to
 538                 // libxml_disable_entity_loader() to avoid local file
 539                 // inclusion attacks (bug 46932).
 540                 $oldDisable = libxml_disable_entity_loader( true );
 541                 $this->reader->read();
 542
 543                 if ( $this->reader->localName != 'mediawiki' ) {
 544                         libxml_disable_entity_loader( $oldDisable );
 545                         throw new MWException( "Expected <mediawiki> tag, got " .
 546                                 $this->reader->localName );
 547                 }
 548                 $this->debug( "<mediawiki> tag is correct." );
 549
 550                 $this->debug( "Starting primary dump processing loop." );
 551
 552                 $keepReading = $this->reader->read();
 553                 $skip = false;
 554                 $rethrow = null;
 555                 try {
 556                         while ( $keepReading ) {
 557                                 $tag = $this->reader->localName;
 558                                 $type = $this->reader->nodeType;
 559
 560                                 if ( !Hooks::run( 'ImportHandleToplevelXMLTag', array( $this ) ) ) {
 561                                         // Do nothing
 562                                 } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) {
 563                                         break;
 564                                 } elseif ( $tag == 'siteinfo' ) {
 565                                         $this->handleSiteInfo();
 566                                 } elseif ( $tag == 'page' ) {
 567                                         $this->handlePage();
 568                                 } elseif ( $tag == 'logitem' ) {
 569                                         $this->handleLogItem();
 570                                 } elseif ( $tag != '#text' ) {
 571                                         $this->warn( "Unhandled top-level XML tag $tag" );
 572
 573                                         $skip = true;
 574                                 }
 575
 576                                 if ( $skip ) {
 577                                         $keepReading = $this->reader->next();
 578                                         $skip = false;
 579                                         $this->debug( "Skip" );
 580                                 } else {
 581                                         $keepReading = $this->reader->read();
 582                                 }
 583                         }
 584                 } catch ( Exception $ex ) {
 585                         $rethrow = $ex;
 586                 }
 587
 588                 // finally
 589                 libxml_disable_entity_loader( $oldDisable );
 590                 $this->reader->close();
 591
 592                 if ( $rethrow ) {
 593                         throw $rethrow;
 594                 }
 595
 596                 return true;
 597         }
 598
 599         private function handleSiteInfo() {
 600                 $this->debug( "Enter site info handler." );
 601                 $siteInfo = array();
 602
 603                 // Fields that can just be stuffed in the siteInfo object
 604                 $normalFields = array( 'sitename', 'base', 'generator', 'case' );
 605
 606                 while ( $this->reader->read() ) {
 607                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 608                                         $this->reader->localName == 'siteinfo' ) {
 609                                 break;
 610                         }
 611
 612                         $tag = $this->reader->localName;
 613
 614                         if ( $tag == 'namespace' ) {
 615                                 $this->foreignNamespaces[$this->nodeAttribute( 'key' )] =
 616                                         $this->nodeContents();
 617                         } elseif ( in_array( $tag, $normalFields ) ) {
 618                                 $siteInfo[$tag] = $this->nodeContents();
 619                         }
 620                 }
 621
 622                 $siteInfo['_namespaces'] = $this->foreignNamespaces;
 623                 $this->siteInfoCallback( $siteInfo );
 624         }
 625
 626         private function handleLogItem() {
 627                 $this->debug( "Enter log item handler." );
 628                 $logInfo = array();
 629
 630                 // Fields that can just be stuffed in the pageInfo object
 631                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 632                                         'logtitle', 'params' );
 633
 634                 while ( $this->reader->read() ) {
 635                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 636                                         $this->reader->localName == 'logitem' ) {
 637                                 break;
 638                         }
 639
 640                         $tag = $this->reader->localName;
 641
 642                         if ( !Hooks::run( 'ImportHandleLogItemXMLTag', array(
 643                                 $this, $logInfo
 644                         ) ) ) {
 645                                 // Do nothing
 646                         } elseif ( in_array( $tag, $normalFields ) ) {
 647                                 $logInfo[$tag] = $this->nodeContents();
 648                         } elseif ( $tag == 'contributor' ) {
 649                                 $logInfo['contributor'] = $this->handleContributor();
 650                         } elseif ( $tag != '#text' ) {
 651                                 $this->warn( "Unhandled log-item XML tag $tag" );
 652                         }
 653                 }
 654
 655                 $this->processLogItem( $logInfo );
 656         }
 657
 658         /**
 659          * @param array $logInfo
 660          * @return bool|mixed
 661          */
 662         private function processLogItem( $logInfo ) {
 663                 $revision = new WikiRevision( $this->config );
 664
 665                 $revision->setID( $logInfo['id'] );
 666                 $revision->setType( $logInfo['type'] );
 667                 $revision->setAction( $logInfo['action'] );
 668                 $revision->setTimestamp( $logInfo['timestamp'] );
 669                 $revision->setParams( $logInfo['params'] );
 670                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 671                 $revision->setNoUpdates( $this->mNoUpdates );
 672
 673                 if ( isset( $logInfo['comment'] ) ) {
 674                         $revision->setComment( $logInfo['comment'] );
 675                 }
 676
 677                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 678                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 679                 }
 680                 if ( isset( $logInfo['contributor']['username'] ) ) {
 681                         $revision->setUserName( $logInfo['contributor']['username'] );
 682                 }
 683
 684                 return $this->logItemCallback( $revision );
 685         }
 686
 687         private function handlePage() {
 688                 // Handle page data.
 689                 $this->debug( "Enter page handler." );
 690                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 691
 692                 // Fields that can just be stuffed in the pageInfo object
 693                 $normalFields = array( 'title', 'ns', 'id', 'redirect', 'restrictions' );
 694
 695                 $skip = false;
 696                 $badTitle = false;
 697
 698                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 699                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 700                                         $this->reader->localName == 'page' ) {
 701                                 break;
 702                         }
 703
 704                         $skip = false;
 705
 706                         $tag = $this->reader->localName;
 707
 708                         if ( $badTitle ) {
 709                                 // The title is invalid, bail out of this page
 710                                 $skip = true;
 711                         } elseif ( !Hooks::run( 'ImportHandlePageXMLTag', array( $this,
 712                                                 &$pageInfo ) ) ) {
 713                                 // Do nothing
 714                         } elseif ( in_array( $tag, $normalFields ) ) {
 715                                 // An XML snippet:
 716                                 // <page>
 717                                 //     <id>123</id>
 718                                 //     <title>Page</title>
 719                                 //     <redirect title="NewTitle"/>
 720                                 //     ...
 721                                 // Because the redirect tag is built differently, we need special handling for that case.
 722                                 if ( $tag == 'redirect' ) {
 723                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 724                                 } else {
 725                                         $pageInfo[$tag] = $this->nodeContents();
 726                                 }
 727                         } elseif ( $tag == 'revision' || $tag == 'upload' ) {
 728                                 if ( !isset( $title ) ) {
 729                                         $title = $this->processTitle( $pageInfo['title'],
 730                                                 isset( $pageInfo['ns'] ) ? $pageInfo['ns'] : null );
 731
 732                                         // $title is either an array of two titles or false.
 733                                         if ( is_array( $title ) ) {
 734                                                 $this->pageCallback( $title );
 735                                                 list( $pageInfo['_title'], $foreignTitle ) = $title;
 736                                         } else {
 737                                                 $badTitle = true;
 738                                                 $skip = true;
 739                                         }
 740                                 }
 741
 742                                 if ( $title ) {
 743                                         if ( $tag == 'revision' ) {
 744                                                 $this->handleRevision( $pageInfo );
 745                                         } else {
 746                                                 $this->handleUpload( $pageInfo );
 747                                         }
 748                                 }
 749                         } elseif ( $tag != '#text' ) {
 750                                 $this->warn( "Unhandled page XML tag $tag" );
 751                                 $skip = true;
 752                         }
 753                 }
 754
 755                 // @note $pageInfo is only set if a valid $title is processed above with
 756                 //       no error. If we have a valid $title, then pageCallback is called
 757                 //       above, $pageInfo['title'] is set and we do pageOutCallback here.
 758                 //       If $pageInfo['_title'] is not set, then $foreignTitle is also not
 759                 //       set since they both come from $title above.
 760                 if ( array_key_exists( '_title', $pageInfo ) ) {
 761                         $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
 762                                         $pageInfo['revisionCount'],
 763                                         $pageInfo['successfulRevisionCount'],
 764                                         $pageInfo );
 765                 }
 766         }
 767
 768         /**
 769          * @param array $pageInfo
 770          */
 771         private function handleRevision( &$pageInfo ) {
 772                 $this->debug( "Enter revision handler" );
 773                 $revisionInfo = array();
 774
 775                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' );
 776
 777                 $skip = false;
 778
 779                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 780                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 781                                         $this->reader->localName == 'revision' ) {
 782                                 break;
 783                         }
 784
 785                         $tag = $this->reader->localName;
 786
 787                         if ( !Hooks::run( 'ImportHandleRevisionXMLTag', array(
 788                                 $this, $pageInfo, $revisionInfo
 789                         ) ) ) {
 790                                 // Do nothing
 791                         } elseif ( in_array( $tag, $normalFields ) ) {
 792                                 $revisionInfo[$tag] = $this->nodeContents();
 793                         } elseif ( $tag == 'contributor' ) {
 794                                 $revisionInfo['contributor'] = $this->handleContributor();
 795                         } elseif ( $tag != '#text' ) {
 796                                 $this->warn( "Unhandled revision XML tag $tag" );
 797                                 $skip = true;
 798                         }
 799                 }
 800
 801                 $pageInfo['revisionCount']++;
 802                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 803                         $pageInfo['successfulRevisionCount']++;
 804                 }
 805         }
 806
 807         /**
 808          * @param array $pageInfo
 809          * @param array $revisionInfo
 810          * @return bool|mixed
 811          */
 812         private function processRevision( $pageInfo, $revisionInfo ) {
 813                 $revision = new WikiRevision( $this->config );
 814
 815                 if ( isset( $revisionInfo['id'] ) ) {
 816                         $revision->setID( $revisionInfo['id'] );
 817                 }
 818                 if ( isset( $revisionInfo['model'] ) ) {
 819                         $revision->setModel( $revisionInfo['model'] );
 820                 }
 821                 if ( isset( $revisionInfo['format'] ) ) {
 822                         $revision->setFormat( $revisionInfo['format'] );
 823                 }
 824                 $revision->setTitle( $pageInfo['_title'] );
 825
 826                 if ( isset( $revisionInfo['text'] ) ) {
 827                         $handler = $revision->getContentHandler();
 828                         $text = $handler->importTransform(
 829                                 $revisionInfo['text'],
 830                                 $revision->getFormat() );
 831
 832                         $revision->setText( $text );
 833                 }
 834                 if ( isset( $revisionInfo['timestamp'] ) ) {
 835                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 836                 } else {
 837                         $revision->setTimestamp( wfTimestampNow() );
 838                 }
 839
 840                 if ( isset( $revisionInfo['comment'] ) ) {
 841                         $revision->setComment( $revisionInfo['comment'] );
 842                 }
 843
 844                 if ( isset( $revisionInfo['minor'] ) ) {
 845                         $revision->setMinor( true );
 846                 }
 847                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 848                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 849                 }
 850                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 851                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 852                 }
 853                 $revision->setNoUpdates( $this->mNoUpdates );
 854
 855                 return $this->revisionCallback( $revision );
 856         }
 857
 858         /**
 859          * @param array $pageInfo
 860          * @return mixed
 861          */
 862         private function handleUpload( &$pageInfo ) {
 863                 $this->debug( "Enter upload handler" );
 864                 $uploadInfo = array();
 865
 866                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 867                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 868
 869                 $skip = false;
 870
 871                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 872                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 873                                         $this->reader->localName == 'upload' ) {
 874                                 break;
 875                         }
 876
 877                         $tag = $this->reader->localName;
 878
 879                         if ( !Hooks::run( 'ImportHandleUploadXMLTag', array(
 880                                 $this, $pageInfo
 881                         ) ) ) {
 882                                 // Do nothing
 883                         } elseif ( in_array( $tag, $normalFields ) ) {
 884                                 $uploadInfo[$tag] = $this->nodeContents();
 885                         } elseif ( $tag == 'contributor' ) {
 886                                 $uploadInfo['contributor'] = $this->handleContributor();
 887                         } elseif ( $tag == 'contents' ) {
 888                                 $contents = $this->nodeContents();
 889                                 $encoding = $this->reader->getAttribute( 'encoding' );
 890                                 if ( $encoding === 'base64' ) {
 891                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 892                                         $uploadInfo['isTempSrc'] = true;
 893                                 }
 894                         } elseif ( $tag != '#text' ) {
 895                                 $this->warn( "Unhandled upload XML tag $tag" );
 896                                 $skip = true;
 897                         }
 898                 }
 899
 900                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 901                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 902                         if ( file_exists( $path ) ) {
 903                                 $uploadInfo['fileSrc'] = $path;
 904                                 $uploadInfo['isTempSrc'] = false;
 905                         }
 906                 }
 907
 908                 if ( $this->mImportUploads ) {
 909                         return $this->processUpload( $pageInfo, $uploadInfo );
 910                 }
 911         }
 912
 913         /**
 914          * @param string $contents
 915          * @return string
 916          */
 917         private function dumpTemp( $contents ) {
 918                 $filename = tempnam( wfTempDir(), 'importupload' );
 919                 file_put_contents( $filename, $contents );
 920                 return $filename;
 921         }
 922
 923         /**
 924          * @param array $pageInfo
 925          * @param array $uploadInfo
 926          * @return mixed
 927          */
 928         private function processUpload( $pageInfo, $uploadInfo ) {
 929                 $revision = new WikiRevision( $this->config );
 930                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 931
 932                 $revision->setTitle( $pageInfo['_title'] );
 933                 $revision->setID( $pageInfo['id'] );
 934                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 935                 $revision->setText( $text );
 936                 $revision->setFilename( $uploadInfo['filename'] );
 937                 if ( isset( $uploadInfo['archivename'] ) ) {
 938                         $revision->setArchiveName( $uploadInfo['archivename'] );
 939                 }
 940                 $revision->setSrc( $uploadInfo['src'] );
 941                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 942                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 943                                 !empty( $uploadInfo['isTempSrc'] ) );
 944                 }
 945                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 946                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 947                 }
 948                 $revision->setSize( intval( $uploadInfo['size'] ) );
 949                 $revision->setComment( $uploadInfo['comment'] );
 950
 951                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 952                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 953                 }
 954                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 955                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 956                 }
 957                 $revision->setNoUpdates( $this->mNoUpdates );
 958
 959                 return call_user_func( $this->mUploadCallback, $revision );
 960         }
 961
 962         /**
 963          * @return array
 964          */
 965         private function handleContributor() {
 966                 $fields = array( 'id', 'ip', 'username' );
 967                 $info = array();
 968
 969                 while ( $this->reader->read() ) {
 970                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 971                                         $this->reader->localName == 'contributor' ) {
 972                                 break;
 973                         }
 974
 975                         $tag = $this->reader->localName;
 976
 977                         if ( in_array( $tag, $fields ) ) {
 978                                 $info[$tag] = $this->nodeContents();
 979                         }
 980                 }
 981
 982                 return $info;
 983         }
 984
 985         /**
 986          * @param string $text
 987          * @param string|null $ns
 988          * @return array|bool
 989          */
 990         private function processTitle( $text, $ns = null ) {
 991                 if ( is_null( $this->foreignNamespaces ) ) {
 992                         $foreignTitleFactory = new NaiveForeignTitleFactory();
 993                 } else {
 994                         $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
 995                                 $this->foreignNamespaces );
 996                 }
 997
 998                 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
 999                         intval( $ns ) );
1000
1001                 $title = $this->importTitleFactory->createTitleFromForeignTitle(
1002                         $foreignTitle );
1003
1004                 $commandLineMode = $this->config->get( 'CommandLineMode' );
1005                 if ( is_null( $title ) ) {
1006                         # Invalid page title? Ignore the page
1007                         $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
1008                         return false;
1009                 } elseif ( $title->isExternal() ) {
1010                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
1011                         return false;
1012                 } elseif ( !$title->canExist() ) {
1013                         $this->notice( 'import-error-special', $title->getPrefixedText() );
1014                         return false;
1015                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
1016                         # Do not import if the importing wiki user cannot edit this page
1017                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
1018                         return false;
1019                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
1020                         # Do not import if the importing wiki user cannot create this page
1021                         $this->notice( 'import-error-create', $title->getPrefixedText() );
1022                         return false;
1023                 }
1024
1025                 return array( $title, $foreignTitle );
1026         }
1027 }
1028
1029 /** This is a horrible hack used to keep source compatibility */
1030 class UploadSourceAdapter {
1031         /** @var array */
1032         public static $sourceRegistrations = array();
1033
1034         /** @var string */
1035         private $mSource;
1036
1037         /** @var string */
1038         private $mBuffer;
1039
1040         /** @var int */
1041         private $mPosition;
1042
1043         /**
1044          * @param ImportSource $source
1045          * @return string
1046          */
1047         static function registerSource( ImportSource $source ) {
1048                 $id = wfRandomString();
1049
1050                 self::$sourceRegistrations[$id] = $source;
1051
1052                 return $id;
1053         }
1054
1055         /**
1056          * @param string $path
1057          * @param string $mode
1058          * @param array $options
1059          * @param string $opened_path
1060          * @return bool
1061          */
1062         function stream_open( $path, $mode, $options, &$opened_path ) {
1063                 $url = parse_url( $path );
1064                 $id = $url['host'];
1065
1066                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
1067                         return false;
1068                 }
1069
1070                 $this->mSource = self::$sourceRegistrations[$id];
1071
1072                 return true;
1073         }
1074
1075         /**
1076          * @param int $count
1077          * @return string
1078          */
1079         function stream_read( $count ) {
1080                 $return = '';
1081                 $leave = false;
1082
1083                 while ( !$leave && !$this->mSource->atEnd() &&
1084                                 strlen( $this->mBuffer ) < $count ) {
1085                         $read = $this->mSource->readChunk();
1086
1087                         if ( !strlen( $read ) ) {
1088                                 $leave = true;
1089                         }
1090
1091                         $this->mBuffer .= $read;
1092                 }
1093
1094                 if ( strlen( $this->mBuffer ) ) {
1095                         $return = substr( $this->mBuffer, 0, $count );
1096                         $this->mBuffer = substr( $this->mBuffer, $count );
1097                 }
1098
1099                 $this->mPosition += strlen( $return );
1100
1101                 return $return;
1102         }
1103
1104         /**
1105          * @param string $data
1106          * @return bool
1107          */
1108         function stream_write( $data ) {
1109                 return false;
1110         }
1111
1112         /**
1113          * @return mixed
1114          */
1115         function stream_tell() {
1116                 return $this->mPosition;
1117         }
1118
1119         /**
1120          * @return bool
1121          */
1122         function stream_eof() {
1123                 return $this->mSource->atEnd();
1124         }
1125
1126         /**
1127          * @return array
1128          */
1129         function url_stat() {
1130                 $result = array();
1131
1132                 $result['dev'] = $result[0] = 0;
1133                 $result['ino'] = $result[1] = 0;
1134                 $result['mode'] = $result[2] = 0;
1135                 $result['nlink'] = $result[3] = 0;
1136                 $result['uid'] = $result[4] = 0;
1137                 $result['gid'] = $result[5] = 0;
1138                 $result['rdev'] = $result[6] = 0;
1139                 $result['size'] = $result[7] = 0;
1140                 $result['atime'] = $result[8] = 0;
1141                 $result['mtime'] = $result[9] = 0;
1142                 $result['ctime'] = $result[10] = 0;
1143                 $result['blksize'] = $result[11] = 0;
1144                 $result['blocks'] = $result[12] = 0;
1145
1146                 return $result;
1147         }
1148 }
1149
1150 /**
1151  * @todo document (e.g. one-sentence class description).
1152  * @ingroup SpecialPage
1153  */
1154 class WikiRevision {
1155         /** @todo Unused? */
1156         public $importer = null;
1157
1158         /** @var Title */
1159         public $title = null;
1160
1161         /** @var int */
1162         public $id = 0;
1163
1164         /** @var string */
1165         public $timestamp = "20010115000000";
1166
1167         /**
1168          * @var int
1169          * @todo Can't find any uses. Public, because that's suspicious. Get clarity. */
1170         public $user = 0;
1171
1172         /** @var string */
1173         public $user_text = "";
1174
1175         /** @var string */
1176         public $model = null;
1177
1178         /** @var string */
1179         public $format = null;
1180
1181         /** @var string */
1182         public $text = "";
1183
1184         /** @var int */
1185         protected $size;
1186
1187         /** @var Content */
1188         public $content = null;
1189
1190         /** @var ContentHandler */
1191         protected $contentHandler = null;
1192
1193         /** @var string */
1194         public $comment = "";
1195
1196         /** @var bool */
1197         public $minor = false;
1198
1199         /** @var string */
1200         public $type = "";
1201
1202         /** @var string */
1203         public $action = "";
1204
1205         /** @var string */
1206         public $params = "";
1207
1208         /** @var string */
1209         public $fileSrc = '';
1210
1211         /** @var bool|string */
1212         public $sha1base36 = false;
1213
1214         /**
1215          * @var bool
1216          * @todo Unused?
1217          */
1218         public $isTemp = false;
1219
1220         /** @var string */
1221         public $archiveName = '';
1222
1223         protected $filename;
1224
1225         /** @var mixed */
1226         protected $src;
1227
1228         /** @todo Unused? */
1229         public $fileIsTemp;
1230
1231         /** @var bool */
1232         private $mNoUpdates = false;
1233
1234         /** @var Config $config */
1235         private $config;
1236
1237         public function __construct( Config $config ) {
1238                 $this->config = $config;
1239         }
1240
1241         /**
1242          * @param Title $title
1243          * @throws MWException
1244          */
1245         function setTitle( $title ) {
1246                 if ( is_object( $title ) ) {
1247                         $this->title = $title;
1248                 } elseif ( is_null( $title ) ) {
1249                         throw new MWException( "WikiRevision given a null title in import. "
1250                                 . "You may need to adjust \$wgLegalTitleChars." );
1251                 } else {
1252                         throw new MWException( "WikiRevision given non-object title in import." );
1253                 }
1254         }
1255
1256         /**
1257          * @param int $id
1258          */
1259         function setID( $id ) {
1260                 $this->id = $id;
1261         }
1262
1263         /**
1264          * @param string $ts
1265          */
1266         function setTimestamp( $ts ) {
1267                 # 2003-08-05T18:30:02Z
1268                 $this->timestamp = wfTimestamp( TS_MW, $ts );
1269         }
1270
1271         /**
1272          * @param string $user
1273          */
1274         function setUsername( $user ) {
1275                 $this->user_text = $user;
1276         }
1277
1278         /**
1279          * @param string $ip
1280          */
1281         function setUserIP( $ip ) {
1282                 $this->user_text = $ip;
1283         }
1284
1285         /**
1286          * @param string $model
1287          */
1288         function setModel( $model ) {
1289                 $this->model = $model;
1290         }
1291
1292         /**
1293          * @param string $format
1294          */
1295         function setFormat( $format ) {
1296                 $this->format = $format;
1297         }
1298
1299         /**
1300          * @param string $text
1301          */
1302         function setText( $text ) {
1303                 $this->text = $text;
1304         }
1305
1306         /**
1307          * @param string $text
1308          */
1309         function setComment( $text ) {
1310                 $this->comment = $text;
1311         }
1312
1313         /**
1314          * @param bool $minor
1315          */
1316         function setMinor( $minor ) {
1317                 $this->minor = (bool)$minor;
1318         }
1319
1320         /**
1321          * @param mixed $src
1322          */
1323         function setSrc( $src ) {
1324                 $this->src = $src;
1325         }
1326
1327         /**
1328          * @param string $src
1329          * @param bool $isTemp
1330          */
1331         function setFileSrc( $src, $isTemp ) {
1332                 $this->fileSrc = $src;
1333                 $this->fileIsTemp = $isTemp;
1334         }
1335
1336         /**
1337          * @param string $sha1base36
1338          */
1339         function setSha1Base36( $sha1base36 ) {
1340                 $this->sha1base36 = $sha1base36;
1341         }
1342
1343         /**
1344          * @param string $filename
1345          */
1346         function setFilename( $filename ) {
1347                 $this->filename = $filename;
1348         }
1349
1350         /**
1351          * @param string $archiveName
1352          */
1353         function setArchiveName( $archiveName ) {
1354                 $this->archiveName = $archiveName;
1355         }
1356
1357         /**
1358          * @param int $size
1359          */
1360         function setSize( $size ) {
1361                 $this->size = intval( $size );
1362         }
1363
1364         /**
1365          * @param string $type
1366          */
1367         function setType( $type ) {
1368                 $this->type = $type;
1369         }
1370
1371         /**
1372          * @param string $action
1373          */
1374         function setAction( $action ) {
1375                 $this->action = $action;
1376         }
1377
1378         /**
1379          * @param array $params
1380          */
1381         function setParams( $params ) {
1382                 $this->params = $params;
1383         }
1384
1385         /**
1386          * @param bool $noupdates
1387          */
1388         public function setNoUpdates( $noupdates ) {
1389                 $this->mNoUpdates = $noupdates;
1390         }
1391
1392         /**
1393          * @return Title
1394          */
1395         function getTitle() {
1396                 return $this->title;
1397         }
1398
1399         /**
1400          * @return int
1401          */
1402         function getID() {
1403                 return $this->id;
1404         }
1405
1406         /**
1407          * @return string
1408          */
1409         function getTimestamp() {
1410                 return $this->timestamp;
1411         }
1412
1413         /**
1414          * @return string
1415          */
1416         function getUser() {
1417                 return $this->user_text;
1418         }
1419
1420         /**
1421          * @return string
1422          *
1423          * @deprecated Since 1.21, use getContent() instead.
1424          */
1425         function getText() {
1426                 ContentHandler::deprecated( __METHOD__, '1.21' );
1427
1428                 return $this->text;
1429         }
1430
1431         /**
1432          * @return ContentHandler
1433          */
1434         function getContentHandler() {
1435                 if ( is_null( $this->contentHandler ) ) {
1436                         $this->contentHandler = ContentHandler::getForModelID( $this->getModel() );
1437                 }
1438
1439                 return $this->contentHandler;
1440         }
1441
1442         /**
1443          * @return Content
1444          */
1445         function getContent() {
1446                 if ( is_null( $this->content ) ) {
1447                         $handler = $this->getContentHandler();
1448                         $this->content = $handler->unserializeContent( $this->text, $this->getFormat() );
1449                 }
1450
1451                 return $this->content;
1452         }
1453
1454         /**
1455          * @return string
1456          */
1457         function getModel() {
1458                 if ( is_null( $this->model ) ) {
1459                         $this->model = $this->getTitle()->getContentModel();
1460                 }
1461
1462                 return $this->model;
1463         }
1464
1465         /**
1466          * @return string
1467          */
1468         function getFormat() {
1469                 if ( is_null( $this->format ) ) {
1470                         $this->format = $this->getContentHandler()->getDefaultFormat();
1471                 }
1472
1473                 return $this->format;
1474         }
1475
1476         /**
1477          * @return string
1478          */
1479         function getComment() {
1480                 return $this->comment;
1481         }
1482
1483         /**
1484          * @return bool
1485          */
1486         function getMinor() {
1487                 return $this->minor;
1488         }
1489
1490         /**
1491          * @return mixed
1492          */
1493         function getSrc() {
1494                 return $this->src;
1495         }
1496
1497         /**
1498          * @return bool|string
1499          */
1500         function getSha1() {
1501                 if ( $this->sha1base36 ) {
1502                         return wfBaseConvert( $this->sha1base36, 36, 16 );
1503                 }
1504                 return false;
1505         }
1506
1507         /**
1508          * @return string
1509          */
1510         function getFileSrc() {
1511                 return $this->fileSrc;
1512         }
1513
1514         /**
1515          * @return bool
1516          */
1517         function isTempSrc() {
1518                 return $this->isTemp;
1519         }
1520
1521         /**
1522          * @return mixed
1523          */
1524         function getFilename() {
1525                 return $this->filename;
1526         }
1527
1528         /**
1529          * @return string
1530          */
1531         function getArchiveName() {
1532                 return $this->archiveName;
1533         }
1534
1535         /**
1536          * @return mixed
1537          */
1538         function getSize() {
1539                 return $this->size;
1540         }
1541
1542         /**
1543          * @return string
1544          */
1545         function getType() {
1546                 return $this->type;
1547         }
1548
1549         /**
1550          * @return string
1551          */
1552         function getAction() {
1553                 return $this->action;
1554         }
1555
1556         /**
1557          * @return string
1558          */
1559         function getParams() {
1560                 return $this->params;
1561         }
1562
1563         /**
1564          * @return bool
1565          */
1566         function importOldRevision() {
1567                 $dbw = wfGetDB( DB_MASTER );
1568
1569                 # Sneak a single revision into place
1570                 $user = User::newFromName( $this->getUser() );
1571                 if ( $user ) {
1572                         $userId = intval( $user->getId() );
1573                         $userText = $user->getName();
1574                         $userObj = $user;
1575                 } else {
1576                         $userId = 0;
1577                         $userText = $this->getUser();
1578                         $userObj = new User;
1579                 }
1580
1581                 // avoid memory leak...?
1582                 Title::clearCaches();
1583
1584                 $page = WikiPage::factory( $this->title );
1585                 $page->loadPageData( 'fromdbmaster' );
1586                 if ( !$page->exists() ) {
1587                         # must create the page...
1588                         $pageId = $page->insertOn( $dbw );
1589                         $created = true;
1590                         $oldcountable = null;
1591                 } else {
1592                         $pageId = $page->getId();
1593                         $created = false;
1594
1595                         $prior = $dbw->selectField( 'revision', '1',
1596                                 array( 'rev_page' => $pageId,
1597                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1598                                         'rev_user_text' => $userText,
1599                                         'rev_comment' => $this->getComment() ),
1600                                 __METHOD__
1601                         );
1602                         if ( $prior ) {
1603                                 // @todo FIXME: This could fail slightly for multiple matches :P
1604                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1605                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1606                                 return false;
1607                         }
1608                 }
1609
1610                 // Select previous version to make size diffs correct
1611                 $prevId = $dbw->selectField( 'revision', 'rev_id',
1612                         array(
1613                                 'rev_page' => $pageId,
1614                                 'rev_timestamp <= ' . $dbw->timestamp( $this->timestamp ),
1615                         ),
1616                         __METHOD__,
1617                         array( 'ORDER BY' => array(
1618                                         'rev_timestamp DESC',
1619                                         'rev_id DESC', // timestamp is not unique per page
1620                                 )
1621                         )
1622                 );
1623
1624                 # @todo FIXME: Use original rev_id optionally (better for backups)
1625                 # Insert the row
1626                 $revision = new Revision( array(
1627                         'title' => $this->title,
1628                         'page' => $pageId,
1629                         'content_model' => $this->getModel(),
1630                         'content_format' => $this->getFormat(),
1631                         // XXX: just set 'content' => $this->getContent()?
1632                         'text' => $this->getContent()->serialize( $this->getFormat() ),
1633                         'comment' => $this->getComment(),
1634                         'user' => $userId,
1635                         'user_text' => $userText,
1636                         'timestamp' => $this->timestamp,
1637                         'minor_edit' => $this->minor,
1638                         'parent_id' => $prevId,
1639                         ) );
1640                 $revision->insertOn( $dbw );
1641                 $changed = $page->updateIfNewerOn( $dbw, $revision );
1642
1643                 if ( $changed !== false && !$this->mNoUpdates ) {
1644                         wfDebug( __METHOD__ . ": running updates\n" );
1645                         // countable/oldcountable stuff is handled in WikiImporter::finishImportPage
1646                         $page->doEditUpdates(
1647                                 $revision,
1648                                 $userObj,
1649                                 array( 'created' => $created, 'oldcountable' => 'no-change' )
1650                         );
1651                 }
1652
1653                 return true;
1654         }
1655
1656         function importLogItem() {
1657                 $dbw = wfGetDB( DB_MASTER );
1658                 # @todo FIXME: This will not record autoblocks
1659                 if ( !$this->getTitle() ) {
1660                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1661                                 $this->timestamp . "\n" );
1662                         return;
1663                 }
1664                 # Check if it exists already
1665                 // @todo FIXME: Use original log ID (better for backups)
1666                 $prior = $dbw->selectField( 'logging', '1',
1667                         array( 'log_type' => $this->getType(),
1668                                 'log_action' => $this->getAction(),
1669                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1670                                 'log_namespace' => $this->getTitle()->getNamespace(),
1671                                 'log_title' => $this->getTitle()->getDBkey(),
1672                                 'log_comment' => $this->getComment(),
1673                                 # 'log_user_text' => $this->user_text,
1674                                 'log_params' => $this->params ),
1675                         __METHOD__
1676                 );
1677                 // @todo FIXME: This could fail slightly for multiple matches :P
1678                 if ( $prior ) {
1679                         wfDebug( __METHOD__
1680                                 . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp "
1681                                 . $this->timestamp . "\n" );
1682                         return;
1683                 }
1684                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1685                 $data = array(
1686                         'log_id' => $log_id,
1687                         'log_type' => $this->type,
1688                         'log_action' => $this->action,
1689                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1690                         'log_user' => User::idFromName( $this->user_text ),
1691                         # 'log_user_text' => $this->user_text,
1692                         'log_namespace' => $this->getTitle()->getNamespace(),
1693                         'log_title' => $this->getTitle()->getDBkey(),
1694                         'log_comment' => $this->getComment(),
1695                         'log_params' => $this->params
1696                 );
1697                 $dbw->insert( 'logging', $data, __METHOD__ );
1698         }
1699
1700         /**
1701          * @return bool
1702          */
1703         function importUpload() {
1704                 # Construct a file
1705                 $archiveName = $this->getArchiveName();
1706                 if ( $archiveName ) {
1707                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1708                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1709                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1710                 } else {
1711                         $file = wfLocalFile( $this->getTitle() );
1712                         $file->load( File::READ_LATEST );
1713                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1714                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1715                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1716                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1717                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1718                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1719                         }
1720                 }
1721                 if ( !$file ) {
1722                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1723                         return false;
1724                 }
1725
1726                 # Get the file source or download if necessary
1727                 $source = $this->getFileSrc();
1728                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1729                 if ( !$source ) {
1730                         $source = $this->downloadSource();
1731                         $flags |= File::DELETE_SOURCE;
1732                 }
1733                 if ( !$source ) {
1734                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1735                         return false;
1736                 }
1737                 $sha1 = $this->getSha1();
1738                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1739                         if ( $flags & File::DELETE_SOURCE ) {
1740                                 # Broken file; delete it if it is a temporary file
1741                                 unlink( $source );
1742                         }
1743                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1744                         return false;
1745                 }
1746
1747                 $user = User::newFromName( $this->user_text );
1748
1749                 # Do the actual upload
1750                 if ( $archiveName ) {
1751                         $status = $file->uploadOld( $source, $archiveName,
1752                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1753                 } else {
1754                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1755                                 $flags, false, $this->getTimestamp(), $user );
1756                 }
1757
1758                 if ( $status->isGood() ) {
1759                         wfDebug( __METHOD__ . ": Successful\n" );
1760                         return true;
1761                 } else {
1762                         wfDebug( __METHOD__ . ': failed: ' . $status->getHTML() . "\n" );
1763                         return false;
1764                 }
1765         }
1766
1767         /**
1768          * @return bool|string
1769          */
1770         function downloadSource() {
1771                 if ( !$this->config->get( 'EnableUploads' ) ) {
1772                         return false;
1773                 }
1774
1775                 $tempo = tempnam( wfTempDir(), 'download' );
1776                 $f = fopen( $tempo, 'wb' );
1777                 if ( !$f ) {
1778                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1779                         return false;
1780                 }
1781
1782                 // @todo FIXME!
1783                 $src = $this->getSrc();
1784                 $data = Http::get( $src, array(), __METHOD__ );
1785                 if ( !$data ) {
1786                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1787                         fclose( $f );
1788                         unlink( $tempo );
1789                         return false;
1790                 }
1791
1792                 fwrite( $f, $data );
1793                 fclose( $f );
1794
1795                 return $tempo;
1796         }
1797
1798 }
1799
1800 /**
1801  * Source interface for XML import.
1802  */
1803 interface ImportSource {
1804
1805         /**
1806          * Indicates whether the end of the input has been reached.
1807          * Will return true after a finite number of calls to readChunk.
1808          *
1809          * @return bool true if there is no more input, false otherwise.
1810          */
1811         function atEnd();
1812
1813         /**
1814          * Return a chunk of the input, as a (possibly empty) string.
1815          * When the end of input is reached, readChunk() returns false.
1816          * If atEnd() returns false, readChunk() will return a string.
1817          * If atEnd() returns true, readChunk() will return false.
1818          *
1819          * @return bool|string
1820          */
1821         function readChunk();
1822 }
1823
1824 /**
1825  * Used for importing XML dumps where the content of the dump is in a string.
1826  * This class is ineffecient, and should only be used for small dumps.
1827  * For larger dumps, ImportStreamSource should be used instead.
1828  *
1829  * @ingroup SpecialPage
1830  */
1831 class ImportStringSource implements ImportSource {
1832         function __construct( $string ) {
1833                 $this->mString = $string;
1834                 $this->mRead = false;
1835         }
1836
1837         /**
1838          * @return bool
1839          */
1840         function atEnd() {
1841                 return $this->mRead;
1842         }
1843
1844         /**
1845          * @return bool|string
1846          */
1847         function readChunk() {
1848                 if ( $this->atEnd() ) {
1849                         return false;
1850                 }
1851                 $this->mRead = true;
1852                 return $this->mString;
1853         }
1854 }
1855
1856 /**
1857  * Imports a XML dump from a file (either from file upload, files on disk, or HTTP)
1858  * @ingroup SpecialPage
1859  */
1860 class ImportStreamSource implements ImportSource {
1861         function __construct( $handle ) {
1862                 $this->mHandle = $handle;
1863         }
1864
1865         /**
1866          * @return bool
1867          */
1868         function atEnd() {
1869                 return feof( $this->mHandle );
1870         }
1871
1872         /**
1873          * @return string
1874          */
1875         function readChunk() {
1876                 return fread( $this->mHandle, 32768 );
1877         }
1878
1879         /**
1880          * @param string $filename
1881          * @return Status
1882          */
1883         static function newFromFile( $filename ) {
1884                 MediaWiki\suppressWarnings();
1885                 $file = fopen( $filename, 'rt' );
1886                 MediaWiki\restoreWarnings();
1887                 if ( !$file ) {
1888                         return Status::newFatal( "importcantopen" );
1889                 }
1890                 return Status::newGood( new ImportStreamSource( $file ) );
1891         }
1892
1893         /**
1894          * @param string $fieldname
1895          * @return Status
1896          */
1897         static function newFromUpload( $fieldname = "xmlimport" ) {
1898                 $upload =& $_FILES[$fieldname];
1899
1900                 if ( $upload === null || !$upload['name'] ) {
1901                         return Status::newFatal( 'importnofile' );
1902                 }
1903                 if ( !empty( $upload['error'] ) ) {
1904                         switch ( $upload['error'] ) {
1905                                 case 1:
1906                                         # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1907                                         return Status::newFatal( 'importuploaderrorsize' );
1908                                 case 2:
1909                                         # The uploaded file exceeds the MAX_FILE_SIZE directive that
1910                                         # was specified in the HTML form.
1911                                         return Status::newFatal( 'importuploaderrorsize' );
1912                                 case 3:
1913                                         # The uploaded file was only partially uploaded
1914                                         return Status::newFatal( 'importuploaderrorpartial' );
1915                                 case 6:
1916                                         # Missing a temporary folder.
1917                                         return Status::newFatal( 'importuploaderrortemp' );
1918                                 # case else: # Currently impossible
1919                         }
1920
1921                 }
1922                 $fname = $upload['tmp_name'];
1923                 if ( is_uploaded_file( $fname ) ) {
1924                         return ImportStreamSource::newFromFile( $fname );
1925                 } else {
1926                         return Status::newFatal( 'importnofile' );
1927                 }
1928         }
1929
1930         /**
1931          * @param string $url
1932          * @param string $method
1933          * @return Status
1934          */
1935         static function newFromURL( $url, $method = 'GET' ) {
1936                 wfDebug( __METHOD__ . ": opening $url\n" );
1937                 # Use the standard HTTP fetch function; it times out
1938                 # quicker and sorts out user-agent problems which might
1939                 # otherwise prevent importing from large sites, such
1940                 # as the Wikimedia cluster, etc.
1941                 $data = Http::request( $method, $url, array( 'followRedirects' => true ), __METHOD__ );
1942                 if ( $data !== false ) {
1943                         $file = tmpfile();
1944                         fwrite( $file, $data );
1945                         fflush( $file );
1946                         fseek( $file, 0 );
1947                         return Status::newGood( new ImportStreamSource( $file ) );
1948                 } else {
1949                         return Status::newFatal( 'importcantopen' );
1950                 }
1951         }
1952
1953         /**
1954          * @param string $interwiki
1955          * @param string $page
1956          * @param bool $history
1957          * @param bool $templates
1958          * @param int $pageLinkDepth
1959          * @return Status
1960          */
1961         public static function newFromInterwiki( $interwiki, $page, $history = false,
1962                 $templates = false, $pageLinkDepth = 0
1963         ) {
1964                 if ( $page == '' ) {
1965                         return Status::newFatal( 'import-noarticle' );
1966                 }
1967
1968                 # Look up the first interwiki prefix, and let the foreign site handle
1969                 # subsequent interwiki prefixes
1970                 $firstIwPrefix = strtok( $interwiki, ':' );
1971                 $firstIw = Interwiki::fetch( $firstIwPrefix );
1972                 if ( !$firstIw ) {
1973                         return Status::newFatal( 'importbadinterwiki' );
1974                 }
1975
1976                 $additionalIwPrefixes = strtok( '' );
1977                 if ( $additionalIwPrefixes ) {
1978                         $additionalIwPrefixes .= ':';
1979                 }
1980                 # Have to do a DB-key replacement ourselves; otherwise spaces get
1981                 # URL-encoded to +, which is wrong in this case. Similar to logic in
1982                 # Title::getLocalURL
1983                 $link = $firstIw->getURL( strtr( "${additionalIwPrefixes}Special:Export/$page",
1984                         ' ', '_' ) );
1985
1986                 $params = array();
1987                 if ( $history ) {
1988                         $params['history'] = 1;
1989                 }
1990                 if ( $templates ) {
1991                         $params['templates'] = 1;
1992                 }
1993                 if ( $pageLinkDepth ) {
1994                         $params['pagelink-depth'] = $pageLinkDepth;
1995                 }
1996
1997                 $url = wfAppendQuery( $link, $params );
1998                 # For interwikis, use POST to avoid redirects.
1999                 return ImportStreamSource::newFromURL( $url, "POST" );
2000         }
2001 }