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::SIGNIFICANT_WHITESPACE:
 519                                 $buffer .= $this->reader->value;
 520                                 break;
 521                         case XMLReader::END_ELEMENT:
 522                                 return $buffer;
 523                         }
 524                 }
 525
 526                 $this->reader->close();
 527                 return '';
 528         }
 529
 530         /**
 531          * Primary entry point
 532          * @throws MWException
 533          * @return bool
 534          */
 535         public function doImport() {
 536                 // Calls to reader->read need to be wrapped in calls to
 537                 // libxml_disable_entity_loader() to avoid local file
 538                 // inclusion attacks (bug 46932).
 539                 $oldDisable = libxml_disable_entity_loader( true );
 540                 $this->reader->read();
 541
 542                 if ( $this->reader->localName != 'mediawiki' ) {
 543                         libxml_disable_entity_loader( $oldDisable );
 544                         throw new MWException( "Expected <mediawiki> tag, got " .
 545                                 $this->reader->localName );
 546                 }
 547                 $this->debug( "<mediawiki> tag is correct." );
 548
 549                 $this->debug( "Starting primary dump processing loop." );
 550
 551                 $keepReading = $this->reader->read();
 552                 $skip = false;
 553                 $rethrow = null;
 554                 try {
 555                         while ( $keepReading ) {
 556                                 $tag = $this->reader->localName;
 557                                 $type = $this->reader->nodeType;
 558
 559                                 if ( !Hooks::run( 'ImportHandleToplevelXMLTag', array( $this ) ) ) {
 560                                         // Do nothing
 561                                 } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) {
 562                                         break;
 563                                 } elseif ( $tag == 'siteinfo' ) {
 564                                         $this->handleSiteInfo();
 565                                 } elseif ( $tag == 'page' ) {
 566                                         $this->handlePage();
 567                                 } elseif ( $tag == 'logitem' ) {
 568                                         $this->handleLogItem();
 569                                 } elseif ( $tag != '#text' ) {
 570                                         $this->warn( "Unhandled top-level XML tag $tag" );
 571
 572                                         $skip = true;
 573                                 }
 574
 575                                 if ( $skip ) {
 576                                         $keepReading = $this->reader->next();
 577                                         $skip = false;
 578                                         $this->debug( "Skip" );
 579                                 } else {
 580                                         $keepReading = $this->reader->read();
 581                                 }
 582                         }
 583                 } catch ( Exception $ex ) {
 584                         $rethrow = $ex;
 585                 }
 586
 587                 // finally
 588                 libxml_disable_entity_loader( $oldDisable );
 589                 $this->reader->close();
 590
 591                 if ( $rethrow ) {
 592                         throw $rethrow;
 593                 }
 594
 595                 return true;
 596         }
 597
 598         private function handleSiteInfo() {
 599                 $this->debug( "Enter site info handler." );
 600                 $siteInfo = array();
 601
 602                 // Fields that can just be stuffed in the siteInfo object
 603                 $normalFields = array( 'sitename', 'base', 'generator', 'case' );
 604
 605                 while ( $this->reader->read() ) {
 606                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 607                                         $this->reader->localName == 'siteinfo' ) {
 608                                 break;
 609                         }
 610
 611                         $tag = $this->reader->localName;
 612
 613                         if ( $tag == 'namespace' ) {
 614                                 $this->foreignNamespaces[$this->nodeAttribute( 'key' )] =
 615                                         $this->nodeContents();
 616                         } elseif ( in_array( $tag, $normalFields ) ) {
 617                                 $siteInfo[$tag] = $this->nodeContents();
 618                         }
 619                 }
 620
 621                 $siteInfo['_namespaces'] = $this->foreignNamespaces;
 622                 $this->siteInfoCallback( $siteInfo );
 623         }
 624
 625         private function handleLogItem() {
 626                 $this->debug( "Enter log item handler." );
 627                 $logInfo = array();
 628
 629                 // Fields that can just be stuffed in the pageInfo object
 630                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 631                                         'logtitle', 'params' );
 632
 633                 while ( $this->reader->read() ) {
 634                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 635                                         $this->reader->localName == 'logitem' ) {
 636                                 break;
 637                         }
 638
 639                         $tag = $this->reader->localName;
 640
 641                         if ( !Hooks::run( 'ImportHandleLogItemXMLTag', array(
 642                                 $this, $logInfo
 643                         ) ) ) {
 644                                 // Do nothing
 645                         } elseif ( in_array( $tag, $normalFields ) ) {
 646                                 $logInfo[$tag] = $this->nodeContents();
 647                         } elseif ( $tag == 'contributor' ) {
 648                                 $logInfo['contributor'] = $this->handleContributor();
 649                         } elseif ( $tag != '#text' ) {
 650                                 $this->warn( "Unhandled log-item XML tag $tag" );
 651                         }
 652                 }
 653
 654                 $this->processLogItem( $logInfo );
 655         }
 656
 657         /**
 658          * @param array $logInfo
 659          * @return bool|mixed
 660          */
 661         private function processLogItem( $logInfo ) {
 662                 $revision = new WikiRevision( $this->config );
 663
 664                 $revision->setID( $logInfo['id'] );
 665                 $revision->setType( $logInfo['type'] );
 666                 $revision->setAction( $logInfo['action'] );
 667                 $revision->setTimestamp( $logInfo['timestamp'] );
 668                 $revision->setParams( $logInfo['params'] );
 669                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 670                 $revision->setNoUpdates( $this->mNoUpdates );
 671
 672                 if ( isset( $logInfo['comment'] ) ) {
 673                         $revision->setComment( $logInfo['comment'] );
 674                 }
 675
 676                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 677                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 678                 }
 679                 if ( isset( $logInfo['contributor']['username'] ) ) {
 680                         $revision->setUserName( $logInfo['contributor']['username'] );
 681                 }
 682
 683                 return $this->logItemCallback( $revision );
 684         }
 685
 686         private function handlePage() {
 687                 // Handle page data.
 688                 $this->debug( "Enter page handler." );
 689                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 690
 691                 // Fields that can just be stuffed in the pageInfo object
 692                 $normalFields = array( 'title', 'ns', 'id', 'redirect', 'restrictions' );
 693
 694                 $skip = false;
 695                 $badTitle = false;
 696
 697                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 698                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 699                                         $this->reader->localName == 'page' ) {
 700                                 break;
 701                         }
 702
 703                         $skip = false;
 704
 705                         $tag = $this->reader->localName;
 706
 707                         if ( $badTitle ) {
 708                                 // The title is invalid, bail out of this page
 709                                 $skip = true;
 710                         } elseif ( !Hooks::run( 'ImportHandlePageXMLTag', array( $this,
 711                                                 &$pageInfo ) ) ) {
 712                                 // Do nothing
 713                         } elseif ( in_array( $tag, $normalFields ) ) {
 714                                 // An XML snippet:
 715                                 // <page>
 716                                 //     <id>123</id>
 717                                 //     <title>Page</title>
 718                                 //     <redirect title="NewTitle"/>
 719                                 //     ...
 720                                 // Because the redirect tag is built differently, we need special handling for that case.
 721                                 if ( $tag == 'redirect' ) {
 722                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 723                                 } else {
 724                                         $pageInfo[$tag] = $this->nodeContents();
 725                                 }
 726                         } elseif ( $tag == 'revision' || $tag == 'upload' ) {
 727                                 if ( !isset( $title ) ) {
 728                                         $title = $this->processTitle( $pageInfo['title'],
 729                                                 isset( $pageInfo['ns'] ) ? $pageInfo['ns'] : null );
 730
 731                                         // $title is either an array of two titles or false.
 732                                         if ( is_array( $title ) ) {
 733                                                 $this->pageCallback( $title );
 734                                                 list( $pageInfo['_title'], $foreignTitle ) = $title;
 735                                         } else {
 736                                                 $badTitle = true;
 737                                                 $skip = true;
 738                                         }
 739                                 }
 740
 741                                 if ( $title ) {
 742                                         if ( $tag == 'revision' ) {
 743                                                 $this->handleRevision( $pageInfo );
 744                                         } else {
 745                                                 $this->handleUpload( $pageInfo );
 746                                         }
 747                                 }
 748                         } elseif ( $tag != '#text' ) {
 749                                 $this->warn( "Unhandled page XML tag $tag" );
 750                                 $skip = true;
 751                         }
 752                 }
 753
 754                 // @note $pageInfo is only set if a valid $title is processed above with
 755                 //       no error. If we have a valid $title, then pageCallback is called
 756                 //       above, $pageInfo['title'] is set and we do pageOutCallback here.
 757                 //       If $pageInfo['_title'] is not set, then $foreignTitle is also not
 758                 //       set since they both come from $title above.
 759                 if ( array_key_exists( '_title', $pageInfo ) ) {
 760                         $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
 761                                         $pageInfo['revisionCount'],
 762                                         $pageInfo['successfulRevisionCount'],
 763                                         $pageInfo );
 764                 }
 765         }
 766
 767         /**
 768          * @param array $pageInfo
 769          */
 770         private function handleRevision( &$pageInfo ) {
 771                 $this->debug( "Enter revision handler" );
 772                 $revisionInfo = array();
 773
 774                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' );
 775
 776                 $skip = false;
 777
 778                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 779                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 780                                         $this->reader->localName == 'revision' ) {
 781                                 break;
 782                         }
 783
 784                         $tag = $this->reader->localName;
 785
 786                         if ( !Hooks::run( 'ImportHandleRevisionXMLTag', array(
 787                                 $this, $pageInfo, $revisionInfo
 788                         ) ) ) {
 789                                 // Do nothing
 790                         } elseif ( in_array( $tag, $normalFields ) ) {
 791                                 $revisionInfo[$tag] = $this->nodeContents();
 792                         } elseif ( $tag == 'contributor' ) {
 793                                 $revisionInfo['contributor'] = $this->handleContributor();
 794                         } elseif ( $tag != '#text' ) {
 795                                 $this->warn( "Unhandled revision XML tag $tag" );
 796                                 $skip = true;
 797                         }
 798                 }
 799
 800                 $pageInfo['revisionCount']++;
 801                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 802                         $pageInfo['successfulRevisionCount']++;
 803                 }
 804         }
 805
 806         /**
 807          * @param array $pageInfo
 808          * @param array $revisionInfo
 809          * @return bool|mixed
 810          */
 811         private function processRevision( $pageInfo, $revisionInfo ) {
 812                 $revision = new WikiRevision( $this->config );
 813
 814                 if ( isset( $revisionInfo['id'] ) ) {
 815                         $revision->setID( $revisionInfo['id'] );
 816                 }
 817                 if ( isset( $revisionInfo['model'] ) ) {
 818                         $revision->setModel( $revisionInfo['model'] );
 819                 }
 820                 if ( isset( $revisionInfo['format'] ) ) {
 821                         $revision->setFormat( $revisionInfo['format'] );
 822                 }
 823                 $revision->setTitle( $pageInfo['_title'] );
 824
 825                 if ( isset( $revisionInfo['text'] ) ) {
 826                         $handler = $revision->getContentHandler();
 827                         $text = $handler->importTransform(
 828                                 $revisionInfo['text'],
 829                                 $revision->getFormat() );
 830
 831                         $revision->setText( $text );
 832                 }
 833                 if ( isset( $revisionInfo['timestamp'] ) ) {
 834                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 835                 } else {
 836                         $revision->setTimestamp( wfTimestampNow() );
 837                 }
 838
 839                 if ( isset( $revisionInfo['comment'] ) ) {
 840                         $revision->setComment( $revisionInfo['comment'] );
 841                 }
 842
 843                 if ( isset( $revisionInfo['minor'] ) ) {
 844                         $revision->setMinor( true );
 845                 }
 846                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 847                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 848                 }
 849                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 850                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 851                 }
 852                 $revision->setNoUpdates( $this->mNoUpdates );
 853
 854                 return $this->revisionCallback( $revision );
 855         }
 856
 857         /**
 858          * @param array $pageInfo
 859          * @return mixed
 860          */
 861         private function handleUpload( &$pageInfo ) {
 862                 $this->debug( "Enter upload handler" );
 863                 $uploadInfo = array();
 864
 865                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 866                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 867
 868                 $skip = false;
 869
 870                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 871                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 872                                         $this->reader->localName == 'upload' ) {
 873                                 break;
 874                         }
 875
 876                         $tag = $this->reader->localName;
 877
 878                         if ( !Hooks::run( 'ImportHandleUploadXMLTag', array(
 879                                 $this, $pageInfo
 880                         ) ) ) {
 881                                 // Do nothing
 882                         } elseif ( in_array( $tag, $normalFields ) ) {
 883                                 $uploadInfo[$tag] = $this->nodeContents();
 884                         } elseif ( $tag == 'contributor' ) {
 885                                 $uploadInfo['contributor'] = $this->handleContributor();
 886                         } elseif ( $tag == 'contents' ) {
 887                                 $contents = $this->nodeContents();
 888                                 $encoding = $this->reader->getAttribute( 'encoding' );
 889                                 if ( $encoding === 'base64' ) {
 890                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 891                                         $uploadInfo['isTempSrc'] = true;
 892                                 }
 893                         } elseif ( $tag != '#text' ) {
 894                                 $this->warn( "Unhandled upload XML tag $tag" );
 895                                 $skip = true;
 896                         }
 897                 }
 898
 899                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 900                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 901                         if ( file_exists( $path ) ) {
 902                                 $uploadInfo['fileSrc'] = $path;
 903                                 $uploadInfo['isTempSrc'] = false;
 904                         }
 905                 }
 906
 907                 if ( $this->mImportUploads ) {
 908                         return $this->processUpload( $pageInfo, $uploadInfo );
 909                 }
 910         }
 911
 912         /**
 913          * @param string $contents
 914          * @return string
 915          */
 916         private function dumpTemp( $contents ) {
 917                 $filename = tempnam( wfTempDir(), 'importupload' );
 918                 file_put_contents( $filename, $contents );
 919                 return $filename;
 920         }
 921
 922         /**
 923          * @param array $pageInfo
 924          * @param array $uploadInfo
 925          * @return mixed
 926          */
 927         private function processUpload( $pageInfo, $uploadInfo ) {
 928                 $revision = new WikiRevision( $this->config );
 929                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 930
 931                 $revision->setTitle( $pageInfo['_title'] );
 932                 $revision->setID( $pageInfo['id'] );
 933                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 934                 $revision->setText( $text );
 935                 $revision->setFilename( $uploadInfo['filename'] );
 936                 if ( isset( $uploadInfo['archivename'] ) ) {
 937                         $revision->setArchiveName( $uploadInfo['archivename'] );
 938                 }
 939                 $revision->setSrc( $uploadInfo['src'] );
 940                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 941                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 942                                 !empty( $uploadInfo['isTempSrc'] ) );
 943                 }
 944                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 945                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 946                 }
 947                 $revision->setSize( intval( $uploadInfo['size'] ) );
 948                 $revision->setComment( $uploadInfo['comment'] );
 949
 950                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 951                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 952                 }
 953                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 954                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 955                 }
 956                 $revision->setNoUpdates( $this->mNoUpdates );
 957
 958                 return call_user_func( $this->mUploadCallback, $revision );
 959         }
 960
 961         /**
 962          * @return array
 963          */
 964         private function handleContributor() {
 965                 $fields = array( 'id', 'ip', 'username' );
 966                 $info = array();
 967
 968                 while ( $this->reader->read() ) {
 969                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 970                                         $this->reader->localName == 'contributor' ) {
 971                                 break;
 972                         }
 973
 974                         $tag = $this->reader->localName;
 975
 976                         if ( in_array( $tag, $fields ) ) {
 977                                 $info[$tag] = $this->nodeContents();
 978                         }
 979                 }
 980
 981                 return $info;
 982         }
 983
 984         /**
 985          * @param string $text
 986          * @param string|null $ns
 987          * @return array|bool
 988          */
 989         private function processTitle( $text, $ns = null ) {
 990                 if ( is_null( $this->foreignNamespaces ) ) {
 991                         $foreignTitleFactory = new NaiveForeignTitleFactory();
 992                 } else {
 993                         $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
 994                                 $this->foreignNamespaces );
 995                 }
 996
 997                 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
 998                         intval( $ns ) );
 999
1000                 $title = $this->importTitleFactory->createTitleFromForeignTitle(
1001                         $foreignTitle );
1002
1003                 $commandLineMode = $this->config->get( 'CommandLineMode' );
1004                 if ( is_null( $title ) ) {
1005                         # Invalid page title? Ignore the page
1006                         $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
1007                         return false;
1008                 } elseif ( $title->isExternal() ) {
1009                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
1010                         return false;
1011                 } elseif ( !$title->canExist() ) {
1012                         $this->notice( 'import-error-special', $title->getPrefixedText() );
1013                         return false;
1014                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
1015                         # Do not import if the importing wiki user cannot edit this page
1016                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
1017                         return false;
1018                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
1019                         # Do not import if the importing wiki user cannot create this page
1020                         $this->notice( 'import-error-create', $title->getPrefixedText() );
1021                         return false;
1022                 }
1023
1024                 return array( $title, $foreignTitle );
1025         }
1026 }
1027
1028 /** This is a horrible hack used to keep source compatibility */
1029 class UploadSourceAdapter {
1030         /** @var array */
1031         public static $sourceRegistrations = array();
1032
1033         /** @var string */
1034         private $mSource;
1035
1036         /** @var string */
1037         private $mBuffer;
1038
1039         /** @var int */
1040         private $mPosition;
1041
1042         /**
1043          * @param ImportSource $source
1044          * @return string
1045          */
1046         static function registerSource( ImportSource $source ) {
1047                 $id = wfRandomString();
1048
1049                 self::$sourceRegistrations[$id] = $source;
1050
1051                 return $id;
1052         }
1053
1054         /**
1055          * @param string $path
1056          * @param string $mode
1057          * @param array $options
1058          * @param string $opened_path
1059          * @return bool
1060          */
1061         function stream_open( $path, $mode, $options, &$opened_path ) {
1062                 $url = parse_url( $path );
1063                 $id = $url['host'];
1064
1065                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
1066                         return false;
1067                 }
1068
1069                 $this->mSource = self::$sourceRegistrations[$id];
1070
1071                 return true;
1072         }
1073
1074         /**
1075          * @param int $count
1076          * @return string
1077          */
1078         function stream_read( $count ) {
1079                 $return = '';
1080                 $leave = false;
1081
1082                 while ( !$leave && !$this->mSource->atEnd() &&
1083                                 strlen( $this->mBuffer ) < $count ) {
1084                         $read = $this->mSource->readChunk();
1085
1086                         if ( !strlen( $read ) ) {
1087                                 $leave = true;
1088                         }
1089
1090                         $this->mBuffer .= $read;
1091                 }
1092
1093                 if ( strlen( $this->mBuffer ) ) {
1094                         $return = substr( $this->mBuffer, 0, $count );
1095                         $this->mBuffer = substr( $this->mBuffer, $count );
1096                 }
1097
1098                 $this->mPosition += strlen( $return );
1099
1100                 return $return;
1101         }
1102
1103         /**
1104          * @param string $data
1105          * @return bool
1106          */
1107         function stream_write( $data ) {
1108                 return false;
1109         }
1110
1111         /**
1112          * @return mixed
1113          */
1114         function stream_tell() {
1115                 return $this->mPosition;
1116         }
1117
1118         /**
1119          * @return bool
1120          */
1121         function stream_eof() {
1122                 return $this->mSource->atEnd();
1123         }
1124
1125         /**
1126          * @return array
1127          */
1128         function url_stat() {
1129                 $result = array();
1130
1131                 $result['dev'] = $result[0] = 0;
1132                 $result['ino'] = $result[1] = 0;
1133                 $result['mode'] = $result[2] = 0;
1134                 $result['nlink'] = $result[3] = 0;
1135                 $result['uid'] = $result[4] = 0;
1136                 $result['gid'] = $result[5] = 0;
1137                 $result['rdev'] = $result[6] = 0;
1138                 $result['size'] = $result[7] = 0;
1139                 $result['atime'] = $result[8] = 0;
1140                 $result['mtime'] = $result[9] = 0;
1141                 $result['ctime'] = $result[10] = 0;
1142                 $result['blksize'] = $result[11] = 0;
1143                 $result['blocks'] = $result[12] = 0;
1144
1145                 return $result;
1146         }
1147 }
1148
1149 /**
1150  * @todo document (e.g. one-sentence class description).
1151  * @ingroup SpecialPage
1152  */
1153 class WikiRevision {
1154         /** @todo Unused? */
1155         public $importer = null;
1156
1157         /** @var Title */
1158         public $title = null;
1159
1160         /** @var int */
1161         public $id = 0;
1162
1163         /** @var string */
1164         public $timestamp = "20010115000000";
1165
1166         /**
1167          * @var int
1168          * @todo Can't find any uses. Public, because that's suspicious. Get clarity. */
1169         public $user = 0;
1170
1171         /** @var string */
1172         public $user_text = "";
1173
1174         /** @var string */
1175         public $model = null;
1176
1177         /** @var string */
1178         public $format = null;
1179
1180         /** @var string */
1181         public $text = "";
1182
1183         /** @var int */
1184         protected $size;
1185
1186         /** @var Content */
1187         public $content = null;
1188
1189         /** @var ContentHandler */
1190         protected $contentHandler = null;
1191
1192         /** @var string */
1193         public $comment = "";
1194
1195         /** @var bool */
1196         public $minor = false;
1197
1198         /** @var string */
1199         public $type = "";
1200
1201         /** @var string */
1202         public $action = "";
1203
1204         /** @var string */
1205         public $params = "";
1206
1207         /** @var string */
1208         public $fileSrc = '';
1209
1210         /** @var bool|string */
1211         public $sha1base36 = false;
1212
1213         /**
1214          * @var bool
1215          * @todo Unused?
1216          */
1217         public $isTemp = false;
1218
1219         /** @var string */
1220         public $archiveName = '';
1221
1222         protected $filename;
1223
1224         /** @var mixed */
1225         protected $src;
1226
1227         /** @todo Unused? */
1228         public $fileIsTemp;
1229
1230         /** @var bool */
1231         private $mNoUpdates = false;
1232
1233         /** @var Config $config */
1234         private $config;
1235
1236         public function __construct( Config $config ) {
1237                 $this->config = $config;
1238         }
1239
1240         /**
1241          * @param Title $title
1242          * @throws MWException
1243          */
1244         function setTitle( $title ) {
1245                 if ( is_object( $title ) ) {
1246                         $this->title = $title;
1247                 } elseif ( is_null( $title ) ) {
1248                         throw new MWException( "WikiRevision given a null title in import. "
1249                                 . "You may need to adjust \$wgLegalTitleChars." );
1250                 } else {
1251                         throw new MWException( "WikiRevision given non-object title in import." );
1252                 }
1253         }
1254
1255         /**
1256          * @param int $id
1257          */
1258         function setID( $id ) {
1259                 $this->id = $id;
1260         }
1261
1262         /**
1263          * @param string $ts
1264          */
1265         function setTimestamp( $ts ) {
1266                 # 2003-08-05T18:30:02Z
1267                 $this->timestamp = wfTimestamp( TS_MW, $ts );
1268         }
1269
1270         /**
1271          * @param string $user
1272          */
1273         function setUsername( $user ) {
1274                 $this->user_text = $user;
1275         }
1276
1277         /**
1278          * @param string $ip
1279          */
1280         function setUserIP( $ip ) {
1281                 $this->user_text = $ip;
1282         }
1283
1284         /**
1285          * @param string $model
1286          */
1287         function setModel( $model ) {
1288                 $this->model = $model;
1289         }
1290
1291         /**
1292          * @param string $format
1293          */
1294         function setFormat( $format ) {
1295                 $this->format = $format;
1296         }
1297
1298         /**
1299          * @param string $text
1300          */
1301         function setText( $text ) {
1302                 $this->text = $text;
1303         }
1304
1305         /**
1306          * @param string $text
1307          */
1308         function setComment( $text ) {
1309                 $this->comment = $text;
1310         }
1311
1312         /**
1313          * @param bool $minor
1314          */
1315         function setMinor( $minor ) {
1316                 $this->minor = (bool)$minor;
1317         }
1318
1319         /**
1320          * @param mixed $src
1321          */
1322         function setSrc( $src ) {
1323                 $this->src = $src;
1324         }
1325
1326         /**
1327          * @param string $src
1328          * @param bool $isTemp
1329          */
1330         function setFileSrc( $src, $isTemp ) {
1331                 $this->fileSrc = $src;
1332                 $this->fileIsTemp = $isTemp;
1333         }
1334
1335         /**
1336          * @param string $sha1base36
1337          */
1338         function setSha1Base36( $sha1base36 ) {
1339                 $this->sha1base36 = $sha1base36;
1340         }
1341
1342         /**
1343          * @param string $filename
1344          */
1345         function setFilename( $filename ) {
1346                 $this->filename = $filename;
1347         }
1348
1349         /**
1350          * @param string $archiveName
1351          */
1352         function setArchiveName( $archiveName ) {
1353                 $this->archiveName = $archiveName;
1354         }
1355
1356         /**
1357          * @param int $size
1358          */
1359         function setSize( $size ) {
1360                 $this->size = intval( $size );
1361         }
1362
1363         /**
1364          * @param string $type
1365          */
1366         function setType( $type ) {
1367                 $this->type = $type;
1368         }
1369
1370         /**
1371          * @param string $action
1372          */
1373         function setAction( $action ) {
1374                 $this->action = $action;
1375         }
1376
1377         /**
1378          * @param array $params
1379          */
1380         function setParams( $params ) {
1381                 $this->params = $params;
1382         }
1383
1384         /**
1385          * @param bool $noupdates
1386          */
1387         public function setNoUpdates( $noupdates ) {
1388                 $this->mNoUpdates = $noupdates;
1389         }
1390
1391         /**
1392          * @return Title
1393          */
1394         function getTitle() {
1395                 return $this->title;
1396         }
1397
1398         /**
1399          * @return int
1400          */
1401         function getID() {
1402                 return $this->id;
1403         }
1404
1405         /**
1406          * @return string
1407          */
1408         function getTimestamp() {
1409                 return $this->timestamp;
1410         }
1411
1412         /**
1413          * @return string
1414          */
1415         function getUser() {
1416                 return $this->user_text;
1417         }
1418
1419         /**
1420          * @return string
1421          *
1422          * @deprecated Since 1.21, use getContent() instead.
1423          */
1424         function getText() {
1425                 ContentHandler::deprecated( __METHOD__, '1.21' );
1426
1427                 return $this->text;
1428         }
1429
1430         /**
1431          * @return ContentHandler
1432          */
1433         function getContentHandler() {
1434                 if ( is_null( $this->contentHandler ) ) {
1435                         $this->contentHandler = ContentHandler::getForModelID( $this->getModel() );
1436                 }
1437
1438                 return $this->contentHandler;
1439         }
1440
1441         /**
1442          * @return Content
1443          */
1444         function getContent() {
1445                 if ( is_null( $this->content ) ) {
1446                         $handler = $this->getContentHandler();
1447                         $this->content = $handler->unserializeContent( $this->text, $this->getFormat() );
1448                 }
1449
1450                 return $this->content;
1451         }
1452
1453         /**
1454          * @return string
1455          */
1456         function getModel() {
1457                 if ( is_null( $this->model ) ) {
1458                         $this->model = $this->getTitle()->getContentModel();
1459                 }
1460
1461                 return $this->model;
1462         }
1463
1464         /**
1465          * @return string
1466          */
1467         function getFormat() {
1468                 if ( is_null( $this->format ) ) {
1469                         $this->format = $this->getContentHandler()->getDefaultFormat();
1470                 }
1471
1472                 return $this->format;
1473         }
1474
1475         /**
1476          * @return string
1477          */
1478         function getComment() {
1479                 return $this->comment;
1480         }
1481
1482         /**
1483          * @return bool
1484          */
1485         function getMinor() {
1486                 return $this->minor;
1487         }
1488
1489         /**
1490          * @return mixed
1491          */
1492         function getSrc() {
1493                 return $this->src;
1494         }
1495
1496         /**
1497          * @return bool|string
1498          */
1499         function getSha1() {
1500                 if ( $this->sha1base36 ) {
1501                         return wfBaseConvert( $this->sha1base36, 36, 16 );
1502                 }
1503                 return false;
1504         }
1505
1506         /**
1507          * @return string
1508          */
1509         function getFileSrc() {
1510                 return $this->fileSrc;
1511         }
1512
1513         /**
1514          * @return bool
1515          */
1516         function isTempSrc() {
1517                 return $this->isTemp;
1518         }
1519
1520         /**
1521          * @return mixed
1522          */
1523         function getFilename() {
1524                 return $this->filename;
1525         }
1526
1527         /**
1528          * @return string
1529          */
1530         function getArchiveName() {
1531                 return $this->archiveName;
1532         }
1533
1534         /**
1535          * @return mixed
1536          */
1537         function getSize() {
1538                 return $this->size;
1539         }
1540
1541         /**
1542          * @return string
1543          */
1544         function getType() {
1545                 return $this->type;
1546         }
1547
1548         /**
1549          * @return string
1550          */
1551         function getAction() {
1552                 return $this->action;
1553         }
1554
1555         /**
1556          * @return string
1557          */
1558         function getParams() {
1559                 return $this->params;
1560         }
1561
1562         /**
1563          * @return bool
1564          */
1565         function importOldRevision() {
1566                 $dbw = wfGetDB( DB_MASTER );
1567
1568                 # Sneak a single revision into place
1569                 $user = User::newFromName( $this->getUser() );
1570                 if ( $user ) {
1571                         $userId = intval( $user->getId() );
1572                         $userText = $user->getName();
1573                         $userObj = $user;
1574                 } else {
1575                         $userId = 0;
1576                         $userText = $this->getUser();
1577                         $userObj = new User;
1578                 }
1579
1580                 // avoid memory leak...?
1581                 Title::clearCaches();
1582
1583                 $page = WikiPage::factory( $this->title );
1584                 $page->loadPageData( 'fromdbmaster' );
1585                 if ( !$page->exists() ) {
1586                         # must create the page...
1587                         $pageId = $page->insertOn( $dbw );
1588                         $created = true;
1589                         $oldcountable = null;
1590                 } else {
1591                         $pageId = $page->getId();
1592                         $created = false;
1593
1594                         $prior = $dbw->selectField( 'revision', '1',
1595                                 array( 'rev_page' => $pageId,
1596                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1597                                         'rev_user_text' => $userText,
1598                                         'rev_comment' => $this->getComment() ),
1599                                 __METHOD__
1600                         );
1601                         if ( $prior ) {
1602                                 // @todo FIXME: This could fail slightly for multiple matches :P
1603                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1604                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1605                                 return false;
1606                         }
1607                 }
1608
1609                 # @todo FIXME: Use original rev_id optionally (better for backups)
1610                 # Insert the row
1611                 $revision = new Revision( array(
1612                         'title' => $this->title,
1613                         'page' => $pageId,
1614                         'content_model' => $this->getModel(),
1615                         'content_format' => $this->getFormat(),
1616                         // XXX: just set 'content' => $this->getContent()?
1617                         'text' => $this->getContent()->serialize( $this->getFormat() ),
1618                         'comment' => $this->getComment(),
1619                         'user' => $userId,
1620                         'user_text' => $userText,
1621                         'timestamp' => $this->timestamp,
1622                         'minor_edit' => $this->minor,
1623                         ) );
1624                 $revision->insertOn( $dbw );
1625                 $changed = $page->updateIfNewerOn( $dbw, $revision );
1626
1627                 if ( $changed !== false && !$this->mNoUpdates ) {
1628                         wfDebug( __METHOD__ . ": running updates\n" );
1629                         // countable/oldcountable stuff is handled in WikiImporter::finishImportPage
1630                         $page->doEditUpdates(
1631                                 $revision,
1632                                 $userObj,
1633                                 array( 'created' => $created, 'oldcountable' => 'no-change' )
1634                         );
1635                 }
1636
1637                 return true;
1638         }
1639
1640         function importLogItem() {
1641                 $dbw = wfGetDB( DB_MASTER );
1642                 # @todo FIXME: This will not record autoblocks
1643                 if ( !$this->getTitle() ) {
1644                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1645                                 $this->timestamp . "\n" );
1646                         return;
1647                 }
1648                 # Check if it exists already
1649                 // @todo FIXME: Use original log ID (better for backups)
1650                 $prior = $dbw->selectField( 'logging', '1',
1651                         array( 'log_type' => $this->getType(),
1652                                 'log_action' => $this->getAction(),
1653                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1654                                 'log_namespace' => $this->getTitle()->getNamespace(),
1655                                 'log_title' => $this->getTitle()->getDBkey(),
1656                                 'log_comment' => $this->getComment(),
1657                                 # 'log_user_text' => $this->user_text,
1658                                 'log_params' => $this->params ),
1659                         __METHOD__
1660                 );
1661                 // @todo FIXME: This could fail slightly for multiple matches :P
1662                 if ( $prior ) {
1663                         wfDebug( __METHOD__
1664                                 . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp "
1665                                 . $this->timestamp . "\n" );
1666                         return;
1667                 }
1668                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1669                 $data = array(
1670                         'log_id' => $log_id,
1671                         'log_type' => $this->type,
1672                         'log_action' => $this->action,
1673                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1674                         'log_user' => User::idFromName( $this->user_text ),
1675                         # 'log_user_text' => $this->user_text,
1676                         'log_namespace' => $this->getTitle()->getNamespace(),
1677                         'log_title' => $this->getTitle()->getDBkey(),
1678                         'log_comment' => $this->getComment(),
1679                         'log_params' => $this->params
1680                 );
1681                 $dbw->insert( 'logging', $data, __METHOD__ );
1682         }
1683
1684         /**
1685          * @return bool
1686          */
1687         function importUpload() {
1688                 # Construct a file
1689                 $archiveName = $this->getArchiveName();
1690                 if ( $archiveName ) {
1691                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1692                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1693                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1694                 } else {
1695                         $file = wfLocalFile( $this->getTitle() );
1696                         $file->load( File::READ_LATEST );
1697                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1698                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1699                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1700                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1701                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1702                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1703                         }
1704                 }
1705                 if ( !$file ) {
1706                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1707                         return false;
1708                 }
1709
1710                 # Get the file source or download if necessary
1711                 $source = $this->getFileSrc();
1712                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1713                 if ( !$source ) {
1714                         $source = $this->downloadSource();
1715                         $flags |= File::DELETE_SOURCE;
1716                 }
1717                 if ( !$source ) {
1718                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1719                         return false;
1720                 }
1721                 $sha1 = $this->getSha1();
1722                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1723                         if ( $flags & File::DELETE_SOURCE ) {
1724                                 # Broken file; delete it if it is a temporary file
1725                                 unlink( $source );
1726                         }
1727                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1728                         return false;
1729                 }
1730
1731                 $user = User::newFromName( $this->user_text );
1732
1733                 # Do the actual upload
1734                 if ( $archiveName ) {
1735                         $status = $file->uploadOld( $source, $archiveName,
1736                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1737                 } else {
1738                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1739                                 $flags, false, $this->getTimestamp(), $user );
1740                 }
1741
1742                 if ( $status->isGood() ) {
1743                         wfDebug( __METHOD__ . ": Successful\n" );
1744                         return true;
1745                 } else {
1746                         wfDebug( __METHOD__ . ': failed: ' . $status->getHTML() . "\n" );
1747                         return false;
1748                 }
1749         }
1750
1751         /**
1752          * @return bool|string
1753          */
1754         function downloadSource() {
1755                 if ( !$this->config->get( 'EnableUploads' ) ) {
1756                         return false;
1757                 }
1758
1759                 $tempo = tempnam( wfTempDir(), 'download' );
1760                 $f = fopen( $tempo, 'wb' );
1761                 if ( !$f ) {
1762                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1763                         return false;
1764                 }
1765
1766                 // @todo FIXME!
1767                 $src = $this->getSrc();
1768                 $data = Http::get( $src, array(), __METHOD__ );
1769                 if ( !$data ) {
1770                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1771                         fclose( $f );
1772                         unlink( $tempo );
1773                         return false;
1774                 }
1775
1776                 fwrite( $f, $data );
1777                 fclose( $f );
1778
1779                 return $tempo;
1780         }
1781
1782 }
1783
1784 /**
1785  * Source interface for XML import.
1786  */
1787 interface ImportSource {
1788
1789         /**
1790          * Indicates whether the end of the input has been reached.
1791          * Will return true after a finite number of calls to readChunk.
1792          *
1793          * @return bool true if there is no more input, false otherwise.
1794          */
1795         function atEnd();
1796
1797         /**
1798          * Return a chunk of the input, as a (possibly empty) string.
1799          * When the end of input is reached, readChunk() returns false.
1800          * If atEnd() returns false, readChunk() will return a string.
1801          * If atEnd() returns true, readChunk() will return false.
1802          *
1803          * @return bool|string
1804          */
1805         function readChunk();
1806 }
1807
1808 /**
1809  * Used for importing XML dumps where the content of the dump is in a string.
1810  * This class is ineffecient, and should only be used for small dumps.
1811  * For larger dumps, ImportStreamSource should be used instead.
1812  *
1813  * @ingroup SpecialPage
1814  */
1815 class ImportStringSource implements ImportSource {
1816         function __construct( $string ) {
1817                 $this->mString = $string;
1818                 $this->mRead = false;
1819         }
1820
1821         /**
1822          * @return bool
1823          */
1824         function atEnd() {
1825                 return $this->mRead;
1826         }
1827
1828         /**
1829          * @return bool|string
1830          */
1831         function readChunk() {
1832                 if ( $this->atEnd() ) {
1833                         return false;
1834                 }
1835                 $this->mRead = true;
1836                 return $this->mString;
1837         }
1838 }
1839
1840 /**
1841  * Imports a XML dump from a file (either from file upload, files on disk, or HTTP)
1842  * @ingroup SpecialPage
1843  */
1844 class ImportStreamSource implements ImportSource {
1845         function __construct( $handle ) {
1846                 $this->mHandle = $handle;
1847         }
1848
1849         /**
1850          * @return bool
1851          */
1852         function atEnd() {
1853                 return feof( $this->mHandle );
1854         }
1855
1856         /**
1857          * @return string
1858          */
1859         function readChunk() {
1860                 return fread( $this->mHandle, 32768 );
1861         }
1862
1863         /**
1864          * @param string $filename
1865          * @return Status
1866          */
1867         static function newFromFile( $filename ) {
1868                 MediaWiki\suppressWarnings();
1869                 $file = fopen( $filename, 'rt' );
1870                 MediaWiki\restoreWarnings();
1871                 if ( !$file ) {
1872                         return Status::newFatal( "importcantopen" );
1873                 }
1874                 return Status::newGood( new ImportStreamSource( $file ) );
1875         }
1876
1877         /**
1878          * @param string $fieldname
1879          * @return Status
1880          */
1881         static function newFromUpload( $fieldname = "xmlimport" ) {
1882                 $upload =& $_FILES[$fieldname];
1883
1884                 if ( $upload === null || !$upload['name'] ) {
1885                         return Status::newFatal( 'importnofile' );
1886                 }
1887                 if ( !empty( $upload['error'] ) ) {
1888                         switch ( $upload['error'] ) {
1889                                 case 1:
1890                                         # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1891                                         return Status::newFatal( 'importuploaderrorsize' );
1892                                 case 2:
1893                                         # The uploaded file exceeds the MAX_FILE_SIZE directive that
1894                                         # was specified in the HTML form.
1895                                         return Status::newFatal( 'importuploaderrorsize' );
1896                                 case 3:
1897                                         # The uploaded file was only partially uploaded
1898                                         return Status::newFatal( 'importuploaderrorpartial' );
1899                                 case 6:
1900                                         # Missing a temporary folder.
1901                                         return Status::newFatal( 'importuploaderrortemp' );
1902                                 # case else: # Currently impossible
1903                         }
1904
1905                 }
1906                 $fname = $upload['tmp_name'];
1907                 if ( is_uploaded_file( $fname ) ) {
1908                         return ImportStreamSource::newFromFile( $fname );
1909                 } else {
1910                         return Status::newFatal( 'importnofile' );
1911                 }
1912         }
1913
1914         /**
1915          * @param string $url
1916          * @param string $method
1917          * @return Status
1918          */
1919         static function newFromURL( $url, $method = 'GET' ) {
1920                 wfDebug( __METHOD__ . ": opening $url\n" );
1921                 # Use the standard HTTP fetch function; it times out
1922                 # quicker and sorts out user-agent problems which might
1923                 # otherwise prevent importing from large sites, such
1924                 # as the Wikimedia cluster, etc.
1925                 $data = Http::request( $method, $url, array( 'followRedirects' => true ), __METHOD__ );
1926                 if ( $data !== false ) {
1927                         $file = tmpfile();
1928                         fwrite( $file, $data );
1929                         fflush( $file );
1930                         fseek( $file, 0 );
1931                         return Status::newGood( new ImportStreamSource( $file ) );
1932                 } else {
1933                         return Status::newFatal( 'importcantopen' );
1934                 }
1935         }
1936
1937         /**
1938          * @param string $interwiki
1939          * @param string $page
1940          * @param bool $history
1941          * @param bool $templates
1942          * @param int $pageLinkDepth
1943          * @return Status
1944          */
1945         public static function newFromInterwiki( $interwiki, $page, $history = false,
1946                 $templates = false, $pageLinkDepth = 0
1947         ) {
1948                 if ( $page == '' ) {
1949                         return Status::newFatal( 'import-noarticle' );
1950                 }
1951                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1952                 if ( is_null( $link ) || !$link->isExternal() ) {
1953                         return Status::newFatal( 'importbadinterwiki' );
1954                 } else {
1955                         $params = array();
1956                         if ( $history ) {
1957                                 $params['history'] = 1;
1958                         }
1959                         if ( $templates ) {
1960                                 $params['templates'] = 1;
1961                         }
1962                         if ( $pageLinkDepth ) {
1963                                 $params['pagelink-depth'] = $pageLinkDepth;
1964                         }
1965                         $url = $link->getFullURL( $params );
1966                         # For interwikis, use POST to avoid redirects.
1967                         return ImportStreamSource::newFromURL( $url, "POST" );
1968                 }
1969         }
1970 }