includes/import/WikiImporter.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 use MediaWiki\MediaWikiServices;
  27
  28 /**
  29  * XML file reader for the page data importer.
  30  *
  31  * implements Special:Import
  32  * @ingroup SpecialPage
  33  */
  34 class WikiImporter {
  35         private $reader = null;
  36         private $foreignNamespaces = null;
  37         private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
  38         private $mSiteInfoCallback, $mPageOutCallback;
  39         private $mNoticeCallback, $mDebug;
  40         private $mImportUploads, $mImageBasePath;
  41         private $mNoUpdates = false;
  42         /** @var Config */
  43         private $config;
  44         /** @var ImportTitleFactory */
  45         private $importTitleFactory;
  46         /** @var array */
  47         private $countableCache = [];
  48         /** @var bool */
  49         private $disableStatisticsUpdate = false;
  50
  51         /**
  52          * Creates an ImportXMLReader drawing from the source provided
  53          * @param ImportSource $source
  54          * @param Config $config
  55          * @throws Exception
  56          */
  57         function __construct( ImportSource $source, Config $config = null ) {
  58                 if ( !class_exists( 'XMLReader' ) ) {
  59                         throw new Exception( 'Import requires PHP to have been compiled with libxml support' );
  60                 }
  61
  62                 $this->reader = new XMLReader();
  63                 if ( !$config ) {
  64                         wfDeprecated( __METHOD__ . ' without a Config instance', '1.25' );
  65                         $config = MediaWikiServices::getInstance()->getMainConfig();
  66                 }
  67                 $this->config = $config;
  68
  69                 if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) {
  70                         stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  71                 }
  72                 $id = UploadSourceAdapter::registerSource( $source );
  73
  74                 // Enable the entity loader, as it is needed for loading external URLs via
  75                 // XMLReader::open (T86036)
  76                 $oldDisable = libxml_disable_entity_loader( false );
  77                 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
  78                         $status = $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  79                 } else {
  80                         $status = $this->reader->open( "uploadsource://$id" );
  81                 }
  82                 if ( !$status ) {
  83                         $error = libxml_get_last_error();
  84                         libxml_disable_entity_loader( $oldDisable );
  85                         throw new MWException( 'Encountered an internal error while initializing WikiImporter object: ' .
  86                                 $error->message );
  87                 }
  88                 libxml_disable_entity_loader( $oldDisable );
  89
  90                 // Default callbacks
  91                 $this->setPageCallback( [ $this, 'beforeImportPage' ] );
  92                 $this->setRevisionCallback( [ $this, "importRevision" ] );
  93                 $this->setUploadCallback( [ $this, 'importUpload' ] );
  94                 $this->setLogItemCallback( [ $this, 'importLogItem' ] );
  95                 $this->setPageOutCallback( [ $this, 'finishImportPage' ] );
  96
  97                 $this->importTitleFactory = new NaiveImportTitleFactory();
  98         }
  99
 100         /**
 101          * @return null|XMLReader
 102          */
 103         public function getReader() {
 104                 return $this->reader;
 105         }
 106
 107         public function throwXmlError( $err ) {
 108                 $this->debug( "FAILURE: $err" );
 109                 wfDebug( "WikiImporter XML error: $err\n" );
 110         }
 111
 112         public function debug( $data ) {
 113                 if ( $this->mDebug ) {
 114                         wfDebug( "IMPORT: $data\n" );
 115                 }
 116         }
 117
 118         public function warn( $data ) {
 119                 wfDebug( "IMPORT: $data\n" );
 120         }
 121
 122         public function notice( $msg /*, $param, ...*/ ) {
 123                 $params = func_get_args();
 124                 array_shift( $params );
 125
 126                 if ( is_callable( $this->mNoticeCallback ) ) {
 127                         call_user_func( $this->mNoticeCallback, $msg, $params );
 128                 } else { # No ImportReporter -> CLI
 129                         echo wfMessage( $msg, $params )->text() . "\n";
 130                 }
 131         }
 132
 133         /**
 134          * Set debug mode...
 135          * @param bool $debug
 136          */
 137         function setDebug( $debug ) {
 138                 $this->mDebug = $debug;
 139         }
 140
 141         /**
 142          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
 143          * @param bool $noupdates
 144          */
 145         function setNoUpdates( $noupdates ) {
 146                 $this->mNoUpdates = $noupdates;
 147         }
 148
 149         /**
 150          * Set a callback that displays notice messages
 151          *
 152          * @param callable $callback
 153          * @return callable
 154          */
 155         public function setNoticeCallback( $callback ) {
 156                 return wfSetVar( $this->mNoticeCallback, $callback );
 157         }
 158
 159         /**
 160          * Sets the action to perform as each new page in the stream is reached.
 161          * @param callable $callback
 162          * @return callable
 163          */
 164         public function setPageCallback( $callback ) {
 165                 $previous = $this->mPageCallback;
 166                 $this->mPageCallback = $callback;
 167                 return $previous;
 168         }
 169
 170         /**
 171          * Sets the action to perform as each page in the stream is completed.
 172          * Callback accepts the page title (as a Title object), a second object
 173          * with the original title form (in case it's been overridden into a
 174          * local namespace), and a count of revisions.
 175          *
 176          * @param callable $callback
 177          * @return callable
 178          */
 179         public function setPageOutCallback( $callback ) {
 180                 $previous = $this->mPageOutCallback;
 181                 $this->mPageOutCallback = $callback;
 182                 return $previous;
 183         }
 184
 185         /**
 186          * Sets the action to perform as each page revision is reached.
 187          * @param callable $callback
 188          * @return callable
 189          */
 190         public function setRevisionCallback( $callback ) {
 191                 $previous = $this->mRevisionCallback;
 192                 $this->mRevisionCallback = $callback;
 193                 return $previous;
 194         }
 195
 196         /**
 197          * Sets the action to perform as each file upload version is reached.
 198          * @param callable $callback
 199          * @return callable
 200          */
 201         public function setUploadCallback( $callback ) {
 202                 $previous = $this->mUploadCallback;
 203                 $this->mUploadCallback = $callback;
 204                 return $previous;
 205         }
 206
 207         /**
 208          * Sets the action to perform as each log item reached.
 209          * @param callable $callback
 210          * @return callable
 211          */
 212         public function setLogItemCallback( $callback ) {
 213                 $previous = $this->mLogItemCallback;
 214                 $this->mLogItemCallback = $callback;
 215                 return $previous;
 216         }
 217
 218         /**
 219          * Sets the action to perform when site info is encountered
 220          * @param callable $callback
 221          * @return callable
 222          */
 223         public function setSiteInfoCallback( $callback ) {
 224                 $previous = $this->mSiteInfoCallback;
 225                 $this->mSiteInfoCallback = $callback;
 226                 return $previous;
 227         }
 228
 229         /**
 230          * Sets the factory object to use to convert ForeignTitle objects into local
 231          * Title objects
 232          * @param ImportTitleFactory $factory
 233          */
 234         public function setImportTitleFactory( $factory ) {
 235                 $this->importTitleFactory = $factory;
 236         }
 237
 238         /**
 239          * Set a target namespace to override the defaults
 240          * @param null|int $namespace
 241          * @return bool
 242          */
 243         public function setTargetNamespace( $namespace ) {
 244                 if ( is_null( $namespace ) ) {
 245                         // Don't override namespaces
 246                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 247                         return true;
 248                 } elseif (
 249                         $namespace >= 0 &&
 250                         MWNamespace::exists( intval( $namespace ) )
 251                 ) {
 252                         $namespace = intval( $namespace );
 253                         $this->setImportTitleFactory( new NamespaceImportTitleFactory( $namespace ) );
 254                         return true;
 255                 } else {
 256                         return false;
 257                 }
 258         }
 259
 260         /**
 261          * Set a target root page under which all pages are imported
 262          * @param null|string $rootpage
 263          * @return Status
 264          */
 265         public function setTargetRootPage( $rootpage ) {
 266                 $status = Status::newGood();
 267                 if ( is_null( $rootpage ) ) {
 268                         // No rootpage
 269                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 270                 } elseif ( $rootpage !== '' ) {
 271                         $rootpage = rtrim( $rootpage, '/' ); // avoid double slashes
 272                         $title = Title::newFromText( $rootpage );
 273
 274                         if ( !$title || $title->isExternal() ) {
 275                                 $status->fatal( 'import-rootpage-invalid' );
 276                         } else {
 277                                 if ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 278                                         global $wgContLang;
 279
 280                                         $displayNSText = $title->getNamespace() == NS_MAIN
 281                                                 ? wfMessage( 'blanknamespace' )->text()
 282                                                 : $wgContLang->getNsText( $title->getNamespace() );
 283                                         $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 284                                 } else {
 285                                         // set namespace to 'all', so the namespace check in processTitle() can pass
 286                                         $this->setTargetNamespace( null );
 287                                         $this->setImportTitleFactory( new SubpageImportTitleFactory( $title ) );
 288                                 }
 289                         }
 290                 }
 291                 return $status;
 292         }
 293
 294         /**
 295          * @param string $dir
 296          */
 297         public function setImageBasePath( $dir ) {
 298                 $this->mImageBasePath = $dir;
 299         }
 300
 301         /**
 302          * @param bool $import
 303          */
 304         public function setImportUploads( $import ) {
 305                 $this->mImportUploads = $import;
 306         }
 307
 308         /**
 309          * Statistics update can cause a lot of time
 310          * @since 1.29
 311          */
 312         public function disableStatisticsUpdate() {
 313                 $this->disableStatisticsUpdate = true;
 314         }
 315
 316         /**
 317          * Default per-page callback. Sets up some things related to site statistics
 318          * @param array $titleAndForeignTitle Two-element array, with Title object at
 319          * index 0 and ForeignTitle object at index 1
 320          * @return bool
 321          */
 322         public function beforeImportPage( $titleAndForeignTitle ) {
 323                 $title = $titleAndForeignTitle[0];
 324                 $page = WikiPage::factory( $title );
 325                 $this->countableCache['title_' . $title->getPrefixedText()] = $page->isCountable();
 326                 return true;
 327         }
 328
 329         /**
 330          * Default per-revision callback, performs the import.
 331          * @param WikiRevision $revision
 332          * @return bool
 333          */
 334         public function importRevision( $revision ) {
 335                 if ( !$revision->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
 336                         $this->notice( 'import-error-bad-location',
 337                                 $revision->getTitle()->getPrefixedText(),
 338                                 $revision->getID(),
 339                                 $revision->getModel(),
 340                                 $revision->getFormat() );
 341
 342                         return false;
 343                 }
 344
 345                 try {
 346                         return $revision->importOldRevision();
 347                 } catch ( MWContentSerializationException $ex ) {
 348                         $this->notice( 'import-error-unserialize',
 349                                 $revision->getTitle()->getPrefixedText(),
 350                                 $revision->getID(),
 351                                 $revision->getModel(),
 352                                 $revision->getFormat() );
 353                 }
 354
 355                 return false;
 356         }
 357
 358         /**
 359          * Default per-revision callback, performs the import.
 360          * @param WikiRevision $revision
 361          * @return bool
 362          */
 363         public function importLogItem( $revision ) {
 364                 return $revision->importLogItem();
 365         }
 366
 367         /**
 368          * Dummy for now...
 369          * @param WikiRevision $revision
 370          * @return bool
 371          */
 372         public function importUpload( $revision ) {
 373                 return $revision->importUpload();
 374         }
 375
 376         /**
 377          * Mostly for hook use
 378          * @param Title $title
 379          * @param ForeignTitle $foreignTitle
 380          * @param int $revCount
 381          * @param int $sRevCount
 382          * @param array $pageInfo
 383          * @return bool
 384          */
 385         public function finishImportPage( $title, $foreignTitle, $revCount,
 386                         $sRevCount, $pageInfo ) {
 387
 388                 // Update article count statistics (T42009)
 389                 // The normal counting logic in WikiPage->doEditUpdates() is designed for
 390                 // one-revision-at-a-time editing, not bulk imports. In this situation it
 391                 // suffers from issues of replica DB lag. We let WikiPage handle the total page
 392                 // and revision count, and we implement our own custom logic for the
 393                 // article (content page) count.
 394                 if ( !$this->disableStatisticsUpdate ) {
 395                         $page = WikiPage::factory( $title );
 396                         $page->loadPageData( 'fromdbmaster' );
 397                         $content = $page->getContent();
 398                         if ( $content === null ) {
 399                                 wfDebug( __METHOD__ . ': Skipping article count adjustment for ' . $title .
 400                                         ' because WikiPage::getContent() returned null' );
 401                         } else {
 402                                 $editInfo = $page->prepareContentForEdit( $content );
 403                                 $countKey = 'title_' . $title->getPrefixedText();
 404                                 $countable = $page->isCountable( $editInfo );
 405                                 if ( array_key_exists( $countKey, $this->countableCache ) &&
 406                                         $countable != $this->countableCache[$countKey] ) {
 407                                         DeferredUpdates::addUpdate( SiteStatsUpdate::factory( [
 408                                                 'articles' => ( (int)$countable - (int)$this->countableCache[$countKey] )
 409                                         ] ) );
 410                                 }
 411                         }
 412                 }
 413
 414                 $args = func_get_args();
 415                 return Hooks::run( 'AfterImportPage', $args );
 416         }
 417
 418         /**
 419          * Alternate per-revision callback, for debugging.
 420          * @param WikiRevision $revision
 421          */
 422         public function debugRevisionHandler( &$revision ) {
 423                 $this->debug( "Got revision:" );
 424                 if ( is_object( $revision->title ) ) {
 425                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 426                 } else {
 427                         $this->debug( "-- Title: <invalid>" );
 428                 }
 429                 $this->debug( "-- User: " . $revision->user_text );
 430                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 431                 $this->debug( "-- Comment: " . $revision->comment );
 432                 $this->debug( "-- Text: " . $revision->text );
 433         }
 434
 435         /**
 436          * Notify the callback function of site info
 437          * @param array $siteInfo
 438          * @return bool|mixed
 439          */
 440         private function siteInfoCallback( $siteInfo ) {
 441                 if ( isset( $this->mSiteInfoCallback ) ) {
 442                         return call_user_func_array( $this->mSiteInfoCallback,
 443                                         [ $siteInfo, $this ] );
 444                 } else {
 445                         return false;
 446                 }
 447         }
 448
 449         /**
 450          * Notify the callback function when a new "<page>" is reached.
 451          * @param Title $title
 452          */
 453         function pageCallback( $title ) {
 454                 if ( isset( $this->mPageCallback ) ) {
 455                         call_user_func( $this->mPageCallback, $title );
 456                 }
 457         }
 458
 459         /**
 460          * Notify the callback function when a "</page>" is closed.
 461          * @param Title $title
 462          * @param ForeignTitle $foreignTitle
 463          * @param int $revCount
 464          * @param int $sucCount Number of revisions for which callback returned true
 465          * @param array $pageInfo Associative array of page information
 466          */
 467         private function pageOutCallback( $title, $foreignTitle, $revCount,
 468                         $sucCount, $pageInfo ) {
 469                 if ( isset( $this->mPageOutCallback ) ) {
 470                         $args = func_get_args();
 471                         call_user_func_array( $this->mPageOutCallback, $args );
 472                 }
 473         }
 474
 475         /**
 476          * Notify the callback function of a revision
 477          * @param WikiRevision $revision
 478          * @return bool|mixed
 479          */
 480         private function revisionCallback( $revision ) {
 481                 if ( isset( $this->mRevisionCallback ) ) {
 482                         return call_user_func_array( $this->mRevisionCallback,
 483                                         [ $revision, $this ] );
 484                 } else {
 485                         return false;
 486                 }
 487         }
 488
 489         /**
 490          * Notify the callback function of a new log item
 491          * @param WikiRevision $revision
 492          * @return bool|mixed
 493          */
 494         private function logItemCallback( $revision ) {
 495                 if ( isset( $this->mLogItemCallback ) ) {
 496                         return call_user_func_array( $this->mLogItemCallback,
 497                                         [ $revision, $this ] );
 498                 } else {
 499                         return false;
 500                 }
 501         }
 502
 503         /**
 504          * Retrieves the contents of the named attribute of the current element.
 505          * @param string $attr The name of the attribute
 506          * @return string The value of the attribute or an empty string if it is not set in the current
 507          * element.
 508          */
 509         public function nodeAttribute( $attr ) {
 510                 return $this->reader->getAttribute( $attr );
 511         }
 512
 513         /**
 514          * Shouldn't something like this be built-in to XMLReader?
 515          * Fetches text contents of the current element, assuming
 516          * no sub-elements or such scary things.
 517          * @return string
 518          * @access private
 519          */
 520         public function nodeContents() {
 521                 if ( $this->reader->isEmptyElement ) {
 522                         return "";
 523                 }
 524                 $buffer = "";
 525                 while ( $this->reader->read() ) {
 526                         switch ( $this->reader->nodeType ) {
 527                         case XMLReader::TEXT:
 528                         case XMLReader::CDATA:
 529                         case XMLReader::SIGNIFICANT_WHITESPACE:
 530                                 $buffer .= $this->reader->value;
 531                                 break;
 532                         case XMLReader::END_ELEMENT:
 533                                 return $buffer;
 534                         }
 535                 }
 536
 537                 $this->reader->close();
 538                 return '';
 539         }
 540
 541         /**
 542          * Primary entry point
 543          * @throws MWException
 544          * @return bool
 545          */
 546         public function doImport() {
 547                 // Calls to reader->read need to be wrapped in calls to
 548                 // libxml_disable_entity_loader() to avoid local file
 549                 // inclusion attacks (T48932).
 550                 $oldDisable = libxml_disable_entity_loader( true );
 551                 $this->reader->read();
 552
 553                 if ( $this->reader->localName != 'mediawiki' ) {
 554                         libxml_disable_entity_loader( $oldDisable );
 555                         throw new MWException( "Expected <mediawiki> tag, got " .
 556                                 $this->reader->localName );
 557                 }
 558                 $this->debug( "<mediawiki> tag is correct." );
 559
 560                 $this->debug( "Starting primary dump processing loop." );
 561
 562                 $keepReading = $this->reader->read();
 563                 $skip = false;
 564                 $rethrow = null;
 565                 try {
 566                         while ( $keepReading ) {
 567                                 $tag = $this->reader->localName;
 568                                 $type = $this->reader->nodeType;
 569
 570                                 if ( !Hooks::run( 'ImportHandleToplevelXMLTag', [ $this ] ) ) {
 571                                         // Do nothing
 572                                 } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) {
 573                                         break;
 574                                 } elseif ( $tag == 'siteinfo' ) {
 575                                         $this->handleSiteInfo();
 576                                 } elseif ( $tag == 'page' ) {
 577                                         $this->handlePage();
 578                                 } elseif ( $tag == 'logitem' ) {
 579                                         $this->handleLogItem();
 580                                 } elseif ( $tag != '#text' ) {
 581                                         $this->warn( "Unhandled top-level XML tag $tag" );
 582
 583                                         $skip = true;
 584                                 }
 585
 586                                 if ( $skip ) {
 587                                         $keepReading = $this->reader->next();
 588                                         $skip = false;
 589                                         $this->debug( "Skip" );
 590                                 } else {
 591                                         $keepReading = $this->reader->read();
 592                                 }
 593                         }
 594                 } catch ( Exception $ex ) {
 595                         $rethrow = $ex;
 596                 }
 597
 598                 // finally
 599                 libxml_disable_entity_loader( $oldDisable );
 600                 $this->reader->close();
 601
 602                 if ( $rethrow ) {
 603                         throw $rethrow;
 604                 }
 605
 606                 return true;
 607         }
 608
 609         private function handleSiteInfo() {
 610                 $this->debug( "Enter site info handler." );
 611                 $siteInfo = [];
 612
 613                 // Fields that can just be stuffed in the siteInfo object
 614                 $normalFields = [ 'sitename', 'base', 'generator', 'case' ];
 615
 616                 while ( $this->reader->read() ) {
 617                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 618                                         $this->reader->localName == 'siteinfo' ) {
 619                                 break;
 620                         }
 621
 622                         $tag = $this->reader->localName;
 623
 624                         if ( $tag == 'namespace' ) {
 625                                 $this->foreignNamespaces[$this->nodeAttribute( 'key' )] =
 626                                         $this->nodeContents();
 627                         } elseif ( in_array( $tag, $normalFields ) ) {
 628                                 $siteInfo[$tag] = $this->nodeContents();
 629                         }
 630                 }
 631
 632                 $siteInfo['_namespaces'] = $this->foreignNamespaces;
 633                 $this->siteInfoCallback( $siteInfo );
 634         }
 635
 636         private function handleLogItem() {
 637                 $this->debug( "Enter log item handler." );
 638                 $logInfo = [];
 639
 640                 // Fields that can just be stuffed in the pageInfo object
 641                 $normalFields = [ 'id', 'comment', 'type', 'action', 'timestamp',
 642                                         'logtitle', 'params' ];
 643
 644                 while ( $this->reader->read() ) {
 645                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 646                                         $this->reader->localName == 'logitem' ) {
 647                                 break;
 648                         }
 649
 650                         $tag = $this->reader->localName;
 651
 652                         if ( !Hooks::run( 'ImportHandleLogItemXMLTag', [
 653                                 $this, $logInfo
 654                         ] ) ) {
 655                                 // Do nothing
 656                         } elseif ( in_array( $tag, $normalFields ) ) {
 657                                 $logInfo[$tag] = $this->nodeContents();
 658                         } elseif ( $tag == 'contributor' ) {
 659                                 $logInfo['contributor'] = $this->handleContributor();
 660                         } elseif ( $tag != '#text' ) {
 661                                 $this->warn( "Unhandled log-item XML tag $tag" );
 662                         }
 663                 }
 664
 665                 $this->processLogItem( $logInfo );
 666         }
 667
 668         /**
 669          * @param array $logInfo
 670          * @return bool|mixed
 671          */
 672         private function processLogItem( $logInfo ) {
 673
 674                 $revision = new WikiRevision( $this->config );
 675
 676                 if ( isset( $logInfo['id'] ) ) {
 677                         $revision->setID( $logInfo['id'] );
 678                 }
 679                 $revision->setType( $logInfo['type'] );
 680                 $revision->setAction( $logInfo['action'] );
 681                 if ( isset( $logInfo['timestamp'] ) ) {
 682                         $revision->setTimestamp( $logInfo['timestamp'] );
 683                 }
 684                 if ( isset( $logInfo['params'] ) ) {
 685                         $revision->setParams( $logInfo['params'] );
 686                 }
 687                 if ( isset( $logInfo['logtitle'] ) ) {
 688                         // @todo Using Title for non-local titles is a recipe for disaster.
 689                         // We should use ForeignTitle here instead.
 690                         $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 691                 }
 692
 693                 $revision->setNoUpdates( $this->mNoUpdates );
 694
 695                 if ( isset( $logInfo['comment'] ) ) {
 696                         $revision->setComment( $logInfo['comment'] );
 697                 }
 698
 699                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 700                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 701                 }
 702
 703                 if ( !isset( $logInfo['contributor']['username'] ) ) {
 704                         $revision->setUsername( 'Unknown user' );
 705                 } else {
 706                         $revision->setUsername( $logInfo['contributor']['username'] );
 707                 }
 708
 709                 return $this->logItemCallback( $revision );
 710         }
 711
 712         private function handlePage() {
 713                 // Handle page data.
 714                 $this->debug( "Enter page handler." );
 715                 $pageInfo = [ 'revisionCount' => 0, 'successfulRevisionCount' => 0 ];
 716
 717                 // Fields that can just be stuffed in the pageInfo object
 718                 $normalFields = [ 'title', 'ns', 'id', 'redirect', 'restrictions' ];
 719
 720                 $skip = false;
 721                 $badTitle = false;
 722
 723                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 724                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 725                                         $this->reader->localName == 'page' ) {
 726                                 break;
 727                         }
 728
 729                         $skip = false;
 730
 731                         $tag = $this->reader->localName;
 732
 733                         if ( $badTitle ) {
 734                                 // The title is invalid, bail out of this page
 735                                 $skip = true;
 736                         } elseif ( !Hooks::run( 'ImportHandlePageXMLTag', [ $this,
 737                                                 &$pageInfo ] ) ) {
 738                                 // Do nothing
 739                         } elseif ( in_array( $tag, $normalFields ) ) {
 740                                 // An XML snippet:
 741                                 // <page>
 742                                 //     <id>123</id>
 743                                 //     <title>Page</title>
 744                                 //     <redirect title="NewTitle"/>
 745                                 //     ...
 746                                 // Because the redirect tag is built differently, we need special handling for that case.
 747                                 if ( $tag == 'redirect' ) {
 748                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 749                                 } else {
 750                                         $pageInfo[$tag] = $this->nodeContents();
 751                                 }
 752                         } elseif ( $tag == 'revision' || $tag == 'upload' ) {
 753                                 if ( !isset( $title ) ) {
 754                                         $title = $this->processTitle( $pageInfo['title'],
 755                                                 isset( $pageInfo['ns'] ) ? $pageInfo['ns'] : null );
 756
 757                                         // $title is either an array of two titles or false.
 758                                         if ( is_array( $title ) ) {
 759                                                 $this->pageCallback( $title );
 760                                                 list( $pageInfo['_title'], $foreignTitle ) = $title;
 761                                         } else {
 762                                                 $badTitle = true;
 763                                                 $skip = true;
 764                                         }
 765                                 }
 766
 767                                 if ( $title ) {
 768                                         if ( $tag == 'revision' ) {
 769                                                 $this->handleRevision( $pageInfo );
 770                                         } else {
 771                                                 $this->handleUpload( $pageInfo );
 772                                         }
 773                                 }
 774                         } elseif ( $tag != '#text' ) {
 775                                 $this->warn( "Unhandled page XML tag $tag" );
 776                                 $skip = true;
 777                         }
 778                 }
 779
 780                 // @note $pageInfo is only set if a valid $title is processed above with
 781                 //       no error. If we have a valid $title, then pageCallback is called
 782                 //       above, $pageInfo['title'] is set and we do pageOutCallback here.
 783                 //       If $pageInfo['_title'] is not set, then $foreignTitle is also not
 784                 //       set since they both come from $title above.
 785                 if ( array_key_exists( '_title', $pageInfo ) ) {
 786                         $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
 787                                         $pageInfo['revisionCount'],
 788                                         $pageInfo['successfulRevisionCount'],
 789                                         $pageInfo );
 790                 }
 791         }
 792
 793         /**
 794          * @param array $pageInfo
 795          */
 796         private function handleRevision( &$pageInfo ) {
 797                 $this->debug( "Enter revision handler" );
 798                 $revisionInfo = [];
 799
 800                 $normalFields = [ 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' ];
 801
 802                 $skip = false;
 803
 804                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 805                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 806                                         $this->reader->localName == 'revision' ) {
 807                                 break;
 808                         }
 809
 810                         $tag = $this->reader->localName;
 811
 812                         if ( !Hooks::run( 'ImportHandleRevisionXMLTag', [
 813                                 $this, $pageInfo, $revisionInfo
 814                         ] ) ) {
 815                                 // Do nothing
 816                         } elseif ( in_array( $tag, $normalFields ) ) {
 817                                 $revisionInfo[$tag] = $this->nodeContents();
 818                         } elseif ( $tag == 'contributor' ) {
 819                                 $revisionInfo['contributor'] = $this->handleContributor();
 820                         } elseif ( $tag != '#text' ) {
 821                                 $this->warn( "Unhandled revision XML tag $tag" );
 822                                 $skip = true;
 823                         }
 824                 }
 825
 826                 $pageInfo['revisionCount']++;
 827                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 828                         $pageInfo['successfulRevisionCount']++;
 829                 }
 830         }
 831
 832         /**
 833          * @param array $pageInfo
 834          * @param array $revisionInfo
 835          * @return bool|mixed
 836          */
 837         private function processRevision( $pageInfo, $revisionInfo ) {
 838                 global $wgMaxArticleSize;
 839
 840                 // Make sure revisions won't violate $wgMaxArticleSize, which could lead to
 841                 // database errors and instability. Testing for revisions with only listed
 842                 // content models, as other content models might use serialization formats
 843                 // which aren't checked against $wgMaxArticleSize.
 844                 if ( ( !isset( $revisionInfo['model'] ) ||
 845                         in_array( $revisionInfo['model'], [
 846                                 'wikitext',
 847                                 'css',
 848                                 'json',
 849                                 'javascript',
 850                                 'text',
 851                                 ''
 852                         ] ) ) &&
 853                         strlen( $revisionInfo['text'] ) > $wgMaxArticleSize * 1024
 854                 ) {
 855                         throw new MWException( 'The text of ' .
 856                                 ( isset( $revisionInfo['id'] ) ?
 857                                         "the revision with ID $revisionInfo[id]" :
 858                                         'a revision'
 859                                 ) . " exceeds the maximum allowable size ($wgMaxArticleSize KB)" );
 860                 }
 861
 862                 $revision = new WikiRevision( $this->config );
 863
 864                 if ( isset( $revisionInfo['id'] ) ) {
 865                         $revision->setID( $revisionInfo['id'] );
 866                 }
 867                 if ( isset( $revisionInfo['model'] ) ) {
 868                         $revision->setModel( $revisionInfo['model'] );
 869                 }
 870                 if ( isset( $revisionInfo['format'] ) ) {
 871                         $revision->setFormat( $revisionInfo['format'] );
 872                 }
 873                 $revision->setTitle( $pageInfo['_title'] );
 874
 875                 if ( isset( $revisionInfo['text'] ) ) {
 876                         $handler = $revision->getContentHandler();
 877                         $text = $handler->importTransform(
 878                                 $revisionInfo['text'],
 879                                 $revision->getFormat() );
 880
 881                         $revision->setText( $text );
 882                 }
 883                 if ( isset( $revisionInfo['timestamp'] ) ) {
 884                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 885                 } else {
 886                         $revision->setTimestamp( wfTimestampNow() );
 887                 }
 888
 889                 if ( isset( $revisionInfo['comment'] ) ) {
 890                         $revision->setComment( $revisionInfo['comment'] );
 891                 }
 892
 893                 if ( isset( $revisionInfo['minor'] ) ) {
 894                         $revision->setMinor( true );
 895                 }
 896                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 897                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 898                 } elseif ( isset( $revisionInfo['contributor']['username'] ) ) {
 899                         $revision->setUsername( $revisionInfo['contributor']['username'] );
 900                 } else {
 901                         $revision->setUsername( 'Unknown user' );
 902                 }
 903                 $revision->setNoUpdates( $this->mNoUpdates );
 904
 905                 return $this->revisionCallback( $revision );
 906         }
 907
 908         /**
 909          * @param array $pageInfo
 910          * @return mixed
 911          */
 912         private function handleUpload( &$pageInfo ) {
 913                 $this->debug( "Enter upload handler" );
 914                 $uploadInfo = [];
 915
 916                 $normalFields = [ 'timestamp', 'comment', 'filename', 'text',
 917                                         'src', 'size', 'sha1base36', 'archivename', 'rel' ];
 918
 919                 $skip = false;
 920
 921                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 922                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 923                                         $this->reader->localName == 'upload' ) {
 924                                 break;
 925                         }
 926
 927                         $tag = $this->reader->localName;
 928
 929                         if ( !Hooks::run( 'ImportHandleUploadXMLTag', [
 930                                 $this, $pageInfo
 931                         ] ) ) {
 932                                 // Do nothing
 933                         } elseif ( in_array( $tag, $normalFields ) ) {
 934                                 $uploadInfo[$tag] = $this->nodeContents();
 935                         } elseif ( $tag == 'contributor' ) {
 936                                 $uploadInfo['contributor'] = $this->handleContributor();
 937                         } elseif ( $tag == 'contents' ) {
 938                                 $contents = $this->nodeContents();
 939                                 $encoding = $this->reader->getAttribute( 'encoding' );
 940                                 if ( $encoding === 'base64' ) {
 941                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 942                                         $uploadInfo['isTempSrc'] = true;
 943                                 }
 944                         } elseif ( $tag != '#text' ) {
 945                                 $this->warn( "Unhandled upload XML tag $tag" );
 946                                 $skip = true;
 947                         }
 948                 }
 949
 950                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 951                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 952                         if ( file_exists( $path ) ) {
 953                                 $uploadInfo['fileSrc'] = $path;
 954                                 $uploadInfo['isTempSrc'] = false;
 955                         }
 956                 }
 957
 958                 if ( $this->mImportUploads ) {
 959                         return $this->processUpload( $pageInfo, $uploadInfo );
 960                 }
 961         }
 962
 963         /**
 964          * @param string $contents
 965          * @return string
 966          */
 967         private function dumpTemp( $contents ) {
 968                 $filename = tempnam( wfTempDir(), 'importupload' );
 969                 file_put_contents( $filename, $contents );
 970                 return $filename;
 971         }
 972
 973         /**
 974          * @param array $pageInfo
 975          * @param array $uploadInfo
 976          * @return mixed
 977          */
 978         private function processUpload( $pageInfo, $uploadInfo ) {
 979                 $revision = new WikiRevision( $this->config );
 980                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 981
 982                 $revision->setTitle( $pageInfo['_title'] );
 983                 $revision->setID( $pageInfo['id'] );
 984                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 985                 $revision->setText( $text );
 986                 $revision->setFilename( $uploadInfo['filename'] );
 987                 if ( isset( $uploadInfo['archivename'] ) ) {
 988                         $revision->setArchiveName( $uploadInfo['archivename'] );
 989                 }
 990                 $revision->setSrc( $uploadInfo['src'] );
 991                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 992                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 993                                 !empty( $uploadInfo['isTempSrc'] ) );
 994                 }
 995                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 996                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 997                 }
 998                 $revision->setSize( intval( $uploadInfo['size'] ) );
 999                 $revision->setComment( $uploadInfo['comment'] );
1000
1001                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
1002                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
1003                 }
1004                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
1005                         $revision->setUsername( $uploadInfo['contributor']['username'] );
1006                 }
1007                 $revision->setNoUpdates( $this->mNoUpdates );
1008
1009                 return call_user_func( $this->mUploadCallback, $revision );
1010         }
1011
1012         /**
1013          * @return array
1014          */
1015         private function handleContributor() {
1016                 $fields = [ 'id', 'ip', 'username' ];
1017                 $info = [];
1018
1019                 if ( $this->reader->isEmptyElement ) {
1020                         return $info;
1021                 }
1022                 while ( $this->reader->read() ) {
1023                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
1024                                         $this->reader->localName == 'contributor' ) {
1025                                 break;
1026                         }
1027
1028                         $tag = $this->reader->localName;
1029
1030                         if ( in_array( $tag, $fields ) ) {
1031                                 $info[$tag] = $this->nodeContents();
1032                         }
1033                 }
1034
1035                 return $info;
1036         }
1037
1038         /**
1039          * @param string $text
1040          * @param string|null $ns
1041          * @return array|bool
1042          */
1043         private function processTitle( $text, $ns = null ) {
1044                 if ( is_null( $this->foreignNamespaces ) ) {
1045                         $foreignTitleFactory = new NaiveForeignTitleFactory();
1046                 } else {
1047                         $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
1048                                 $this->foreignNamespaces );
1049                 }
1050
1051                 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
1052                         intval( $ns ) );
1053
1054                 $title = $this->importTitleFactory->createTitleFromForeignTitle(
1055                         $foreignTitle );
1056
1057                 $commandLineMode = $this->config->get( 'CommandLineMode' );
1058                 if ( is_null( $title ) ) {
1059                         # Invalid page title? Ignore the page
1060                         $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
1061                         return false;
1062                 } elseif ( $title->isExternal() ) {
1063                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
1064                         return false;
1065                 } elseif ( !$title->canExist() ) {
1066                         $this->notice( 'import-error-special', $title->getPrefixedText() );
1067                         return false;
1068                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
1069                         # Do not import if the importing wiki user cannot edit this page
1070                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
1071                         return false;
1072                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
1073                         # Do not import if the importing wiki user cannot create this page
1074                         $this->notice( 'import-error-create', $title->getPrefixedText() );
1075                         return false;
1076                 }
1077
1078                 return [ $title, $foreignTitle ];
1079         }
1080 }