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
  27 use MediaWiki\MediaWikiServices;
  28
  29 /**
  30  * XML file reader for the page data importer.
  31  *
  32  * implements Special:Import
  33  * @ingroup SpecialPage
  34  */
  35 class WikiImporter {
  36         /** @var XMLReader */
  37         private $reader;
  38         private $foreignNamespaces = null;
  39         private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
  40         private $mSiteInfoCallback, $mPageOutCallback;
  41         private $mNoticeCallback, $mDebug;
  42         private $mImportUploads, $mImageBasePath;
  43         private $mNoUpdates = false;
  44         private $pageOffset = 0;
  45         /** @var Config */
  46         private $config;
  47         /** @var ImportTitleFactory */
  48         private $importTitleFactory;
  49         /** @var array */
  50         private $countableCache = [];
  51         /** @var bool */
  52         private $disableStatisticsUpdate = false;
  53         /** @var ExternalUserNames */
  54         private $externalUserNames;
  55
  56         /**
  57          * Creates an ImportXMLReader drawing from the source provided
  58          * @param ImportSource $source
  59          * @param Config $config
  60          * @throws Exception
  61          */
  62         function __construct( ImportSource $source, Config $config ) {
  63                 if ( !class_exists( 'XMLReader' ) ) {
  64                         throw new Exception( 'Import requires PHP to have been compiled with libxml support' );
  65                 }
  66
  67                 $this->reader = new XMLReader();
  68                 $this->config = $config;
  69
  70                 if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) {
  71                         stream_wrapper_register( 'uploadsource', UploadSourceAdapter::class );
  72                 }
  73                 $id = UploadSourceAdapter::registerSource( $source );
  74
  75                 // Enable the entity loader, as it is needed for loading external URLs via
  76                 // XMLReader::open (T86036)
  77                 $oldDisable = libxml_disable_entity_loader( false );
  78                 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
  79                         $status = $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  80                 } else {
  81                         $status = $this->reader->open( "uploadsource://$id" );
  82                 }
  83                 if ( !$status ) {
  84                         $error = libxml_get_last_error();
  85                         libxml_disable_entity_loader( $oldDisable );
  86                         throw new MWException( 'Encountered an internal error while initializing WikiImporter object: ' .
  87                                 $error->message );
  88                 }
  89                 libxml_disable_entity_loader( $oldDisable );
  90
  91                 // Default callbacks
  92                 $this->setPageCallback( [ $this, 'beforeImportPage' ] );
  93                 $this->setRevisionCallback( [ $this, "importRevision" ] );
  94                 $this->setUploadCallback( [ $this, 'importUpload' ] );
  95                 $this->setLogItemCallback( [ $this, 'importLogItem' ] );
  96                 $this->setPageOutCallback( [ $this, 'finishImportPage' ] );
  97
  98                 $this->importTitleFactory = new NaiveImportTitleFactory();
  99                 $this->externalUserNames = new ExternalUserNames( 'imported', false );
 100         }
 101
 102         /**
 103          * @return null|XMLReader
 104          */
 105         public function getReader() {
 106                 return $this->reader;
 107         }
 108
 109         public function throwXmlError( $err ) {
 110                 $this->debug( "FAILURE: $err" );
 111                 wfDebug( "WikiImporter XML error: $err\n" );
 112         }
 113
 114         public function debug( $data ) {
 115                 if ( $this->mDebug ) {
 116                         wfDebug( "IMPORT: $data\n" );
 117                 }
 118         }
 119
 120         public function warn( $data ) {
 121                 wfDebug( "IMPORT: $data\n" );
 122         }
 123
 124         public function notice( $msg, ...$params ) {
 125                 if ( is_callable( $this->mNoticeCallback ) ) {
 126                         call_user_func( $this->mNoticeCallback, $msg, $params );
 127                 } else { # No ImportReporter -> CLI
 128                         // T177997: the command line importers should call setNoticeCallback()
 129                         // for their own custom callback to echo the notice
 130                         wfDebug( wfMessage( $msg, $params )->text() . "\n" );
 131                 }
 132         }
 133
 134         /**
 135          * Set debug mode...
 136          * @param bool $debug
 137          */
 138         function setDebug( $debug ) {
 139                 $this->mDebug = $debug;
 140         }
 141
 142         /**
 143          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
 144          * @param bool $noupdates
 145          */
 146         function setNoUpdates( $noupdates ) {
 147                 $this->mNoUpdates = $noupdates;
 148         }
 149
 150         /**
 151          * Sets 'pageOffset' value. So it will skip the first n-1 pages
 152          * and start from the nth page. It's 1-based indexing.
 153          * @param int $nthPage
 154          * @since 1.29
 155          */
 156         function setPageOffset( $nthPage ) {
 157                 $this->pageOffset = $nthPage;
 158         }
 159
 160         /**
 161          * Set a callback that displays notice messages
 162          *
 163          * @param callable $callback
 164          * @return callable
 165          */
 166         public function setNoticeCallback( $callback ) {
 167                 return wfSetVar( $this->mNoticeCallback, $callback );
 168         }
 169
 170         /**
 171          * Sets the action to perform as each new page in the stream is reached.
 172          * @param callable $callback
 173          * @return callable
 174          */
 175         public function setPageCallback( $callback ) {
 176                 $previous = $this->mPageCallback;
 177                 $this->mPageCallback = $callback;
 178                 return $previous;
 179         }
 180
 181         /**
 182          * Sets the action to perform as each page in the stream is completed.
 183          * Callback accepts the page title (as a Title object), a second object
 184          * with the original title form (in case it's been overridden into a
 185          * local namespace), and a count of revisions.
 186          *
 187          * @param callable $callback
 188          * @return callable
 189          */
 190         public function setPageOutCallback( $callback ) {
 191                 $previous = $this->mPageOutCallback;
 192                 $this->mPageOutCallback = $callback;
 193                 return $previous;
 194         }
 195
 196         /**
 197          * Sets the action to perform as each page revision is reached.
 198          * @param callable $callback
 199          * @return callable
 200          */
 201         public function setRevisionCallback( $callback ) {
 202                 $previous = $this->mRevisionCallback;
 203                 $this->mRevisionCallback = $callback;
 204                 return $previous;
 205         }
 206
 207         /**
 208          * Sets the action to perform as each file upload version is reached.
 209          * @param callable $callback
 210          * @return callable
 211          */
 212         public function setUploadCallback( $callback ) {
 213                 $previous = $this->mUploadCallback;
 214                 $this->mUploadCallback = $callback;
 215                 return $previous;
 216         }
 217
 218         /**
 219          * Sets the action to perform as each log item reached.
 220          * @param callable $callback
 221          * @return callable
 222          */
 223         public function setLogItemCallback( $callback ) {
 224                 $previous = $this->mLogItemCallback;
 225                 $this->mLogItemCallback = $callback;
 226                 return $previous;
 227         }
 228
 229         /**
 230          * Sets the action to perform when site info is encountered
 231          * @param callable $callback
 232          * @return callable
 233          */
 234         public function setSiteInfoCallback( $callback ) {
 235                 $previous = $this->mSiteInfoCallback;
 236                 $this->mSiteInfoCallback = $callback;
 237                 return $previous;
 238         }
 239
 240         /**
 241          * Sets the factory object to use to convert ForeignTitle objects into local
 242          * Title objects
 243          * @param ImportTitleFactory $factory
 244          */
 245         public function setImportTitleFactory( $factory ) {
 246                 $this->importTitleFactory = $factory;
 247         }
 248
 249         /**
 250          * Set a target namespace to override the defaults
 251          * @param null|int $namespace
 252          * @return bool
 253          */
 254         public function setTargetNamespace( $namespace ) {
 255                 if ( is_null( $namespace ) ) {
 256                         // Don't override namespaces
 257                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 258                         return true;
 259                 } elseif (
 260                         $namespace >= 0 &&
 261                         MediaWikiServices::getInstance()->getNamespaceInfo()->exists( intval( $namespace ) )
 262                 ) {
 263                         $namespace = intval( $namespace );
 264                         $this->setImportTitleFactory( new NamespaceImportTitleFactory( $namespace ) );
 265                         return true;
 266                 } else {
 267                         return false;
 268                 }
 269         }
 270
 271         /**
 272          * Set a target root page under which all pages are imported
 273          * @param null|string $rootpage
 274          * @return Status
 275          */
 276         public function setTargetRootPage( $rootpage ) {
 277                 $status = Status::newGood();
 278                 if ( is_null( $rootpage ) ) {
 279                         // No rootpage
 280                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 281                 } elseif ( $rootpage !== '' ) {
 282                         $rootpage = rtrim( $rootpage, '/' ); // avoid double slashes
 283                         $title = Title::newFromText( $rootpage );
 284
 285                         if ( !$title || $title->isExternal() ) {
 286                                 $status->fatal( 'import-rootpage-invalid' );
 287                         } elseif (
 288                                 !MediaWikiServices::getInstance()->getNamespaceInfo()->
 289                                 hasSubpages( $title->getNamespace() )
 290                         ) {
 291                                 $displayNSText = $title->getNamespace() == NS_MAIN
 292                                         ? wfMessage( 'blanknamespace' )->text()
 293                                         : MediaWikiServices::getInstance()->getContentLanguage()->
 294                                                 getNsText( $title->getNamespace() );
 295                                 $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 296                         } else {
 297                                 // set namespace to 'all', so the namespace check in processTitle() can pass
 298                                 $this->setTargetNamespace( null );
 299                                 $this->setImportTitleFactory( new SubpageImportTitleFactory( $title ) );
 300                         }
 301                 }
 302                 return $status;
 303         }
 304
 305         /**
 306          * @param string $dir
 307          */
 308         public function setImageBasePath( $dir ) {
 309                 $this->mImageBasePath = $dir;
 310         }
 311
 312         /**
 313          * @param bool $import
 314          */
 315         public function setImportUploads( $import ) {
 316                 $this->mImportUploads = $import;
 317         }
 318
 319         /**
 320          * @since 1.31
 321          * @param string $usernamePrefix Prefix to apply to unknown (and possibly also known) usernames
 322          * @param bool $assignKnownUsers Whether to apply the prefix to usernames that exist locally
 323          */
 324         public function setUsernamePrefix( $usernamePrefix, $assignKnownUsers ) {
 325                 $this->externalUserNames = new ExternalUserNames( $usernamePrefix, $assignKnownUsers );
 326         }
 327
 328         /**
 329          * Statistics update can cause a lot of time
 330          * @since 1.29
 331          */
 332         public function disableStatisticsUpdate() {
 333                 $this->disableStatisticsUpdate = true;
 334         }
 335
 336         /**
 337          * Default per-page callback. Sets up some things related to site statistics
 338          * @param array $titleAndForeignTitle Two-element array, with Title object at
 339          * index 0 and ForeignTitle object at index 1
 340          * @return bool
 341          */
 342         public function beforeImportPage( $titleAndForeignTitle ) {
 343                 $title = $titleAndForeignTitle[0];
 344                 $page = WikiPage::factory( $title );
 345                 $this->countableCache['title_' . $title->getPrefixedText()] = $page->isCountable();
 346                 return true;
 347         }
 348
 349         /**
 350          * Default per-revision callback, performs the import.
 351          * @param WikiRevision $revision
 352          * @return bool
 353          */
 354         public function importRevision( $revision ) {
 355                 if ( !$revision->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
 356                         $this->notice( 'import-error-bad-location',
 357                                 $revision->getTitle()->getPrefixedText(),
 358                                 $revision->getID(),
 359                                 $revision->getModel(),
 360                                 $revision->getFormat() );
 361
 362                         return false;
 363                 }
 364
 365                 try {
 366                         return $revision->importOldRevision();
 367                 } catch ( MWContentSerializationException $ex ) {
 368                         $this->notice( 'import-error-unserialize',
 369                                 $revision->getTitle()->getPrefixedText(),
 370                                 $revision->getID(),
 371                                 $revision->getModel(),
 372                                 $revision->getFormat() );
 373                 }
 374
 375                 return false;
 376         }
 377
 378         /**
 379          * Default per-revision callback, performs the import.
 380          * @param WikiRevision $revision
 381          * @return bool
 382          */
 383         public function importLogItem( $revision ) {
 384                 return $revision->importLogItem();
 385         }
 386
 387         /**
 388          * Dummy for now...
 389          * @param WikiRevision $revision
 390          * @return bool
 391          */
 392         public function importUpload( $revision ) {
 393                 return $revision->importUpload();
 394         }
 395
 396         /**
 397          * Mostly for hook use
 398          * @param Title $title
 399          * @param ForeignTitle $foreignTitle
 400          * @param int $revCount
 401          * @param int $sRevCount
 402          * @param array $pageInfo
 403          * @return bool
 404          */
 405         public function finishImportPage( $title, $foreignTitle, $revCount,
 406                 $sRevCount, $pageInfo
 407         ) {
 408                 // Update article count statistics (T42009)
 409                 // The normal counting logic in WikiPage->doEditUpdates() is designed for
 410                 // one-revision-at-a-time editing, not bulk imports. In this situation it
 411                 // suffers from issues of replica DB lag. We let WikiPage handle the total page
 412                 // and revision count, and we implement our own custom logic for the
 413                 // article (content page) count.
 414                 if ( !$this->disableStatisticsUpdate ) {
 415                         $page = WikiPage::factory( $title );
 416                         $page->loadPageData( 'fromdbmaster' );
 417                         $content = $page->getContent();
 418                         if ( $content === null ) {
 419                                 wfDebug( __METHOD__ . ': Skipping article count adjustment for ' . $title .
 420                                         ' because WikiPage::getContent() returned null' );
 421                         } else {
 422                                 $editInfo = $page->prepareContentForEdit( $content );
 423                                 $countKey = 'title_' . $title->getPrefixedText();
 424                                 $countable = $page->isCountable( $editInfo );
 425                                 if ( array_key_exists( $countKey, $this->countableCache ) &&
 426                                         $countable != $this->countableCache[$countKey] ) {
 427                                         DeferredUpdates::addUpdate( SiteStatsUpdate::factory( [
 428                                                 'articles' => ( (int)$countable - (int)$this->countableCache[$countKey] )
 429                                         ] ) );
 430                                 }
 431                         }
 432                 }
 433
 434                 return Hooks::run( 'AfterImportPage', func_get_args() );
 435         }
 436
 437         /**
 438          * Alternate per-revision callback, for debugging.
 439          * @param WikiRevision &$revision
 440          */
 441         public function debugRevisionHandler( &$revision ) {
 442                 $this->debug( "Got revision:" );
 443                 if ( is_object( $revision->title ) ) {
 444                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 445                 } else {
 446                         $this->debug( "-- Title: <invalid>" );
 447                 }
 448                 $this->debug( "-- User: " . $revision->user_text );
 449                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 450                 $this->debug( "-- Comment: " . $revision->comment );
 451                 $this->debug( "-- Text: " . $revision->text );
 452         }
 453
 454         /**
 455          * Notify the callback function of site info
 456          * @param array $siteInfo
 457          * @return bool|mixed
 458          */
 459         private function siteInfoCallback( $siteInfo ) {
 460                 if ( isset( $this->mSiteInfoCallback ) ) {
 461                         return call_user_func_array( $this->mSiteInfoCallback,
 462                                         [ $siteInfo, $this ] );
 463                 } else {
 464                         return false;
 465                 }
 466         }
 467
 468         /**
 469          * Notify the callback function when a new "<page>" is reached.
 470          * @param array $title
 471          */
 472         function pageCallback( $title ) {
 473                 if ( isset( $this->mPageCallback ) ) {
 474                         call_user_func( $this->mPageCallback, $title );
 475                 }
 476         }
 477
 478         /**
 479          * Notify the callback function when a "</page>" is closed.
 480          * @param Title $title
 481          * @param ForeignTitle $foreignTitle
 482          * @param int $revCount
 483          * @param int $sucCount Number of revisions for which callback returned true
 484          * @param array $pageInfo Associative array of page information
 485          */
 486         private function pageOutCallback( $title, $foreignTitle, $revCount,
 487                         $sucCount, $pageInfo ) {
 488                 if ( isset( $this->mPageOutCallback ) ) {
 489                         call_user_func_array( $this->mPageOutCallback, func_get_args() );
 490                 }
 491         }
 492
 493         /**
 494          * Notify the callback function of a revision
 495          * @param WikiRevision $revision
 496          * @return bool|mixed
 497          */
 498         private function revisionCallback( $revision ) {
 499                 if ( isset( $this->mRevisionCallback ) ) {
 500                         return call_user_func_array( $this->mRevisionCallback,
 501                                         [ $revision, $this ] );
 502                 } else {
 503                         return false;
 504                 }
 505         }
 506
 507         /**
 508          * Notify the callback function of a new log item
 509          * @param WikiRevision $revision
 510          * @return bool|mixed
 511          */
 512         private function logItemCallback( $revision ) {
 513                 if ( isset( $this->mLogItemCallback ) ) {
 514                         return call_user_func_array( $this->mLogItemCallback,
 515                                         [ $revision, $this ] );
 516                 } else {
 517                         return false;
 518                 }
 519         }
 520
 521         /**
 522          * Retrieves the contents of the named attribute of the current element.
 523          * @param string $attr The name of the attribute
 524          * @return string The value of the attribute or an empty string if it is not set in the current
 525          * element.
 526          */
 527         public function nodeAttribute( $attr ) {
 528                 return $this->reader->getAttribute( $attr );
 529         }
 530
 531         /**
 532          * Shouldn't something like this be built-in to XMLReader?
 533          * Fetches text contents of the current element, assuming
 534          * no sub-elements or such scary things.
 535          * @return string
 536          * @private
 537          */
 538         public function nodeContents() {
 539                 if ( $this->reader->isEmptyElement ) {
 540                         return "";
 541                 }
 542                 $buffer = "";
 543                 while ( $this->reader->read() ) {
 544                         switch ( $this->reader->nodeType ) {
 545                                 case XMLReader::TEXT:
 546                                 case XMLReader::CDATA:
 547                                 case XMLReader::SIGNIFICANT_WHITESPACE:
 548                                         $buffer .= $this->reader->value;
 549                                         break;
 550                                 case XMLReader::END_ELEMENT:
 551                                         return $buffer;
 552                         }
 553                 }
 554
 555                 $this->reader->close();
 556                 return '';
 557         }
 558
 559         /**
 560          * Primary entry point
 561          * @throws Exception
 562          * @throws MWException
 563          * @return bool
 564          */
 565         public function doImport() {
 566                 // Calls to reader->read need to be wrapped in calls to
 567                 // libxml_disable_entity_loader() to avoid local file
 568                 // inclusion attacks (T48932).
 569                 $oldDisable = libxml_disable_entity_loader( true );
 570                 $this->reader->read();
 571
 572                 if ( $this->reader->localName != 'mediawiki' ) {
 573                         libxml_disable_entity_loader( $oldDisable );
 574                         throw new MWException( "Expected <mediawiki> tag, got " .
 575                                 $this->reader->localName );
 576                 }
 577                 $this->debug( "<mediawiki> tag is correct." );
 578
 579                 $this->debug( "Starting primary dump processing loop." );
 580
 581                 $keepReading = $this->reader->read();
 582                 $skip = false;
 583                 $rethrow = null;
 584                 $pageCount = 0;
 585                 try {
 586                         while ( $keepReading ) {
 587                                 $tag = $this->reader->localName;
 588                                 if ( $this->pageOffset ) {
 589                                         if ( $tag === 'page' ) {
 590                                                 $pageCount++;
 591                                         }
 592                                         if ( $pageCount < $this->pageOffset ) {
 593                                                 $keepReading = $this->reader->next();
 594                                                 continue;
 595                                         }
 596                                 }
 597                                 $type = $this->reader->nodeType;
 598
 599                                 if ( !Hooks::run( 'ImportHandleToplevelXMLTag', [ $this ] ) ) {
 600                                         // Do nothing
 601                                 } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) {
 602                                         break;
 603                                 } elseif ( $tag == 'siteinfo' ) {
 604                                         $this->handleSiteInfo();
 605                                 } elseif ( $tag == 'page' ) {
 606                                         $this->handlePage();
 607                                 } elseif ( $tag == 'logitem' ) {
 608                                         $this->handleLogItem();
 609                                 } elseif ( $tag != '#text' ) {
 610                                         $this->warn( "Unhandled top-level XML tag $tag" );
 611
 612                                         $skip = true;
 613                                 }
 614
 615                                 if ( $skip ) {
 616                                         $keepReading = $this->reader->next();
 617                                         $skip = false;
 618                                         $this->debug( "Skip" );
 619                                 } else {
 620                                         $keepReading = $this->reader->read();
 621                                 }
 622                         }
 623                 } catch ( Exception $ex ) {
 624                         $rethrow = $ex;
 625                 }
 626
 627                 // finally
 628                 libxml_disable_entity_loader( $oldDisable );
 629                 $this->reader->close();
 630
 631                 if ( $rethrow ) {
 632                         throw $rethrow;
 633                 }
 634
 635                 return true;
 636         }
 637
 638         private function handleSiteInfo() {
 639                 $this->debug( "Enter site info handler." );
 640                 $siteInfo = [];
 641
 642                 // Fields that can just be stuffed in the siteInfo object
 643                 $normalFields = [ 'sitename', 'base', 'generator', 'case' ];
 644
 645                 while ( $this->reader->read() ) {
 646                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 647                                         $this->reader->localName == 'siteinfo' ) {
 648                                 break;
 649                         }
 650
 651                         $tag = $this->reader->localName;
 652
 653                         if ( $tag == 'namespace' ) {
 654                                 $this->foreignNamespaces[$this->nodeAttribute( 'key' )] =
 655                                         $this->nodeContents();
 656                         } elseif ( in_array( $tag, $normalFields ) ) {
 657                                 $siteInfo[$tag] = $this->nodeContents();
 658                         }
 659                 }
 660
 661                 $siteInfo['_namespaces'] = $this->foreignNamespaces;
 662                 $this->siteInfoCallback( $siteInfo );
 663         }
 664
 665         private function handleLogItem() {
 666                 $this->debug( "Enter log item handler." );
 667                 $logInfo = [];
 668
 669                 // Fields that can just be stuffed in the pageInfo object
 670                 $normalFields = [ 'id', 'comment', 'type', 'action', 'timestamp',
 671                                         'logtitle', 'params' ];
 672
 673                 while ( $this->reader->read() ) {
 674                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 675                                         $this->reader->localName == 'logitem' ) {
 676                                 break;
 677                         }
 678
 679                         $tag = $this->reader->localName;
 680
 681                         if ( !Hooks::run( 'ImportHandleLogItemXMLTag', [
 682                                 $this, $logInfo
 683                         ] ) ) {
 684                                 // Do nothing
 685                         } elseif ( in_array( $tag, $normalFields ) ) {
 686                                 $logInfo[$tag] = $this->nodeContents();
 687                         } elseif ( $tag == 'contributor' ) {
 688                                 $logInfo['contributor'] = $this->handleContributor();
 689                         } elseif ( $tag != '#text' ) {
 690                                 $this->warn( "Unhandled log-item XML tag $tag" );
 691                         }
 692                 }
 693
 694                 $this->processLogItem( $logInfo );
 695         }
 696
 697         /**
 698          * @param array $logInfo
 699          * @return bool|mixed
 700          */
 701         private function processLogItem( $logInfo ) {
 702                 $revision = new WikiRevision( $this->config );
 703
 704                 if ( isset( $logInfo['id'] ) ) {
 705                         $revision->setID( $logInfo['id'] );
 706                 }
 707                 $revision->setType( $logInfo['type'] );
 708                 $revision->setAction( $logInfo['action'] );
 709                 if ( isset( $logInfo['timestamp'] ) ) {
 710                         $revision->setTimestamp( $logInfo['timestamp'] );
 711                 }
 712                 if ( isset( $logInfo['params'] ) ) {
 713                         $revision->setParams( $logInfo['params'] );
 714                 }
 715                 if ( isset( $logInfo['logtitle'] ) ) {
 716                         // @todo Using Title for non-local titles is a recipe for disaster.
 717                         // We should use ForeignTitle here instead.
 718                         $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 719                 }
 720
 721                 $revision->setNoUpdates( $this->mNoUpdates );
 722
 723                 if ( isset( $logInfo['comment'] ) ) {
 724                         $revision->setComment( $logInfo['comment'] );
 725                 }
 726
 727                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 728                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 729                 }
 730
 731                 if ( !isset( $logInfo['contributor']['username'] ) ) {
 732                         $revision->setUsername( $this->externalUserNames->addPrefix( 'Unknown user' ) );
 733                 } else {
 734                         $revision->setUsername(
 735                                 $this->externalUserNames->applyPrefix( $logInfo['contributor']['username'] )
 736                         );
 737                 }
 738
 739                 return $this->logItemCallback( $revision );
 740         }
 741
 742         /**
 743          * @suppress PhanTypeInvalidDimOffset Phan not reading the reference inside the hook
 744          */
 745         private function handlePage() {
 746                 // Handle page data.
 747                 $this->debug( "Enter page handler." );
 748                 $pageInfo = [ 'revisionCount' => 0, 'successfulRevisionCount' => 0 ];
 749
 750                 // Fields that can just be stuffed in the pageInfo object
 751                 $normalFields = [ 'title', 'ns', 'id', 'redirect', 'restrictions' ];
 752
 753                 $skip = false;
 754                 $badTitle = false;
 755
 756                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 757                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 758                                         $this->reader->localName == 'page' ) {
 759                                 break;
 760                         }
 761
 762                         $skip = false;
 763
 764                         $tag = $this->reader->localName;
 765
 766                         if ( $badTitle ) {
 767                                 // The title is invalid, bail out of this page
 768                                 $skip = true;
 769                         } elseif ( !Hooks::run( 'ImportHandlePageXMLTag', [ $this,
 770                                                 &$pageInfo ] ) ) {
 771                                 // Do nothing
 772                         } elseif ( in_array( $tag, $normalFields ) ) {
 773                                 // An XML snippet:
 774                                 // <page>
 775                                 //     <id>123</id>
 776                                 //     <title>Page</title>
 777                                 //     <redirect title="NewTitle"/>
 778                                 //     ...
 779                                 // Because the redirect tag is built differently, we need special handling for that case.
 780                                 if ( $tag == 'redirect' ) {
 781                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 782                                 } else {
 783                                         $pageInfo[$tag] = $this->nodeContents();
 784                                 }
 785                         } elseif ( $tag == 'revision' || $tag == 'upload' ) {
 786                                 if ( !isset( $title ) ) {
 787                                         $title = $this->processTitle( $pageInfo['title'],
 788                                                 $pageInfo['ns'] ?? null );
 789
 790                                         // $title is either an array of two titles or false.
 791                                         if ( is_array( $title ) ) {
 792                                                 $this->pageCallback( $title );
 793                                                 list( $pageInfo['_title'], $foreignTitle ) = $title;
 794                                         } else {
 795                                                 $badTitle = true;
 796                                                 $skip = true;
 797                                         }
 798                                 }
 799
 800                                 if ( $title ) {
 801                                         if ( $tag == 'revision' ) {
 802                                                 $this->handleRevision( $pageInfo );
 803                                         } else {
 804                                                 $this->handleUpload( $pageInfo );
 805                                         }
 806                                 }
 807                         } elseif ( $tag != '#text' ) {
 808                                 $this->warn( "Unhandled page XML tag $tag" );
 809                                 $skip = true;
 810                         }
 811                 }
 812
 813                 // @note $pageInfo is only set if a valid $title is processed above with
 814                 //       no error. If we have a valid $title, then pageCallback is called
 815                 //       above, $pageInfo['title'] is set and we do pageOutCallback here.
 816                 //       If $pageInfo['_title'] is not set, then $foreignTitle is also not
 817                 //       set since they both come from $title above.
 818                 if ( array_key_exists( '_title', $pageInfo ) ) {
 819                         $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
 820                                         $pageInfo['revisionCount'],
 821                                         $pageInfo['successfulRevisionCount'],
 822                                         $pageInfo );
 823                 }
 824         }
 825
 826         /**
 827          * @param array $pageInfo
 828          */
 829         private function handleRevision( &$pageInfo ) {
 830                 $this->debug( "Enter revision handler" );
 831                 $revisionInfo = [];
 832
 833                 $normalFields = [ 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text', 'sha1' ];
 834
 835                 $skip = false;
 836
 837                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 838                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 839                                         $this->reader->localName == 'revision' ) {
 840                                 break;
 841                         }
 842
 843                         $tag = $this->reader->localName;
 844
 845                         if ( !Hooks::run( 'ImportHandleRevisionXMLTag', [
 846                                 $this, $pageInfo, $revisionInfo
 847                         ] ) ) {
 848                                 // Do nothing
 849                         } elseif ( in_array( $tag, $normalFields ) ) {
 850                                 $revisionInfo[$tag] = $this->nodeContents();
 851                         } elseif ( $tag == 'contributor' ) {
 852                                 $revisionInfo['contributor'] = $this->handleContributor();
 853                         } elseif ( $tag != '#text' ) {
 854                                 $this->warn( "Unhandled revision XML tag $tag" );
 855                                 $skip = true;
 856                         }
 857                 }
 858
 859                 $pageInfo['revisionCount']++;
 860                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 861                         $pageInfo['successfulRevisionCount']++;
 862                 }
 863         }
 864
 865         /**
 866          * @param array $pageInfo
 867          * @param array $revisionInfo
 868          * @throws MWException
 869          * @return bool|mixed
 870          */
 871         private function processRevision( $pageInfo, $revisionInfo ) {
 872                 global $wgMaxArticleSize;
 873
 874                 // Make sure revisions won't violate $wgMaxArticleSize, which could lead to
 875                 // database errors and instability. Testing for revisions with only listed
 876                 // content models, as other content models might use serialization formats
 877                 // which aren't checked against $wgMaxArticleSize.
 878                 if ( ( !isset( $revisionInfo['model'] ) ||
 879                         in_array( $revisionInfo['model'], [
 880                                 'wikitext',
 881                                 'css',
 882                                 'json',
 883                                 'javascript',
 884                                 'text',
 885                                 ''
 886                         ] ) ) &&
 887                         strlen( $revisionInfo['text'] ) > $wgMaxArticleSize * 1024
 888                 ) {
 889                         throw new MWException( 'The text of ' .
 890                                 ( isset( $revisionInfo['id'] ) ?
 891                                         "the revision with ID $revisionInfo[id]" :
 892                                         'a revision'
 893                                 ) . " exceeds the maximum allowable size ($wgMaxArticleSize KB)" );
 894                 }
 895
 896                 // FIXME: process schema version 11!
 897                 $revision = new WikiRevision( $this->config );
 898
 899                 if ( isset( $revisionInfo['id'] ) ) {
 900                         $revision->setID( $revisionInfo['id'] );
 901                 }
 902                 if ( isset( $revisionInfo['model'] ) ) {
 903                         $revision->setModel( $revisionInfo['model'] );
 904                 }
 905                 if ( isset( $revisionInfo['format'] ) ) {
 906                         $revision->setFormat( $revisionInfo['format'] );
 907                 }
 908                 $revision->setTitle( $pageInfo['_title'] );
 909
 910                 if ( isset( $revisionInfo['text'] ) ) {
 911                         $handler = $revision->getContentHandler();
 912                         $text = $handler->importTransform(
 913                                 $revisionInfo['text'],
 914                                 $revision->getFormat() );
 915
 916                         $revision->setText( $text );
 917                 }
 918                 $revision->setTimestamp( $revisionInfo['timestamp'] ?? wfTimestampNow() );
 919
 920                 if ( isset( $revisionInfo['comment'] ) ) {
 921                         $revision->setComment( $revisionInfo['comment'] );
 922                 }
 923
 924                 if ( isset( $revisionInfo['minor'] ) ) {
 925                         $revision->setMinor( true );
 926                 }
 927                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 928                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 929                 } elseif ( isset( $revisionInfo['contributor']['username'] ) ) {
 930                         $revision->setUsername(
 931                                 $this->externalUserNames->applyPrefix( $revisionInfo['contributor']['username'] )
 932                         );
 933                 } else {
 934                         $revision->setUsername( $this->externalUserNames->addPrefix( 'Unknown user' ) );
 935                 }
 936                 if ( isset( $revisionInfo['sha1'] ) ) {
 937                         $revision->setSha1Base36( $revisionInfo['sha1'] );
 938                 }
 939                 $revision->setNoUpdates( $this->mNoUpdates );
 940
 941                 return $this->revisionCallback( $revision );
 942         }
 943
 944         /**
 945          * @param array $pageInfo
 946          * @return mixed
 947          */
 948         private function handleUpload( &$pageInfo ) {
 949                 $this->debug( "Enter upload handler" );
 950                 $uploadInfo = [];
 951
 952                 $normalFields = [ 'timestamp', 'comment', 'filename', 'text',
 953                                         'src', 'size', 'sha1base36', 'archivename', 'rel' ];
 954
 955                 $skip = false;
 956
 957                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 958                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 959                                         $this->reader->localName == 'upload' ) {
 960                                 break;
 961                         }
 962
 963                         $tag = $this->reader->localName;
 964
 965                         if ( !Hooks::run( 'ImportHandleUploadXMLTag', [
 966                                 $this, $pageInfo
 967                         ] ) ) {
 968                                 // Do nothing
 969                         } elseif ( in_array( $tag, $normalFields ) ) {
 970                                 $uploadInfo[$tag] = $this->nodeContents();
 971                         } elseif ( $tag == 'contributor' ) {
 972                                 $uploadInfo['contributor'] = $this->handleContributor();
 973                         } elseif ( $tag == 'contents' ) {
 974                                 $contents = $this->nodeContents();
 975                                 $encoding = $this->reader->getAttribute( 'encoding' );
 976                                 if ( $encoding === 'base64' ) {
 977                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 978                                         $uploadInfo['isTempSrc'] = true;
 979                                 }
 980                         } elseif ( $tag != '#text' ) {
 981                                 $this->warn( "Unhandled upload XML tag $tag" );
 982                                 $skip = true;
 983                         }
 984                 }
 985
 986                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 987                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 988                         if ( file_exists( $path ) ) {
 989                                 $uploadInfo['fileSrc'] = $path;
 990                                 $uploadInfo['isTempSrc'] = false;
 991                         }
 992                 }
 993
 994                 if ( $this->mImportUploads ) {
 995                         return $this->processUpload( $pageInfo, $uploadInfo );
 996                 }
 997         }
 998
 999         /**
1000          * @param string $contents
1001          * @return string
1002          */
1003         private function dumpTemp( $contents ) {
1004                 $filename = tempnam( wfTempDir(), 'importupload' );
1005                 file_put_contents( $filename, $contents );
1006                 return $filename;
1007         }
1008
1009         /**
1010          * @param array $pageInfo
1011          * @param array $uploadInfo
1012          * @return mixed
1013          */
1014         private function processUpload( $pageInfo, $uploadInfo ) {
1015                 $revision = new WikiRevision( $this->config );
1016                 $text = $uploadInfo['text'] ?? '';
1017
1018                 $revision->setTitle( $pageInfo['_title'] );
1019                 $revision->setID( $pageInfo['id'] );
1020                 $revision->setTimestamp( $uploadInfo['timestamp'] );
1021                 $revision->setText( $text );
1022                 $revision->setFilename( $uploadInfo['filename'] );
1023                 if ( isset( $uploadInfo['archivename'] ) ) {
1024                         $revision->setArchiveName( $uploadInfo['archivename'] );
1025                 }
1026                 $revision->setSrc( $uploadInfo['src'] );
1027                 if ( isset( $uploadInfo['fileSrc'] ) ) {
1028                         $revision->setFileSrc( $uploadInfo['fileSrc'],
1029                                 !empty( $uploadInfo['isTempSrc'] ) );
1030                 }
1031                 if ( isset( $uploadInfo['sha1base36'] ) ) {
1032                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
1033                 }
1034                 $revision->setSize( intval( $uploadInfo['size'] ) );
1035                 $revision->setComment( $uploadInfo['comment'] );
1036
1037                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
1038                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
1039                 }
1040                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
1041                         $revision->setUsername(
1042                                 $this->externalUserNames->applyPrefix( $uploadInfo['contributor']['username'] )
1043                         );
1044                 }
1045                 $revision->setNoUpdates( $this->mNoUpdates );
1046
1047                 return call_user_func( $this->mUploadCallback, $revision );
1048         }
1049
1050         /**
1051          * @return array
1052          */
1053         private function handleContributor() {
1054                 $fields = [ 'id', 'ip', 'username' ];
1055                 $info = [];
1056
1057                 if ( $this->reader->isEmptyElement ) {
1058                         return $info;
1059                 }
1060                 while ( $this->reader->read() ) {
1061                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
1062                                         $this->reader->localName == 'contributor' ) {
1063                                 break;
1064                         }
1065
1066                         $tag = $this->reader->localName;
1067
1068                         if ( in_array( $tag, $fields ) ) {
1069                                 $info[$tag] = $this->nodeContents();
1070                         }
1071                 }
1072
1073                 return $info;
1074         }
1075
1076         /**
1077          * @param string $text
1078          * @param string|null $ns
1079          * @return array|bool
1080          */
1081         private function processTitle( $text, $ns = null ) {
1082                 if ( is_null( $this->foreignNamespaces ) ) {
1083                         $foreignTitleFactory = new NaiveForeignTitleFactory();
1084                 } else {
1085                         $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
1086                                 $this->foreignNamespaces );
1087                 }
1088
1089                 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
1090                         intval( $ns ) );
1091
1092                 $title = $this->importTitleFactory->createTitleFromForeignTitle(
1093                         $foreignTitle );
1094
1095                 $commandLineMode = $this->config->get( 'CommandLineMode' );
1096                 if ( is_null( $title ) ) {
1097                         # Invalid page title? Ignore the page
1098                         $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
1099                         return false;
1100                 } elseif ( $title->isExternal() ) {
1101                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
1102                         return false;
1103                 } elseif ( !$title->canExist() ) {
1104                         $this->notice( 'import-error-special', $title->getPrefixedText() );
1105                         return false;
1106                 } elseif ( !$commandLineMode ) {
1107                         $permissionManager = MediaWikiServices::getInstance()->getPermissionManager();
1108                         $user = RequestContext::getMain()->getUser();
1109
1110                         if ( !$permissionManager->userCan( 'edit', $user, $title ) ) {
1111                                 # Do not import if the importing wiki user cannot edit this page
1112                                 $this->notice( 'import-error-edit', $title->getPrefixedText() );
1113
1114                                 return false;
1115                         }
1116
1117                         if ( !$title->exists() && !$permissionManager->userCan( 'create', $user, $title ) ) {
1118                                 # Do not import if the importing wiki user cannot create this page
1119                                 $this->notice( 'import-error-create', $title->getPrefixedText() );
1120
1121                                 return false;
1122                         }
1123                 }
1124
1125                 return [ $title, $foreignTitle ];
1126         }
1127 }