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         private $reader = null;
  37         private $foreignNamespaces = null;
  38         private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
  39         private $mSiteInfoCallback, $mPageOutCallback;
  40         private $mNoticeCallback, $mDebug;
  41         private $mImportUploads, $mImageBasePath;
  42         private $mNoUpdates = false;
  43         private $pageOffset = 0;
  44         /** @var Config */
  45         private $config;
  46         /** @var ImportTitleFactory */
  47         private $importTitleFactory;
  48         /** @var array */
  49         private $countableCache = [];
  50         /** @var bool */
  51         private $disableStatisticsUpdate = false;
  52         /** @var ExternalUserNames */
  53         private $externalUserNames;
  54
  55         /**
  56          * Creates an ImportXMLReader drawing from the source provided
  57          * @param ImportSource $source
  58          * @param Config $config
  59          * @throws Exception
  60          */
  61         function __construct( ImportSource $source, Config $config ) {
  62                 if ( !class_exists( 'XMLReader' ) ) {
  63                         throw new Exception( 'Import requires PHP to have been compiled with libxml support' );
  64                 }
  65
  66                 $this->reader = new XMLReader();
  67                 $this->config = $config;
  68
  69                 if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) {
  70                         stream_wrapper_register( 'uploadsource', UploadSourceAdapter::class );
  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                 $this->externalUserNames = new ExternalUserNames( 'imported', false );
  99         }
 100
 101         /**
 102          * @return null|XMLReader
 103          */
 104         public function getReader() {
 105                 return $this->reader;
 106         }
 107
 108         public function throwXmlError( $err ) {
 109                 $this->debug( "FAILURE: $err" );
 110                 wfDebug( "WikiImporter XML error: $err\n" );
 111         }
 112
 113         public function debug( $data ) {
 114                 if ( $this->mDebug ) {
 115                         wfDebug( "IMPORT: $data\n" );
 116                 }
 117         }
 118
 119         public function warn( $data ) {
 120                 wfDebug( "IMPORT: $data\n" );
 121         }
 122
 123         public function notice( $msg /*, $param, ...*/ ) {
 124                 $params = func_get_args();
 125                 array_shift( $params );
 126
 127                 if ( is_callable( $this->mNoticeCallback ) ) {
 128                         call_user_func( $this->mNoticeCallback, $msg, $params );
 129                 } else { # No ImportReporter -> CLI
 130                         // T177997: the command line importers should call setNoticeCallback()
 131                         // for their own custom callback to echo the notice
 132                         wfDebug( wfMessage( $msg, $params )->text() . "\n" );
 133                 }
 134         }
 135
 136         /**
 137          * Set debug mode...
 138          * @param bool $debug
 139          */
 140         function setDebug( $debug ) {
 141                 $this->mDebug = $debug;
 142         }
 143
 144         /**
 145          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
 146          * @param bool $noupdates
 147          */
 148         function setNoUpdates( $noupdates ) {
 149                 $this->mNoUpdates = $noupdates;
 150         }
 151
 152         /**
 153          * Sets 'pageOffset' value. So it will skip the first n-1 pages
 154          * and start from the nth page. It's 1-based indexing.
 155          * @param int $nthPage
 156          * @since 1.29
 157          */
 158         function setPageOffset( $nthPage ) {
 159                 $this->pageOffset = $nthPage;
 160         }
 161
 162         /**
 163          * Set a callback that displays notice messages
 164          *
 165          * @param callable $callback
 166          * @return callable
 167          */
 168         public function setNoticeCallback( $callback ) {
 169                 return wfSetVar( $this->mNoticeCallback, $callback );
 170         }
 171
 172         /**
 173          * Sets the action to perform as each new page in the stream is reached.
 174          * @param callable $callback
 175          * @return callable
 176          */
 177         public function setPageCallback( $callback ) {
 178                 $previous = $this->mPageCallback;
 179                 $this->mPageCallback = $callback;
 180                 return $previous;
 181         }
 182
 183         /**
 184          * Sets the action to perform as each page in the stream is completed.
 185          * Callback accepts the page title (as a Title object), a second object
 186          * with the original title form (in case it's been overridden into a
 187          * local namespace), and a count of revisions.
 188          *
 189          * @param callable $callback
 190          * @return callable
 191          */
 192         public function setPageOutCallback( $callback ) {
 193                 $previous = $this->mPageOutCallback;
 194                 $this->mPageOutCallback = $callback;
 195                 return $previous;
 196         }
 197
 198         /**
 199          * Sets the action to perform as each page revision is reached.
 200          * @param callable $callback
 201          * @return callable
 202          */
 203         public function setRevisionCallback( $callback ) {
 204                 $previous = $this->mRevisionCallback;
 205                 $this->mRevisionCallback = $callback;
 206                 return $previous;
 207         }
 208
 209         /**
 210          * Sets the action to perform as each file upload version is reached.
 211          * @param callable $callback
 212          * @return callable
 213          */
 214         public function setUploadCallback( $callback ) {
 215                 $previous = $this->mUploadCallback;
 216                 $this->mUploadCallback = $callback;
 217                 return $previous;
 218         }
 219
 220         /**
 221          * Sets the action to perform as each log item reached.
 222          * @param callable $callback
 223          * @return callable
 224          */
 225         public function setLogItemCallback( $callback ) {
 226                 $previous = $this->mLogItemCallback;
 227                 $this->mLogItemCallback = $callback;
 228                 return $previous;
 229         }
 230
 231         /**
 232          * Sets the action to perform when site info is encountered
 233          * @param callable $callback
 234          * @return callable
 235          */
 236         public function setSiteInfoCallback( $callback ) {
 237                 $previous = $this->mSiteInfoCallback;
 238                 $this->mSiteInfoCallback = $callback;
 239                 return $previous;
 240         }
 241
 242         /**
 243          * Sets the factory object to use to convert ForeignTitle objects into local
 244          * Title objects
 245          * @param ImportTitleFactory $factory
 246          */
 247         public function setImportTitleFactory( $factory ) {
 248                 $this->importTitleFactory = $factory;
 249         }
 250
 251         /**
 252          * Set a target namespace to override the defaults
 253          * @param null|int $namespace
 254          * @return bool
 255          */
 256         public function setTargetNamespace( $namespace ) {
 257                 if ( is_null( $namespace ) ) {
 258                         // Don't override namespaces
 259                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 260                         return true;
 261                 } elseif (
 262                         $namespace >= 0 &&
 263                         MWNamespace::exists( intval( $namespace ) )
 264                 ) {
 265                         $namespace = intval( $namespace );
 266                         $this->setImportTitleFactory( new NamespaceImportTitleFactory( $namespace ) );
 267                         return true;
 268                 } else {
 269                         return false;
 270                 }
 271         }
 272
 273         /**
 274          * Set a target root page under which all pages are imported
 275          * @param null|string $rootpage
 276          * @return Status
 277          */
 278         public function setTargetRootPage( $rootpage ) {
 279                 $status = Status::newGood();
 280                 if ( is_null( $rootpage ) ) {
 281                         // No rootpage
 282                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 283                 } elseif ( $rootpage !== '' ) {
 284                         $rootpage = rtrim( $rootpage, '/' ); // avoid double slashes
 285                         $title = Title::newFromText( $rootpage );
 286
 287                         if ( !$title || $title->isExternal() ) {
 288                                 $status->fatal( 'import-rootpage-invalid' );
 289                         } else {
 290                                 if ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 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                 }
 303                 return $status;
 304         }
 305
 306         /**
 307          * @param string $dir
 308          */
 309         public function setImageBasePath( $dir ) {
 310                 $this->mImageBasePath = $dir;
 311         }
 312
 313         /**
 314          * @param bool $import
 315          */
 316         public function setImportUploads( $import ) {
 317                 $this->mImportUploads = $import;
 318         }
 319
 320         /**
 321          * @since 1.31
 322          * @param string $usernamePrefix Prefix to apply to unknown (and possibly also known) usernames
 323          * @param bool $assignKnownUsers Whether to apply the prefix to usernames that exist locally
 324          */
 325         public function setUsernamePrefix( $usernamePrefix, $assignKnownUsers ) {
 326                 $this->externalUserNames = new ExternalUserNames( $usernamePrefix, $assignKnownUsers );
 327         }
 328
 329         /**
 330          * Statistics update can cause a lot of time
 331          * @since 1.29
 332          */
 333         public function disableStatisticsUpdate() {
 334                 $this->disableStatisticsUpdate = true;
 335         }
 336
 337         /**
 338          * Default per-page callback. Sets up some things related to site statistics
 339          * @param array $titleAndForeignTitle Two-element array, with Title object at
 340          * index 0 and ForeignTitle object at index 1
 341          * @return bool
 342          */
 343         public function beforeImportPage( $titleAndForeignTitle ) {
 344                 $title = $titleAndForeignTitle[0];
 345                 $page = WikiPage::factory( $title );
 346                 $this->countableCache['title_' . $title->getPrefixedText()] = $page->isCountable();
 347                 return true;
 348         }
 349
 350         /**
 351          * Default per-revision callback, performs the import.
 352          * @param WikiRevision $revision
 353          * @return bool
 354          */
 355         public function importRevision( $revision ) {
 356                 if ( !$revision->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
 357                         $this->notice( 'import-error-bad-location',
 358                                 $revision->getTitle()->getPrefixedText(),
 359                                 $revision->getID(),
 360                                 $revision->getModel(),
 361                                 $revision->getFormat() );
 362
 363                         return false;
 364                 }
 365
 366                 try {
 367                         return $revision->importOldRevision();
 368                 } catch ( MWContentSerializationException $ex ) {
 369                         $this->notice( 'import-error-unserialize',
 370                                 $revision->getTitle()->getPrefixedText(),
 371                                 $revision->getID(),
 372                                 $revision->getModel(),
 373                                 $revision->getFormat() );
 374                 }
 375
 376                 return false;
 377         }
 378
 379         /**
 380          * Default per-revision callback, performs the import.
 381          * @param WikiRevision $revision
 382          * @return bool
 383          */
 384         public function importLogItem( $revision ) {
 385                 return $revision->importLogItem();
 386         }
 387
 388         /**
 389          * Dummy for now...
 390          * @param WikiRevision $revision
 391          * @return bool
 392          */
 393         public function importUpload( $revision ) {
 394                 return $revision->importUpload();
 395         }
 396
 397         /**
 398          * Mostly for hook use
 399          * @param Title $title
 400          * @param ForeignTitle $foreignTitle
 401          * @param int $revCount
 402          * @param int $sRevCount
 403          * @param array $pageInfo
 404          * @return bool
 405          */
 406         public function finishImportPage( $title, $foreignTitle, $revCount,
 407                 $sRevCount, $pageInfo
 408         ) {
 409                 // Update article count statistics (T42009)
 410                 // The normal counting logic in WikiPage->doEditUpdates() is designed for
 411                 // one-revision-at-a-time editing, not bulk imports. In this situation it
 412                 // suffers from issues of replica DB lag. We let WikiPage handle the total page
 413                 // and revision count, and we implement our own custom logic for the
 414                 // article (content page) count.
 415                 if ( !$this->disableStatisticsUpdate ) {
 416                         $page = WikiPage::factory( $title );
 417                         $page->loadPageData( 'fromdbmaster' );
 418                         $content = $page->getContent();
 419                         if ( $content === null ) {
 420                                 wfDebug( __METHOD__ . ': Skipping article count adjustment for ' . $title .
 421                                         ' because WikiPage::getContent() returned null' );
 422                         } else {
 423                                 $editInfo = $page->prepareContentForEdit( $content );
 424                                 $countKey = 'title_' . $title->getPrefixedText();
 425                                 $countable = $page->isCountable( $editInfo );
 426                                 if ( array_key_exists( $countKey, $this->countableCache ) &&
 427                                         $countable != $this->countableCache[$countKey] ) {
 428                                         DeferredUpdates::addUpdate( SiteStatsUpdate::factory( [
 429                                                 'articles' => ( (int)$countable - (int)$this->countableCache[$countKey] )
 430                                         ] ) );
 431                                 }
 432                         }
 433                 }
 434
 435                 $args = func_get_args();
 436                 return Hooks::run( 'AfterImportPage', $args );
 437         }
 438
 439         /**
 440          * Alternate per-revision callback, for debugging.
 441          * @param WikiRevision &$revision
 442          */
 443         public function debugRevisionHandler( &$revision ) {
 444                 $this->debug( "Got revision:" );
 445                 if ( is_object( $revision->title ) ) {
 446                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 447                 } else {
 448                         $this->debug( "-- Title: <invalid>" );
 449                 }
 450                 $this->debug( "-- User: " . $revision->user_text );
 451                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 452                 $this->debug( "-- Comment: " . $revision->comment );
 453                 $this->debug( "-- Text: " . $revision->text );
 454         }
 455
 456         /**
 457          * Notify the callback function of site info
 458          * @param array $siteInfo
 459          * @return bool|mixed
 460          */
 461         private function siteInfoCallback( $siteInfo ) {
 462                 if ( isset( $this->mSiteInfoCallback ) ) {
 463                         return call_user_func_array( $this->mSiteInfoCallback,
 464                                         [ $siteInfo, $this ] );
 465                 } else {
 466                         return false;
 467                 }
 468         }
 469
 470         /**
 471          * Notify the callback function when a new "<page>" is reached.
 472          * @param Title $title
 473          */
 474         function pageCallback( $title ) {
 475                 if ( isset( $this->mPageCallback ) ) {
 476                         call_user_func( $this->mPageCallback, $title );
 477                 }
 478         }
 479
 480         /**
 481          * Notify the callback function when a "</page>" is closed.
 482          * @param Title $title
 483          * @param ForeignTitle $foreignTitle
 484          * @param int $revCount
 485          * @param int $sucCount Number of revisions for which callback returned true
 486          * @param array $pageInfo Associative array of page information
 487          */
 488         private function pageOutCallback( $title, $foreignTitle, $revCount,
 489                         $sucCount, $pageInfo ) {
 490                 if ( isset( $this->mPageOutCallback ) ) {
 491                         $args = func_get_args();
 492                         call_user_func_array( $this->mPageOutCallback, $args );
 493                 }
 494         }
 495
 496         /**
 497          * Notify the callback function of a revision
 498          * @param WikiRevision $revision
 499          * @return bool|mixed
 500          */
 501         private function revisionCallback( $revision ) {
 502                 if ( isset( $this->mRevisionCallback ) ) {
 503                         return call_user_func_array( $this->mRevisionCallback,
 504                                         [ $revision, $this ] );
 505                 } else {
 506                         return false;
 507                 }
 508         }
 509
 510         /**
 511          * Notify the callback function of a new log item
 512          * @param WikiRevision $revision
 513          * @return bool|mixed
 514          */
 515         private function logItemCallback( $revision ) {
 516                 if ( isset( $this->mLogItemCallback ) ) {
 517                         return call_user_func_array( $this->mLogItemCallback,
 518                                         [ $revision, $this ] );
 519                 } else {
 520                         return false;
 521                 }
 522         }
 523
 524         /**
 525          * Retrieves the contents of the named attribute of the current element.
 526          * @param string $attr The name of the attribute
 527          * @return string The value of the attribute or an empty string if it is not set in the current
 528          * element.
 529          */
 530         public function nodeAttribute( $attr ) {
 531                 return $this->reader->getAttribute( $attr );
 532         }
 533
 534         /**
 535          * Shouldn't something like this be built-in to XMLReader?
 536          * Fetches text contents of the current element, assuming
 537          * no sub-elements or such scary things.
 538          * @return string
 539          * @private
 540          */
 541         public function nodeContents() {
 542                 if ( $this->reader->isEmptyElement ) {
 543                         return "";
 544                 }
 545                 $buffer = "";
 546                 while ( $this->reader->read() ) {
 547                         switch ( $this->reader->nodeType ) {
 548                                 case XMLReader::TEXT:
 549                                 case XMLReader::CDATA:
 550                                 case XMLReader::SIGNIFICANT_WHITESPACE:
 551                                         $buffer .= $this->reader->value;
 552                                         break;
 553                                 case XMLReader::END_ELEMENT:
 554                                         return $buffer;
 555                         }
 556                 }
 557
 558                 $this->reader->close();
 559                 return '';
 560         }
 561
 562         /**
 563          * Primary entry point
 564          * @throws Exception
 565          * @throws MWException
 566          * @return bool
 567          */
 568         public function doImport() {
 569                 // Calls to reader->read need to be wrapped in calls to
 570                 // libxml_disable_entity_loader() to avoid local file
 571                 // inclusion attacks (T48932).
 572                 $oldDisable = libxml_disable_entity_loader( true );
 573                 $this->reader->read();
 574
 575                 if ( $this->reader->localName != 'mediawiki' ) {
 576                         libxml_disable_entity_loader( $oldDisable );
 577                         throw new MWException( "Expected <mediawiki> tag, got " .
 578                                 $this->reader->localName );
 579                 }
 580                 $this->debug( "<mediawiki> tag is correct." );
 581
 582                 $this->debug( "Starting primary dump processing loop." );
 583
 584                 $keepReading = $this->reader->read();
 585                 $skip = false;
 586                 $rethrow = null;
 587                 $pageCount = 0;
 588                 try {
 589                         while ( $keepReading ) {
 590                                 $tag = $this->reader->localName;
 591                                 if ( $this->pageOffset ) {
 592                                         if ( $tag === 'page' ) {
 593                                                 $pageCount++;
 594                                         }
 595                                         if ( $pageCount < $this->pageOffset ) {
 596                                                 $keepReading = $this->reader->next();
 597                                                 continue;
 598                                         }
 599                                 }
 600                                 $type = $this->reader->nodeType;
 601
 602                                 if ( !Hooks::run( 'ImportHandleToplevelXMLTag', [ $this ] ) ) {
 603                                         // Do nothing
 604                                 } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) {
 605                                         break;
 606                                 } elseif ( $tag == 'siteinfo' ) {
 607                                         $this->handleSiteInfo();
 608                                 } elseif ( $tag == 'page' ) {
 609                                         $this->handlePage();
 610                                 } elseif ( $tag == 'logitem' ) {
 611                                         $this->handleLogItem();
 612                                 } elseif ( $tag != '#text' ) {
 613                                         $this->warn( "Unhandled top-level XML tag $tag" );
 614
 615                                         $skip = true;
 616                                 }
 617
 618                                 if ( $skip ) {
 619                                         $keepReading = $this->reader->next();
 620                                         $skip = false;
 621                                         $this->debug( "Skip" );
 622                                 } else {
 623                                         $keepReading = $this->reader->read();
 624                                 }
 625                         }
 626                 } catch ( Exception $ex ) {
 627                         $rethrow = $ex;
 628                 }
 629
 630                 // finally
 631                 libxml_disable_entity_loader( $oldDisable );
 632                 $this->reader->close();
 633
 634                 if ( $rethrow ) {
 635                         throw $rethrow;
 636                 }
 637
 638                 return true;
 639         }
 640
 641         private function handleSiteInfo() {
 642                 $this->debug( "Enter site info handler." );
 643                 $siteInfo = [];
 644
 645                 // Fields that can just be stuffed in the siteInfo object
 646                 $normalFields = [ 'sitename', 'base', 'generator', 'case' ];
 647
 648                 while ( $this->reader->read() ) {
 649                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 650                                         $this->reader->localName == 'siteinfo' ) {
 651                                 break;
 652                         }
 653
 654                         $tag = $this->reader->localName;
 655
 656                         if ( $tag == 'namespace' ) {
 657                                 $this->foreignNamespaces[$this->nodeAttribute( 'key' )] =
 658                                         $this->nodeContents();
 659                         } elseif ( in_array( $tag, $normalFields ) ) {
 660                                 $siteInfo[$tag] = $this->nodeContents();
 661                         }
 662                 }
 663
 664                 $siteInfo['_namespaces'] = $this->foreignNamespaces;
 665                 $this->siteInfoCallback( $siteInfo );
 666         }
 667
 668         private function handleLogItem() {
 669                 $this->debug( "Enter log item handler." );
 670                 $logInfo = [];
 671
 672                 // Fields that can just be stuffed in the pageInfo object
 673                 $normalFields = [ 'id', 'comment', 'type', 'action', 'timestamp',
 674                                         'logtitle', 'params' ];
 675
 676                 while ( $this->reader->read() ) {
 677                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 678                                         $this->reader->localName == 'logitem' ) {
 679                                 break;
 680                         }
 681
 682                         $tag = $this->reader->localName;
 683
 684                         if ( !Hooks::run( 'ImportHandleLogItemXMLTag', [
 685                                 $this, $logInfo
 686                         ] ) ) {
 687                                 // Do nothing
 688                         } elseif ( in_array( $tag, $normalFields ) ) {
 689                                 $logInfo[$tag] = $this->nodeContents();
 690                         } elseif ( $tag == 'contributor' ) {
 691                                 $logInfo['contributor'] = $this->handleContributor();
 692                         } elseif ( $tag != '#text' ) {
 693                                 $this->warn( "Unhandled log-item XML tag $tag" );
 694                         }
 695                 }
 696
 697                 $this->processLogItem( $logInfo );
 698         }
 699
 700         /**
 701          * @param array $logInfo
 702          * @return bool|mixed
 703          */
 704         private function processLogItem( $logInfo ) {
 705                 $revision = new WikiRevision( $this->config );
 706
 707                 if ( isset( $logInfo['id'] ) ) {
 708                         $revision->setID( $logInfo['id'] );
 709                 }
 710                 $revision->setType( $logInfo['type'] );
 711                 $revision->setAction( $logInfo['action'] );
 712                 if ( isset( $logInfo['timestamp'] ) ) {
 713                         $revision->setTimestamp( $logInfo['timestamp'] );
 714                 }
 715                 if ( isset( $logInfo['params'] ) ) {
 716                         $revision->setParams( $logInfo['params'] );
 717                 }
 718                 if ( isset( $logInfo['logtitle'] ) ) {
 719                         // @todo Using Title for non-local titles is a recipe for disaster.
 720                         // We should use ForeignTitle here instead.
 721                         $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 722                 }
 723
 724                 $revision->setNoUpdates( $this->mNoUpdates );
 725
 726                 if ( isset( $logInfo['comment'] ) ) {
 727                         $revision->setComment( $logInfo['comment'] );
 728                 }
 729
 730                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 731                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 732                 }
 733
 734                 if ( !isset( $logInfo['contributor']['username'] ) ) {
 735                         $revision->setUsername( $this->externalUserNames->addPrefix( 'Unknown user' ) );
 736                 } else {
 737                         $revision->setUsername(
 738                                 $this->externalUserNames->applyPrefix( $logInfo['contributor']['username'] )
 739                         );
 740                 }
 741
 742                 return $this->logItemCallback( $revision );
 743         }
 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', $title->getPrefixedText() );
1105                         return false;
1106                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
1107                         # Do not import if the importing wiki user cannot edit this page
1108                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
1109                         return false;
1110                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
1111                         # Do not import if the importing wiki user cannot create this page
1112                         $this->notice( 'import-error-create', $title->getPrefixedText() );
1113                         return false;
1114                 }
1115
1116                 return [ $title, $foreignTitle ];
1117         }
1118 }