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