includes/import/WikiImporter.php

   1 <?php
   2 /**
   3  * MediaWiki page data importer.
   4  *
   5  * Copyright © 2003,2005 Brion Vibber <brion@pobox.com>
   6  * https://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup SpecialPage
  25  */
  26 use MediaWiki\MediaWikiServices;
  27
  28 /**
  29  * XML file reader for the page data importer.
  30  *
  31  * implements Special:Import
  32  * @ingroup SpecialPage
  33  */
  34 class WikiImporter {
  35         private $reader = null;
  36         private $foreignNamespaces = null;
  37         private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
  38         private $mSiteInfoCallback, $mPageOutCallback;
  39         private $mNoticeCallback, $mDebug;
  40         private $mImportUploads, $mImageBasePath;
  41         private $mNoUpdates = false;
  42         /** @var Config */
  43         private $config;
  44         /** @var ImportTitleFactory */
  45         private $importTitleFactory;
  46         /** @var array */
  47         private $countableCache = [];
  48
  49         /**
  50          * Creates an ImportXMLReader drawing from the source provided
  51          * @param ImportSource $source
  52          * @param Config $config
  53          * @throws Exception
  54          */
  55         function __construct( ImportSource $source, Config $config = null ) {
  56                 if ( !class_exists( 'XMLReader' ) ) {
  57                         throw new Exception( 'Import requires PHP to have been compiled with libxml support' );
  58                 }
  59
  60                 $this->reader = new XMLReader();
  61                 if ( !$config ) {
  62                         wfDeprecated( __METHOD__ . ' without a Config instance', '1.25' );
  63                         $config = MediaWikiServices::getInstance()->getMainConfig();
  64                 }
  65                 $this->config = $config;
  66
  67                 if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) {
  68                         stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  69                 }
  70                 $id = UploadSourceAdapter::registerSource( $source );
  71
  72                 // Enable the entity loader, as it is needed for loading external URLs via
  73                 // XMLReader::open (T86036)
  74                 $oldDisable = libxml_disable_entity_loader( false );
  75                 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
  76                         $status = $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  77                 } else {
  78                         $status = $this->reader->open( "uploadsource://$id" );
  79                 }
  80                 if ( !$status ) {
  81                         $error = libxml_get_last_error();
  82                         libxml_disable_entity_loader( $oldDisable );
  83                         throw new MWException( 'Encountered an internal error while initializing WikiImporter object: ' .
  84                                 $error->message );
  85                 }
  86                 libxml_disable_entity_loader( $oldDisable );
  87
  88                 // Default callbacks
  89                 $this->setPageCallback( [ $this, 'beforeImportPage' ] );
  90                 $this->setRevisionCallback( [ $this, "importRevision" ] );
  91                 $this->setUploadCallback( [ $this, 'importUpload' ] );
  92                 $this->setLogItemCallback( [ $this, 'importLogItem' ] );
  93                 $this->setPageOutCallback( [ $this, 'finishImportPage' ] );
  94
  95                 $this->importTitleFactory = new NaiveImportTitleFactory();
  96         }
  97
  98         /**
  99          * @return null|XMLReader
 100          */
 101         public function getReader() {
 102                 return $this->reader;
 103         }
 104
 105         public function throwXmlError( $err ) {
 106                 $this->debug( "FAILURE: $err" );
 107                 wfDebug( "WikiImporter XML error: $err\n" );
 108         }
 109
 110         public function debug( $data ) {
 111                 if ( $this->mDebug ) {
 112                         wfDebug( "IMPORT: $data\n" );
 113                 }
 114         }
 115
 116         public function warn( $data ) {
 117                 wfDebug( "IMPORT: $data\n" );
 118         }
 119
 120         public function notice( $msg /*, $param, ...*/ ) {
 121                 $params = func_get_args();
 122                 array_shift( $params );
 123
 124                 if ( is_callable( $this->mNoticeCallback ) ) {
 125                         call_user_func( $this->mNoticeCallback, $msg, $params );
 126                 } else { # No ImportReporter -> CLI
 127                         echo wfMessage( $msg, $params )->text() . "\n";
 128                 }
 129         }
 130
 131         /**
 132          * Set debug mode...
 133          * @param bool $debug
 134          */
 135         function setDebug( $debug ) {
 136                 $this->mDebug = $debug;
 137         }
 138
 139         /**
 140          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
 141          * @param bool $noupdates
 142          */
 143         function setNoUpdates( $noupdates ) {
 144                 $this->mNoUpdates = $noupdates;
 145         }
 146
 147         /**
 148          * Set a callback that displays notice messages
 149          *
 150          * @param callable $callback
 151          * @return callable
 152          */
 153         public function setNoticeCallback( $callback ) {
 154                 return wfSetVar( $this->mNoticeCallback, $callback );
 155         }
 156
 157         /**
 158          * Sets the action to perform as each new page in the stream is reached.
 159          * @param callable $callback
 160          * @return callable
 161          */
 162         public function setPageCallback( $callback ) {
 163                 $previous = $this->mPageCallback;
 164                 $this->mPageCallback = $callback;
 165                 return $previous;
 166         }
 167
 168         /**
 169          * Sets the action to perform as each page in the stream is completed.
 170          * Callback accepts the page title (as a Title object), a second object
 171          * with the original title form (in case it's been overridden into a
 172          * local namespace), and a count of revisions.
 173          *
 174          * @param callable $callback
 175          * @return callable
 176          */
 177         public function setPageOutCallback( $callback ) {
 178                 $previous = $this->mPageOutCallback;
 179                 $this->mPageOutCallback = $callback;
 180                 return $previous;
 181         }
 182
 183         /**
 184          * Sets the action to perform as each page revision is reached.
 185          * @param callable $callback
 186          * @return callable
 187          */
 188         public function setRevisionCallback( $callback ) {
 189                 $previous = $this->mRevisionCallback;
 190                 $this->mRevisionCallback = $callback;
 191                 return $previous;
 192         }
 193
 194         /**
 195          * Sets the action to perform as each file upload version is reached.
 196          * @param callable $callback
 197          * @return callable
 198          */
 199         public function setUploadCallback( $callback ) {
 200                 $previous = $this->mUploadCallback;
 201                 $this->mUploadCallback = $callback;
 202                 return $previous;
 203         }
 204
 205         /**
 206          * Sets the action to perform as each log item reached.
 207          * @param callable $callback
 208          * @return callable
 209          */
 210         public function setLogItemCallback( $callback ) {
 211                 $previous = $this->mLogItemCallback;
 212                 $this->mLogItemCallback = $callback;
 213                 return $previous;
 214         }
 215
 216         /**
 217          * Sets the action to perform when site info is encountered
 218          * @param callable $callback
 219          * @return callable
 220          */
 221         public function setSiteInfoCallback( $callback ) {
 222                 $previous = $this->mSiteInfoCallback;
 223                 $this->mSiteInfoCallback = $callback;
 224                 return $previous;
 225         }
 226
 227         /**
 228          * Sets the factory object to use to convert ForeignTitle objects into local
 229          * Title objects
 230          * @param ImportTitleFactory $factory
 231          */
 232         public function setImportTitleFactory( $factory ) {
 233                 $this->importTitleFactory = $factory;
 234         }
 235
 236         /**
 237          * Set a target namespace to override the defaults
 238          * @param null|int $namespace
 239          * @return bool
 240          */
 241         public function setTargetNamespace( $namespace ) {
 242                 if ( is_null( $namespace ) ) {
 243                         // Don't override namespaces
 244                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 245                         return true;
 246                 } elseif (
 247                         $namespace >= 0 &&
 248                         MWNamespace::exists( intval( $namespace ) )
 249                 ) {
 250                         $namespace = intval( $namespace );
 251                         $this->setImportTitleFactory( new NamespaceImportTitleFactory( $namespace ) );
 252                         return true;
 253                 } else {
 254                         return false;
 255                 }
 256         }
 257
 258         /**
 259          * Set a target root page under which all pages are imported
 260          * @param null|string $rootpage
 261          * @return Status
 262          */
 263         public function setTargetRootPage( $rootpage ) {
 264                 $status = Status::newGood();
 265                 if ( is_null( $rootpage ) ) {
 266                         // No rootpage
 267                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 268                 } elseif ( $rootpage !== '' ) {
 269                         $rootpage = rtrim( $rootpage, '/' ); // avoid double slashes
 270                         $title = Title::newFromText( $rootpage );
 271
 272                         if ( !$title || $title->isExternal() ) {
 273                                 $status->fatal( 'import-rootpage-invalid' );
 274                         } else {
 275                                 if ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 276                                         global $wgContLang;
 277
 278                                         $displayNSText = $title->getNamespace() == NS_MAIN
 279                                                 ? wfMessage( 'blanknamespace' )->text()
 280                                                 : $wgContLang->getNsText( $title->getNamespace() );
 281                                         $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 282                                 } else {
 283                                         // set namespace to 'all', so the namespace check in processTitle() can pass
 284                                         $this->setTargetNamespace( null );
 285                                         $this->setImportTitleFactory( new SubpageImportTitleFactory( $title ) );
 286                                 }
 287                         }
 288                 }
 289                 return $status;
 290         }
 291
 292         /**
 293          * @param string $dir
 294          */
 295         public function setImageBasePath( $dir ) {
 296                 $this->mImageBasePath = $dir;
 297         }
 298
 299         /**
 300          * @param bool $import
 301          */
 302         public function setImportUploads( $import ) {
 303                 $this->mImportUploads = $import;
 304         }
 305
 306         /**
 307          * Default per-page callback. Sets up some things related to site statistics
 308          * @param array $titleAndForeignTitle Two-element array, with Title object at
 309          * index 0 and ForeignTitle object at index 1
 310          * @return bool
 311          */
 312         public function beforeImportPage( $titleAndForeignTitle ) {
 313                 $title = $titleAndForeignTitle[0];
 314                 $page = WikiPage::factory( $title );
 315                 $this->countableCache['title_' . $title->getPrefixedText()] = $page->isCountable();
 316                 return true;
 317         }
 318
 319         /**
 320          * Default per-revision callback, performs the import.
 321          * @param WikiRevision $revision
 322          * @return bool
 323          */
 324         public function importRevision( $revision ) {
 325                 if ( !$revision->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
 326                         $this->notice( 'import-error-bad-location',
 327                                 $revision->getTitle()->getPrefixedText(),
 328                                 $revision->getID(),
 329                                 $revision->getModel(),
 330                                 $revision->getFormat() );
 331
 332                         return false;
 333                 }
 334
 335                 try {
 336                         return $revision->importOldRevision();
 337                 } catch ( MWContentSerializationException $ex ) {
 338                         $this->notice( 'import-error-unserialize',
 339                                 $revision->getTitle()->getPrefixedText(),
 340                                 $revision->getID(),
 341                                 $revision->getModel(),
 342                                 $revision->getFormat() );
 343                 }
 344
 345                 return false;
 346         }
 347
 348         /**
 349          * Default per-revision callback, performs the import.
 350          * @param WikiRevision $revision
 351          * @return bool
 352          */
 353         public function importLogItem( $revision ) {
 354                 return $revision->importLogItem();
 355         }
 356
 357         /**
 358          * Dummy for now...
 359          * @param WikiRevision $revision
 360          * @return bool
 361          */
 362         public function importUpload( $revision ) {
 363                 return $revision->importUpload();
 364         }
 365
 366         /**
 367          * Mostly for hook use
 368          * @param Title $title
 369          * @param ForeignTitle $foreignTitle
 370          * @param int $revCount
 371          * @param int $sRevCount
 372          * @param array $pageInfo
 373          * @return bool
 374          */
 375         public function finishImportPage( $title, $foreignTitle, $revCount,
 376                         $sRevCount, $pageInfo ) {
 377
 378                 // Update article count statistics (T42009)
 379                 // The normal counting logic in WikiPage->doEditUpdates() is designed for
 380                 // one-revision-at-a-time editing, not bulk imports. In this situation it
 381                 // suffers from issues of replica DB lag. We let WikiPage handle the total page
 382                 // and revision count, and we implement our own custom logic for the
 383                 // article (content page) count.
 384                 $page = WikiPage::factory( $title );
 385                 $page->loadPageData( 'fromdbmaster' );
 386                 $content = $page->getContent();
 387                 if ( $content === null ) {
 388                         wfDebug( __METHOD__ . ': Skipping article count adjustment for ' . $title .
 389                                 ' because WikiPage::getContent() returned null' );
 390                 } else {
 391                         $editInfo = $page->prepareContentForEdit( $content );
 392                         $countKey = 'title_' . $title->getPrefixedText();
 393                         $countable = $page->isCountable( $editInfo );
 394                         if ( array_key_exists( $countKey, $this->countableCache ) &&
 395                                 $countable != $this->countableCache[$countKey] ) {
 396                                 DeferredUpdates::addUpdate( SiteStatsUpdate::factory( [
 397                                         'articles' => ( (int)$countable - (int)$this->countableCache[$countKey] )
 398                                 ] ) );
 399                         }
 400                 }
 401
 402                 $args = func_get_args();
 403                 return Hooks::run( 'AfterImportPage', $args );
 404         }
 405
 406         /**
 407          * Alternate per-revision callback, for debugging.
 408          * @param WikiRevision $revision
 409          */
 410         public function debugRevisionHandler( &$revision ) {
 411                 $this->debug( "Got revision:" );
 412                 if ( is_object( $revision->title ) ) {
 413                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 414                 } else {
 415                         $this->debug( "-- Title: <invalid>" );
 416                 }
 417                 $this->debug( "-- User: " . $revision->user_text );
 418                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 419                 $this->debug( "-- Comment: " . $revision->comment );
 420                 $this->debug( "-- Text: " . $revision->text );
 421         }
 422
 423         /**
 424          * Notify the callback function of site info
 425          * @param array $siteInfo
 426          * @return bool|mixed
 427          */
 428         private function siteInfoCallback( $siteInfo ) {
 429                 if ( isset( $this->mSiteInfoCallback ) ) {
 430                         return call_user_func_array( $this->mSiteInfoCallback,
 431                                         [ $siteInfo, $this ] );
 432                 } else {
 433                         return false;
 434                 }
 435         }
 436
 437         /**
 438          * Notify the callback function when a new "<page>" is reached.
 439          * @param Title $title
 440          */
 441         function pageCallback( $title ) {
 442                 if ( isset( $this->mPageCallback ) ) {
 443                         call_user_func( $this->mPageCallback, $title );
 444                 }
 445         }
 446
 447         /**
 448          * Notify the callback function when a "</page>" is closed.
 449          * @param Title $title
 450          * @param ForeignTitle $foreignTitle
 451          * @param int $revCount
 452          * @param int $sucCount Number of revisions for which callback returned true
 453          * @param array $pageInfo Associative array of page information
 454          */
 455         private function pageOutCallback( $title, $foreignTitle, $revCount,
 456                         $sucCount, $pageInfo ) {
 457                 if ( isset( $this->mPageOutCallback ) ) {
 458                         $args = func_get_args();
 459                         call_user_func_array( $this->mPageOutCallback, $args );
 460                 }
 461         }
 462
 463         /**
 464          * Notify the callback function of a revision
 465          * @param WikiRevision $revision
 466          * @return bool|mixed
 467          */
 468         private function revisionCallback( $revision ) {
 469                 if ( isset( $this->mRevisionCallback ) ) {
 470                         return call_user_func_array( $this->mRevisionCallback,
 471                                         [ $revision, $this ] );
 472                 } else {
 473                         return false;
 474                 }
 475         }
 476
 477         /**
 478          * Notify the callback function of a new log item
 479          * @param WikiRevision $revision
 480          * @return bool|mixed
 481          */
 482         private function logItemCallback( $revision ) {
 483                 if ( isset( $this->mLogItemCallback ) ) {
 484                         return call_user_func_array( $this->mLogItemCallback,
 485                                         [ $revision, $this ] );
 486                 } else {
 487                         return false;
 488                 }
 489         }
 490
 491         /**
 492          * Retrieves the contents of the named attribute of the current element.
 493          * @param string $attr The name of the attribute
 494          * @return string The value of the attribute or an empty string if it is not set in the current
 495          * element.
 496          */
 497         public function nodeAttribute( $attr ) {
 498                 return $this->reader->getAttribute( $attr );
 499         }
 500
 501         /**
 502          * Shouldn't something like this be built-in to XMLReader?
 503          * Fetches text contents of the current element, assuming
 504          * no sub-elements or such scary things.
 505          * @return string
 506          * @access private
 507          */
 508         public function nodeContents() {
 509                 if ( $this->reader->isEmptyElement ) {
 510                         return "";
 511                 }
 512                 $buffer = "";
 513                 while ( $this->reader->read() ) {
 514                         switch ( $this->reader->nodeType ) {
 515                         case XMLReader::TEXT:
 516                         case XMLReader::CDATA:
 517                         case XMLReader::SIGNIFICANT_WHITESPACE:
 518                                 $buffer .= $this->reader->value;
 519                                 break;
 520                         case XMLReader::END_ELEMENT:
 521                                 return $buffer;
 522                         }
 523                 }
 524
 525                 $this->reader->close();
 526                 return '';
 527         }
 528
 529         /**
 530          * Primary entry point
 531          * @throws MWException
 532          * @return bool
 533          */
 534         public function doImport() {
 535                 // Calls to reader->read need to be wrapped in calls to
 536                 // libxml_disable_entity_loader() to avoid local file
 537                 // inclusion attacks (bug 46932).
 538                 $oldDisable = libxml_disable_entity_loader( true );
 539                 $this->reader->read();
 540
 541                 if ( $this->reader->localName != 'mediawiki' ) {
 542                         libxml_disable_entity_loader( $oldDisable );
 543                         throw new MWException( "Expected <mediawiki> tag, got " .
 544                                 $this->reader->localName );
 545                 }
 546                 $this->debug( "<mediawiki> tag is correct." );
 547
 548                 $this->debug( "Starting primary dump processing loop." );
 549
 550                 $keepReading = $this->reader->read();
 551                 $skip = false;
 552                 $rethrow = null;
 553                 try {
 554                         while ( $keepReading ) {
 555                                 $tag = $this->reader->localName;
 556                                 $type = $this->reader->nodeType;
 557
 558                                 if ( !Hooks::run( 'ImportHandleToplevelXMLTag', [ $this ] ) ) {
 559                                         // Do nothing
 560                                 } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) {
 561                                         break;
 562                                 } elseif ( $tag == 'siteinfo' ) {
 563                                         $this->handleSiteInfo();
 564                                 } elseif ( $tag == 'page' ) {
 565                                         $this->handlePage();
 566                                 } elseif ( $tag == 'logitem' ) {
 567                                         $this->handleLogItem();
 568                                 } elseif ( $tag != '#text' ) {
 569                                         $this->warn( "Unhandled top-level XML tag $tag" );
 570
 571                                         $skip = true;
 572                                 }
 573
 574                                 if ( $skip ) {
 575                                         $keepReading = $this->reader->next();
 576                                         $skip = false;
 577                                         $this->debug( "Skip" );
 578                                 } else {
 579                                         $keepReading = $this->reader->read();
 580                                 }
 581                         }
 582                 } catch ( Exception $ex ) {
 583                         $rethrow = $ex;
 584                 }
 585
 586                 // finally
 587                 libxml_disable_entity_loader( $oldDisable );
 588                 $this->reader->close();
 589
 590                 if ( $rethrow ) {
 591                         throw $rethrow;
 592                 }
 593
 594                 return true;
 595         }
 596
 597         private function handleSiteInfo() {
 598                 $this->debug( "Enter site info handler." );
 599                 $siteInfo = [];
 600
 601                 // Fields that can just be stuffed in the siteInfo object
 602                 $normalFields = [ 'sitename', 'base', 'generator', 'case' ];
 603
 604                 while ( $this->reader->read() ) {
 605                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 606                                         $this->reader->localName == 'siteinfo' ) {
 607                                 break;
 608                         }
 609
 610                         $tag = $this->reader->localName;
 611
 612                         if ( $tag == 'namespace' ) {
 613                                 $this->foreignNamespaces[$this->nodeAttribute( 'key' )] =
 614                                         $this->nodeContents();
 615                         } elseif ( in_array( $tag, $normalFields ) ) {
 616                                 $siteInfo[$tag] = $this->nodeContents();
 617                         }
 618                 }
 619
 620                 $siteInfo['_namespaces'] = $this->foreignNamespaces;
 621                 $this->siteInfoCallback( $siteInfo );
 622         }
 623
 624         private function handleLogItem() {
 625                 $this->debug( "Enter log item handler." );
 626                 $logInfo = [];
 627
 628                 // Fields that can just be stuffed in the pageInfo object
 629                 $normalFields = [ 'id', 'comment', 'type', 'action', 'timestamp',
 630                                         'logtitle', 'params' ];
 631
 632                 while ( $this->reader->read() ) {
 633                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 634                                         $this->reader->localName == 'logitem' ) {
 635                                 break;
 636                         }
 637
 638                         $tag = $this->reader->localName;
 639
 640                         if ( !Hooks::run( 'ImportHandleLogItemXMLTag', [
 641                                 $this, $logInfo
 642                         ] ) ) {
 643                                 // Do nothing
 644                         } elseif ( in_array( $tag, $normalFields ) ) {
 645                                 $logInfo[$tag] = $this->nodeContents();
 646                         } elseif ( $tag == 'contributor' ) {
 647                                 $logInfo['contributor'] = $this->handleContributor();
 648                         } elseif ( $tag != '#text' ) {
 649                                 $this->warn( "Unhandled log-item XML tag $tag" );
 650                         }
 651                 }
 652
 653                 $this->processLogItem( $logInfo );
 654         }
 655
 656         /**
 657          * @param array $logInfo
 658          * @return bool|mixed
 659          */
 660         private function processLogItem( $logInfo ) {
 661
 662                 $revision = new WikiRevision( $this->config );
 663
 664                 if ( isset( $logInfo['id'] ) ) {
 665                         $revision->setID( $logInfo['id'] );
 666                 }
 667                 $revision->setType( $logInfo['type'] );
 668                 $revision->setAction( $logInfo['action'] );
 669                 if ( isset( $logInfo['timestamp'] ) ) {
 670                         $revision->setTimestamp( $logInfo['timestamp'] );
 671                 }
 672                 if ( isset( $logInfo['params'] ) ) {
 673                         $revision->setParams( $logInfo['params'] );
 674                 }
 675                 if ( isset( $logInfo['logtitle'] ) ) {
 676                         // @todo Using Title for non-local titles is a recipe for disaster.
 677                         // We should use ForeignTitle here instead.
 678                         $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 679                 }
 680
 681                 $revision->setNoUpdates( $this->mNoUpdates );
 682
 683                 if ( isset( $logInfo['comment'] ) ) {
 684                         $revision->setComment( $logInfo['comment'] );
 685                 }
 686
 687                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 688                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 689                 }
 690
 691                 if ( !isset( $logInfo['contributor']['username'] ) ) {
 692                         $revision->setUsername( 'Unknown user' );
 693                 } else {
 694                         $revision->setUsername( $logInfo['contributor']['username'] );
 695                 }
 696
 697                 return $this->logItemCallback( $revision );
 698         }
 699
 700         private function handlePage() {
 701                 // Handle page data.
 702                 $this->debug( "Enter page handler." );
 703                 $pageInfo = [ 'revisionCount' => 0, 'successfulRevisionCount' => 0 ];
 704
 705                 // Fields that can just be stuffed in the pageInfo object
 706                 $normalFields = [ 'title', 'ns', 'id', 'redirect', 'restrictions' ];
 707
 708                 $skip = false;
 709                 $badTitle = false;
 710
 711                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 712                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 713                                         $this->reader->localName == 'page' ) {
 714                                 break;
 715                         }
 716
 717                         $skip = false;
 718
 719                         $tag = $this->reader->localName;
 720
 721                         if ( $badTitle ) {
 722                                 // The title is invalid, bail out of this page
 723                                 $skip = true;
 724                         } elseif ( !Hooks::run( 'ImportHandlePageXMLTag', [ $this,
 725                                                 &$pageInfo ] ) ) {
 726                                 // Do nothing
 727                         } elseif ( in_array( $tag, $normalFields ) ) {
 728                                 // An XML snippet:
 729                                 // <page>
 730                                 //     <id>123</id>
 731                                 //     <title>Page</title>
 732                                 //     <redirect title="NewTitle"/>
 733                                 //     ...
 734                                 // Because the redirect tag is built differently, we need special handling for that case.
 735                                 if ( $tag == 'redirect' ) {
 736                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 737                                 } else {
 738                                         $pageInfo[$tag] = $this->nodeContents();
 739                                 }
 740                         } elseif ( $tag == 'revision' || $tag == 'upload' ) {
 741                                 if ( !isset( $title ) ) {
 742                                         $title = $this->processTitle( $pageInfo['title'],
 743                                                 isset( $pageInfo['ns'] ) ? $pageInfo['ns'] : null );
 744
 745                                         // $title is either an array of two titles or false.
 746                                         if ( is_array( $title ) ) {
 747                                                 $this->pageCallback( $title );
 748                                                 list( $pageInfo['_title'], $foreignTitle ) = $title;
 749                                         } else {
 750                                                 $badTitle = true;
 751                                                 $skip = true;
 752                                         }
 753                                 }
 754
 755                                 if ( $title ) {
 756                                         if ( $tag == 'revision' ) {
 757                                                 $this->handleRevision( $pageInfo );
 758                                         } else {
 759                                                 $this->handleUpload( $pageInfo );
 760                                         }
 761                                 }
 762                         } elseif ( $tag != '#text' ) {
 763                                 $this->warn( "Unhandled page XML tag $tag" );
 764                                 $skip = true;
 765                         }
 766                 }
 767
 768                 // @note $pageInfo is only set if a valid $title is processed above with
 769                 //       no error. If we have a valid $title, then pageCallback is called
 770                 //       above, $pageInfo['title'] is set and we do pageOutCallback here.
 771                 //       If $pageInfo['_title'] is not set, then $foreignTitle is also not
 772                 //       set since they both come from $title above.
 773                 if ( array_key_exists( '_title', $pageInfo ) ) {
 774                         $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
 775                                         $pageInfo['revisionCount'],
 776                                         $pageInfo['successfulRevisionCount'],
 777                                         $pageInfo );
 778                 }
 779         }
 780
 781         /**
 782          * @param array $pageInfo
 783          */
 784         private function handleRevision( &$pageInfo ) {
 785                 $this->debug( "Enter revision handler" );
 786                 $revisionInfo = [];
 787
 788                 $normalFields = [ 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' ];
 789
 790                 $skip = false;
 791
 792                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 793                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 794                                         $this->reader->localName == 'revision' ) {
 795                                 break;
 796                         }
 797
 798                         $tag = $this->reader->localName;
 799
 800                         if ( !Hooks::run( 'ImportHandleRevisionXMLTag', [
 801                                 $this, $pageInfo, $revisionInfo
 802                         ] ) ) {
 803                                 // Do nothing
 804                         } elseif ( in_array( $tag, $normalFields ) ) {
 805                                 $revisionInfo[$tag] = $this->nodeContents();
 806                         } elseif ( $tag == 'contributor' ) {
 807                                 $revisionInfo['contributor'] = $this->handleContributor();
 808                         } elseif ( $tag != '#text' ) {
 809                                 $this->warn( "Unhandled revision XML tag $tag" );
 810                                 $skip = true;
 811                         }
 812                 }
 813
 814                 $pageInfo['revisionCount']++;
 815                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 816                         $pageInfo['successfulRevisionCount']++;
 817                 }
 818         }
 819
 820         /**
 821          * @param array $pageInfo
 822          * @param array $revisionInfo
 823          * @return bool|mixed
 824          */
 825         private function processRevision( $pageInfo, $revisionInfo ) {
 826                 global $wgMaxArticleSize;
 827
 828                 // Make sure revisions won't violate $wgMaxArticleSize, which could lead to
 829                 // database errors and instability. Testing for revisions with only listed
 830                 // content models, as other content models might use serialization formats
 831                 // which aren't checked against $wgMaxArticleSize.
 832                 if ( ( !isset( $revisionInfo['model'] ) ||
 833                         in_array( $revisionInfo['model'], [
 834                                 'wikitext',
 835                                 'css',
 836                                 'json',
 837                                 'javascript',
 838                                 'text',
 839                                 ''
 840                         ] ) ) &&
 841                         strlen( $revisionInfo['text'] ) > $wgMaxArticleSize * 1024
 842                 ) {
 843                         throw new MWException( 'The text of ' .
 844                                 ( isset( $revisionInfo['id'] ) ?
 845                                         "the revision with ID $revisionInfo[id]" :
 846                                         'a revision'
 847                                 ) . " exceeds the maximum allowable size ($wgMaxArticleSize KB)" );
 848                 }
 849
 850                 $revision = new WikiRevision( $this->config );
 851
 852                 if ( isset( $revisionInfo['id'] ) ) {
 853                         $revision->setID( $revisionInfo['id'] );
 854                 }
 855                 if ( isset( $revisionInfo['model'] ) ) {
 856                         $revision->setModel( $revisionInfo['model'] );
 857                 }
 858                 if ( isset( $revisionInfo['format'] ) ) {
 859                         $revision->setFormat( $revisionInfo['format'] );
 860                 }
 861                 $revision->setTitle( $pageInfo['_title'] );
 862
 863                 if ( isset( $revisionInfo['text'] ) ) {
 864                         $handler = $revision->getContentHandler();
 865                         $text = $handler->importTransform(
 866                                 $revisionInfo['text'],
 867                                 $revision->getFormat() );
 868
 869                         $revision->setText( $text );
 870                 }
 871                 if ( isset( $revisionInfo['timestamp'] ) ) {
 872                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 873                 } else {
 874                         $revision->setTimestamp( wfTimestampNow() );
 875                 }
 876
 877                 if ( isset( $revisionInfo['comment'] ) ) {
 878                         $revision->setComment( $revisionInfo['comment'] );
 879                 }
 880
 881                 if ( isset( $revisionInfo['minor'] ) ) {
 882                         $revision->setMinor( true );
 883                 }
 884                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 885                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 886                 } elseif ( isset( $revisionInfo['contributor']['username'] ) ) {
 887                         $revision->setUsername( $revisionInfo['contributor']['username'] );
 888                 } else {
 889                         $revision->setUsername( 'Unknown user' );
 890                 }
 891                 $revision->setNoUpdates( $this->mNoUpdates );
 892
 893                 return $this->revisionCallback( $revision );
 894         }
 895
 896         /**
 897          * @param array $pageInfo
 898          * @return mixed
 899          */
 900         private function handleUpload( &$pageInfo ) {
 901                 $this->debug( "Enter upload handler" );
 902                 $uploadInfo = [];
 903
 904                 $normalFields = [ 'timestamp', 'comment', 'filename', 'text',
 905                                         'src', 'size', 'sha1base36', 'archivename', 'rel' ];
 906
 907                 $skip = false;
 908
 909                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 910                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 911                                         $this->reader->localName == 'upload' ) {
 912                                 break;
 913                         }
 914
 915                         $tag = $this->reader->localName;
 916
 917                         if ( !Hooks::run( 'ImportHandleUploadXMLTag', [
 918                                 $this, $pageInfo
 919                         ] ) ) {
 920                                 // Do nothing
 921                         } elseif ( in_array( $tag, $normalFields ) ) {
 922                                 $uploadInfo[$tag] = $this->nodeContents();
 923                         } elseif ( $tag == 'contributor' ) {
 924                                 $uploadInfo['contributor'] = $this->handleContributor();
 925                         } elseif ( $tag == 'contents' ) {
 926                                 $contents = $this->nodeContents();
 927                                 $encoding = $this->reader->getAttribute( 'encoding' );
 928                                 if ( $encoding === 'base64' ) {
 929                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 930                                         $uploadInfo['isTempSrc'] = true;
 931                                 }
 932                         } elseif ( $tag != '#text' ) {
 933                                 $this->warn( "Unhandled upload XML tag $tag" );
 934                                 $skip = true;
 935                         }
 936                 }
 937
 938                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 939                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 940                         if ( file_exists( $path ) ) {
 941                                 $uploadInfo['fileSrc'] = $path;
 942                                 $uploadInfo['isTempSrc'] = false;
 943                         }
 944                 }
 945
 946                 if ( $this->mImportUploads ) {
 947                         return $this->processUpload( $pageInfo, $uploadInfo );
 948                 }
 949         }
 950
 951         /**
 952          * @param string $contents
 953          * @return string
 954          */
 955         private function dumpTemp( $contents ) {
 956                 $filename = tempnam( wfTempDir(), 'importupload' );
 957                 file_put_contents( $filename, $contents );
 958                 return $filename;
 959         }
 960
 961         /**
 962          * @param array $pageInfo
 963          * @param array $uploadInfo
 964          * @return mixed
 965          */
 966         private function processUpload( $pageInfo, $uploadInfo ) {
 967                 $revision = new WikiRevision( $this->config );
 968                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 969
 970                 $revision->setTitle( $pageInfo['_title'] );
 971                 $revision->setID( $pageInfo['id'] );
 972                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 973                 $revision->setText( $text );
 974                 $revision->setFilename( $uploadInfo['filename'] );
 975                 if ( isset( $uploadInfo['archivename'] ) ) {
 976                         $revision->setArchiveName( $uploadInfo['archivename'] );
 977                 }
 978                 $revision->setSrc( $uploadInfo['src'] );
 979                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 980                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 981                                 !empty( $uploadInfo['isTempSrc'] ) );
 982                 }
 983                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 984                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 985                 }
 986                 $revision->setSize( intval( $uploadInfo['size'] ) );
 987                 $revision->setComment( $uploadInfo['comment'] );
 988
 989                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 990                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 991                 }
 992                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 993                         $revision->setUsername( $uploadInfo['contributor']['username'] );
 994                 }
 995                 $revision->setNoUpdates( $this->mNoUpdates );
 996
 997                 return call_user_func( $this->mUploadCallback, $revision );
 998         }
 999
1000         /**
1001          * @return array
1002          */
1003         private function handleContributor() {
1004                 $fields = [ 'id', 'ip', 'username' ];
1005                 $info = [];
1006
1007                 if ( $this->reader->isEmptyElement ) {
1008                         return $info;
1009                 }
1010                 while ( $this->reader->read() ) {
1011                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
1012                                         $this->reader->localName == 'contributor' ) {
1013                                 break;
1014                         }
1015
1016                         $tag = $this->reader->localName;
1017
1018                         if ( in_array( $tag, $fields ) ) {
1019                                 $info[$tag] = $this->nodeContents();
1020                         }
1021                 }
1022
1023                 return $info;
1024         }
1025
1026         /**
1027          * @param string $text
1028          * @param string|null $ns
1029          * @return array|bool
1030          */
1031         private function processTitle( $text, $ns = null ) {
1032                 if ( is_null( $this->foreignNamespaces ) ) {
1033                         $foreignTitleFactory = new NaiveForeignTitleFactory();
1034                 } else {
1035                         $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
1036                                 $this->foreignNamespaces );
1037                 }
1038
1039                 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
1040                         intval( $ns ) );
1041
1042                 $title = $this->importTitleFactory->createTitleFromForeignTitle(
1043                         $foreignTitle );
1044
1045                 $commandLineMode = $this->config->get( 'CommandLineMode' );
1046                 if ( is_null( $title ) ) {
1047                         # Invalid page title? Ignore the page
1048                         $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
1049                         return false;
1050                 } elseif ( $title->isExternal() ) {
1051                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
1052                         return false;
1053                 } elseif ( !$title->canExist() ) {
1054                         $this->notice( 'import-error-special', $title->getPrefixedText() );
1055                         return false;
1056                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
1057                         # Do not import if the importing wiki user cannot edit this page
1058                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
1059                         return false;
1060                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
1061                         # Do not import if the importing wiki user cannot create this page
1062                         $this->notice( 'import-error-create', $title->getPrefixedText() );
1063                         return false;
1064                 }
1065
1066                 return [ $title, $foreignTitle ];
1067         }
1068 }