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 /**
  28  * XML file reader for the page data importer.
  29  *
  30  * implements Special:Import
  31  * @ingroup SpecialPage
  32  */
  33 class WikiImporter {
  34         private $reader = null;
  35         private $foreignNamespaces = null;
  36         private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
  37         private $mSiteInfoCallback, $mPageOutCallback;
  38         private $mNoticeCallback, $mDebug;
  39         private $mImportUploads, $mImageBasePath;
  40         private $mNoUpdates = false;
  41         /** @var Config */
  42         private $config;
  43         /** @var ImportTitleFactory */
  44         private $importTitleFactory;
  45         /** @var array */
  46         private $countableCache = [];
  47
  48         /**
  49          * Creates an ImportXMLReader drawing from the source provided
  50          * @param ImportSource $source
  51          * @param Config $config
  52          * @throws Exception
  53          */
  54         function __construct( ImportSource $source, Config $config = null ) {
  55                 if ( !class_exists( 'XMLReader' ) ) {
  56                         throw new Exception( 'Import requires PHP to have been compiled with libxml support' );
  57                 }
  58
  59                 $this->reader = new XMLReader();
  60                 if ( !$config ) {
  61                         wfDeprecated( __METHOD__ . ' without a Config instance', '1.25' );
  62                         $config = ConfigFactory::getDefaultInstance()->makeConfig( 'main' );
  63                 }
  64                 $this->config = $config;
  65
  66                 if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) {
  67                         stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  68                 }
  69                 $id = UploadSourceAdapter::registerSource( $source );
  70
  71                 // Enable the entity loader, as it is needed for loading external URLs via
  72                 // XMLReader::open (T86036)
  73                 $oldDisable = libxml_disable_entity_loader( false );
  74                 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
  75                         $status = $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  76                 } else {
  77                         $status = $this->reader->open( "uploadsource://$id" );
  78                 }
  79                 if ( !$status ) {
  80                         $error = libxml_get_last_error();
  81                         libxml_disable_entity_loader( $oldDisable );
  82                         throw new MWException( 'Encountered an internal error while initializing WikiImporter object: ' .
  83                                 $error->message );
  84                 }
  85                 libxml_disable_entity_loader( $oldDisable );
  86
  87                 // Default callbacks
  88                 $this->setPageCallback( [ $this, 'beforeImportPage' ] );
  89                 $this->setRevisionCallback( [ $this, "importRevision" ] );
  90                 $this->setUploadCallback( [ $this, 'importUpload' ] );
  91                 $this->setLogItemCallback( [ $this, 'importLogItem' ] );
  92                 $this->setPageOutCallback( [ $this, 'finishImportPage' ] );
  93
  94                 $this->importTitleFactory = new NaiveImportTitleFactory();
  95         }
  96
  97         /**
  98          * @return null|XMLReader
  99          */
 100         public function getReader() {
 101                 return $this->reader;
 102         }
 103
 104         public function throwXmlError( $err ) {
 105                 $this->debug( "FAILURE: $err" );
 106                 wfDebug( "WikiImporter XML error: $err\n" );
 107         }
 108
 109         public function debug( $data ) {
 110                 if ( $this->mDebug ) {
 111                         wfDebug( "IMPORT: $data\n" );
 112                 }
 113         }
 114
 115         public function warn( $data ) {
 116                 wfDebug( "IMPORT: $data\n" );
 117         }
 118
 119         public function notice( $msg /*, $param, ...*/ ) {
 120                 $params = func_get_args();
 121                 array_shift( $params );
 122
 123                 if ( is_callable( $this->mNoticeCallback ) ) {
 124                         call_user_func( $this->mNoticeCallback, $msg, $params );
 125                 } else { # No ImportReporter -> CLI
 126                         echo wfMessage( $msg, $params )->text() . "\n";
 127                 }
 128         }
 129
 130         /**
 131          * Set debug mode...
 132          * @param bool $debug
 133          */
 134         function setDebug( $debug ) {
 135                 $this->mDebug = $debug;
 136         }
 137
 138         /**
 139          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
 140          * @param bool $noupdates
 141          */
 142         function setNoUpdates( $noupdates ) {
 143                 $this->mNoUpdates = $noupdates;
 144         }
 145
 146         /**
 147          * Set a callback that displays notice messages
 148          *
 149          * @param callable $callback
 150          * @return callable
 151          */
 152         public function setNoticeCallback( $callback ) {
 153                 return wfSetVar( $this->mNoticeCallback, $callback );
 154         }
 155
 156         /**
 157          * Sets the action to perform as each new page in the stream is reached.
 158          * @param callable $callback
 159          * @return callable
 160          */
 161         public function setPageCallback( $callback ) {
 162                 $previous = $this->mPageCallback;
 163                 $this->mPageCallback = $callback;
 164                 return $previous;
 165         }
 166
 167         /**
 168          * Sets the action to perform as each page in the stream is completed.
 169          * Callback accepts the page title (as a Title object), a second object
 170          * with the original title form (in case it's been overridden into a
 171          * local namespace), and a count of revisions.
 172          *
 173          * @param callable $callback
 174          * @return callable
 175          */
 176         public function setPageOutCallback( $callback ) {
 177                 $previous = $this->mPageOutCallback;
 178                 $this->mPageOutCallback = $callback;
 179                 return $previous;
 180         }
 181
 182         /**
 183          * Sets the action to perform as each page revision is reached.
 184          * @param callable $callback
 185          * @return callable
 186          */
 187         public function setRevisionCallback( $callback ) {
 188                 $previous = $this->mRevisionCallback;
 189                 $this->mRevisionCallback = $callback;
 190                 return $previous;
 191         }
 192
 193         /**
 194          * Sets the action to perform as each file upload version is reached.
 195          * @param callable $callback
 196          * @return callable
 197          */
 198         public function setUploadCallback( $callback ) {
 199                 $previous = $this->mUploadCallback;
 200                 $this->mUploadCallback = $callback;
 201                 return $previous;
 202         }
 203
 204         /**
 205          * Sets the action to perform as each log item reached.
 206          * @param callable $callback
 207          * @return callable
 208          */
 209         public function setLogItemCallback( $callback ) {
 210                 $previous = $this->mLogItemCallback;
 211                 $this->mLogItemCallback = $callback;
 212                 return $previous;
 213         }
 214
 215         /**
 216          * Sets the action to perform when site info is encountered
 217          * @param callable $callback
 218          * @return callable
 219          */
 220         public function setSiteInfoCallback( $callback ) {
 221                 $previous = $this->mSiteInfoCallback;
 222                 $this->mSiteInfoCallback = $callback;
 223                 return $previous;
 224         }
 225
 226         /**
 227          * Sets the factory object to use to convert ForeignTitle objects into local
 228          * Title objects
 229          * @param ImportTitleFactory $factory
 230          */
 231         public function setImportTitleFactory( $factory ) {
 232                 $this->importTitleFactory = $factory;
 233         }
 234
 235         /**
 236          * Set a target namespace to override the defaults
 237          * @param null|int $namespace
 238          * @return bool
 239          */
 240         public function setTargetNamespace( $namespace ) {
 241                 if ( is_null( $namespace ) ) {
 242                         // Don't override namespaces
 243                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 244                         return true;
 245                 } elseif (
 246                         $namespace >= 0 &&
 247                         MWNamespace::exists( intval( $namespace ) )
 248                 ) {
 249                         $namespace = intval( $namespace );
 250                         $this->setImportTitleFactory( new NamespaceImportTitleFactory( $namespace ) );
 251                         return true;
 252                 } else {
 253                         return false;
 254                 }
 255         }
 256
 257         /**
 258          * Set a target root page under which all pages are imported
 259          * @param null|string $rootpage
 260          * @return Status
 261          */
 262         public function setTargetRootPage( $rootpage ) {
 263                 $status = Status::newGood();
 264                 if ( is_null( $rootpage ) ) {
 265                         // No rootpage
 266                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 267                 } elseif ( $rootpage !== '' ) {
 268                         $rootpage = rtrim( $rootpage, '/' ); // avoid double slashes
 269                         $title = Title::newFromText( $rootpage );
 270
 271                         if ( !$title || $title->isExternal() ) {
 272                                 $status->fatal( 'import-rootpage-invalid' );
 273                         } else {
 274                                 if ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 275                                         global $wgContLang;
 276
 277                                         $displayNSText = $title->getNamespace() == NS_MAIN
 278                                                 ? wfMessage( 'blanknamespace' )->text()
 279                                                 : $wgContLang->getNsText( $title->getNamespace() );
 280                                         $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 281                                 } else {
 282                                         // set namespace to 'all', so the namespace check in processTitle() can pass
 283                                         $this->setTargetNamespace( null );
 284                                         $this->setImportTitleFactory( new SubpageImportTitleFactory( $title ) );
 285                                 }
 286                         }
 287                 }
 288                 return $status;
 289         }
 290
 291         /**
 292          * @param string $dir
 293          */
 294         public function setImageBasePath( $dir ) {
 295                 $this->mImageBasePath = $dir;
 296         }
 297
 298         /**
 299          * @param bool $import
 300          */
 301         public function setImportUploads( $import ) {
 302                 $this->mImportUploads = $import;
 303         }
 304
 305         /**
 306          * Default per-page callback. Sets up some things related to site statistics
 307          * @param array $titleAndForeignTitle Two-element array, with Title object at
 308          * index 0 and ForeignTitle object at index 1
 309          * @return bool
 310          */
 311         public function beforeImportPage( $titleAndForeignTitle ) {
 312                 $title = $titleAndForeignTitle[0];
 313                 $page = WikiPage::factory( $title );
 314                 $this->countableCache['title_' . $title->getPrefixedText()] = $page->isCountable();
 315                 return true;
 316         }
 317
 318         /**
 319          * Default per-revision callback, performs the import.
 320          * @param WikiRevision $revision
 321          * @return bool
 322          */
 323         public function importRevision( $revision ) {
 324                 if ( !$revision->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
 325                         $this->notice( 'import-error-bad-location',
 326                                 $revision->getTitle()->getPrefixedText(),
 327                                 $revision->getID(),
 328                                 $revision->getModel(),
 329                                 $revision->getFormat() );
 330
 331                         return false;
 332                 }
 333
 334                 try {
 335                         $dbw = wfGetDB( DB_MASTER );
 336                         return $dbw->deadlockLoop( [ $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                 $dbw = wfGetDB( DB_MASTER );
 355                 return $dbw->deadlockLoop( [ $revision, 'importLogItem' ] );
 356         }
 357
 358         /**
 359          * Dummy for now...
 360          * @param WikiRevision $revision
 361          * @return bool
 362          */
 363         public function importUpload( $revision ) {
 364                 $dbw = wfGetDB( DB_MASTER );
 365                 return $dbw->deadlockLoop( [ $revision, 'importUpload' ] );
 366         }
 367
 368         /**
 369          * Mostly for hook use
 370          * @param Title $title
 371          * @param ForeignTitle $foreignTitle
 372          * @param int $revCount
 373          * @param int $sRevCount
 374          * @param array $pageInfo
 375          * @return bool
 376          */
 377         public function finishImportPage( $title, $foreignTitle, $revCount,
 378                         $sRevCount, $pageInfo ) {
 379
 380                 // Update article count statistics (T42009)
 381                 // The normal counting logic in WikiPage->doEditUpdates() is designed for
 382                 // one-revision-at-a-time editing, not bulk imports. In this situation it
 383                 // suffers from issues of slave lag. We let WikiPage handle the total page
 384                 // and revision count, and we implement our own custom logic for the
 385                 // article (content page) count.
 386                 $page = WikiPage::factory( $title );
 387                 $page->loadPageData( 'fromdbmaster' );
 388                 $content = $page->getContent();
 389                 if ( $content === null ) {
 390                         wfDebug( __METHOD__ . ': Skipping article count adjustment for ' . $title .
 391                                 ' because WikiPage::getContent() returned null' );
 392                 } else {
 393                         $editInfo = $page->prepareContentForEdit( $content );
 394                         $countKey = 'title_' . $title->getPrefixedText();
 395                         $countable = $page->isCountable( $editInfo );
 396                         if ( array_key_exists( $countKey, $this->countableCache ) &&
 397                                 $countable != $this->countableCache[$countKey] ) {
 398                                 DeferredUpdates::addUpdate( SiteStatsUpdate::factory( [
 399                                         'articles' => ( (int)$countable - (int)$this->countableCache[$countKey] )
 400                                 ] ) );
 401                         }
 402                 }
 403
 404                 $args = func_get_args();
 405                 return Hooks::run( 'AfterImportPage', $args );
 406         }
 407
 408         /**
 409          * Alternate per-revision callback, for debugging.
 410          * @param WikiRevision $revision
 411          */
 412         public function debugRevisionHandler( &$revision ) {
 413                 $this->debug( "Got revision:" );
 414                 if ( is_object( $revision->title ) ) {
 415                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 416                 } else {
 417                         $this->debug( "-- Title: <invalid>" );
 418                 }
 419                 $this->debug( "-- User: " . $revision->user_text );
 420                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 421                 $this->debug( "-- Comment: " . $revision->comment );
 422                 $this->debug( "-- Text: " . $revision->text );
 423         }
 424
 425         /**
 426          * Notify the callback function of site info
 427          * @param array $siteInfo
 428          * @return bool|mixed
 429          */
 430         private function siteInfoCallback( $siteInfo ) {
 431                 if ( isset( $this->mSiteInfoCallback ) ) {
 432                         return call_user_func_array( $this->mSiteInfoCallback,
 433                                         [ $siteInfo, $this ] );
 434                 } else {
 435                         return false;
 436                 }
 437         }
 438
 439         /**
 440          * Notify the callback function when a new "<page>" is reached.
 441          * @param Title $title
 442          */
 443         function pageCallback( $title ) {
 444                 if ( isset( $this->mPageCallback ) ) {
 445                         call_user_func( $this->mPageCallback, $title );
 446                 }
 447         }
 448
 449         /**
 450          * Notify the callback function when a "</page>" is closed.
 451          * @param Title $title
 452          * @param ForeignTitle $foreignTitle
 453          * @param int $revCount
 454          * @param int $sucCount Number of revisions for which callback returned true
 455          * @param array $pageInfo Associative array of page information
 456          */
 457         private function pageOutCallback( $title, $foreignTitle, $revCount,
 458                         $sucCount, $pageInfo ) {
 459                 if ( isset( $this->mPageOutCallback ) ) {
 460                         $args = func_get_args();
 461                         call_user_func_array( $this->mPageOutCallback, $args );
 462                 }
 463         }
 464
 465         /**
 466          * Notify the callback function of a revision
 467          * @param WikiRevision $revision
 468          * @return bool|mixed
 469          */
 470         private function revisionCallback( $revision ) {
 471                 if ( isset( $this->mRevisionCallback ) ) {
 472                         return call_user_func_array( $this->mRevisionCallback,
 473                                         [ $revision, $this ] );
 474                 } else {
 475                         return false;
 476                 }
 477         }
 478
 479         /**
 480          * Notify the callback function of a new log item
 481          * @param WikiRevision $revision
 482          * @return bool|mixed
 483          */
 484         private function logItemCallback( $revision ) {
 485                 if ( isset( $this->mLogItemCallback ) ) {
 486                         return call_user_func_array( $this->mLogItemCallback,
 487                                         [ $revision, $this ] );
 488                 } else {
 489                         return false;
 490                 }
 491         }
 492
 493         /**
 494          * Retrieves the contents of the named attribute of the current element.
 495          * @param string $attr The name of the attribute
 496          * @return string The value of the attribute or an empty string if it is not set in the current
 497          * element.
 498          */
 499         public function nodeAttribute( $attr ) {
 500                 return $this->reader->getAttribute( $attr );
 501         }
 502
 503         /**
 504          * Shouldn't something like this be built-in to XMLReader?
 505          * Fetches text contents of the current element, assuming
 506          * no sub-elements or such scary things.
 507          * @return string
 508          * @access private
 509          */
 510         public function nodeContents() {
 511                 if ( $this->reader->isEmptyElement ) {
 512                         return "";
 513                 }
 514                 $buffer = "";
 515                 while ( $this->reader->read() ) {
 516                         switch ( $this->reader->nodeType ) {
 517                         case XMLReader::TEXT:
 518                         case XMLReader::CDATA:
 519                         case XMLReader::SIGNIFICANT_WHITESPACE:
 520                                 $buffer .= $this->reader->value;
 521                                 break;
 522                         case XMLReader::END_ELEMENT:
 523                                 return $buffer;
 524                         }
 525                 }
 526
 527                 $this->reader->close();
 528                 return '';
 529         }
 530
 531         /**
 532          * Primary entry point
 533          * @throws MWException
 534          * @return bool
 535          */
 536         public function doImport() {
 537                 // Calls to reader->read need to be wrapped in calls to
 538                 // libxml_disable_entity_loader() to avoid local file
 539                 // inclusion attacks (bug 46932).
 540                 $oldDisable = libxml_disable_entity_loader( true );
 541                 $this->reader->read();
 542
 543                 if ( $this->reader->localName != 'mediawiki' ) {
 544                         libxml_disable_entity_loader( $oldDisable );
 545                         throw new MWException( "Expected <mediawiki> tag, got " .
 546                                 $this->reader->localName );
 547                 }
 548                 $this->debug( "<mediawiki> tag is correct." );
 549
 550                 $this->debug( "Starting primary dump processing loop." );
 551
 552                 $keepReading = $this->reader->read();
 553                 $skip = false;
 554                 $rethrow = null;
 555                 try {
 556                         while ( $keepReading ) {
 557                                 $tag = $this->reader->localName;
 558                                 $type = $this->reader->nodeType;
 559
 560                                 if ( !Hooks::run( 'ImportHandleToplevelXMLTag', [ $this ] ) ) {
 561                                         // Do nothing
 562                                 } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) {
 563                                         break;
 564                                 } elseif ( $tag == 'siteinfo' ) {
 565                                         $this->handleSiteInfo();
 566                                 } elseif ( $tag == 'page' ) {
 567                                         $this->handlePage();
 568                                 } elseif ( $tag == 'logitem' ) {
 569                                         $this->handleLogItem();
 570                                 } elseif ( $tag != '#text' ) {
 571                                         $this->warn( "Unhandled top-level XML tag $tag" );
 572
 573                                         $skip = true;
 574                                 }
 575
 576                                 if ( $skip ) {
 577                                         $keepReading = $this->reader->next();
 578                                         $skip = false;
 579                                         $this->debug( "Skip" );
 580                                 } else {
 581                                         $keepReading = $this->reader->read();
 582                                 }
 583                         }
 584                 } catch ( Exception $ex ) {
 585                         $rethrow = $ex;
 586                 }
 587
 588                 // finally
 589                 libxml_disable_entity_loader( $oldDisable );
 590                 $this->reader->close();
 591
 592                 if ( $rethrow ) {
 593                         throw $rethrow;
 594                 }
 595
 596                 return true;
 597         }
 598
 599         private function handleSiteInfo() {
 600                 $this->debug( "Enter site info handler." );
 601                 $siteInfo = [];
 602
 603                 // Fields that can just be stuffed in the siteInfo object
 604                 $normalFields = [ 'sitename', 'base', 'generator', 'case' ];
 605
 606                 while ( $this->reader->read() ) {
 607                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 608                                         $this->reader->localName == 'siteinfo' ) {
 609                                 break;
 610                         }
 611
 612                         $tag = $this->reader->localName;
 613
 614                         if ( $tag == 'namespace' ) {
 615                                 $this->foreignNamespaces[$this->nodeAttribute( 'key' )] =
 616                                         $this->nodeContents();
 617                         } elseif ( in_array( $tag, $normalFields ) ) {
 618                                 $siteInfo[$tag] = $this->nodeContents();
 619                         }
 620                 }
 621
 622                 $siteInfo['_namespaces'] = $this->foreignNamespaces;
 623                 $this->siteInfoCallback( $siteInfo );
 624         }
 625
 626         private function handleLogItem() {
 627                 $this->debug( "Enter log item handler." );
 628                 $logInfo = [];
 629
 630                 // Fields that can just be stuffed in the pageInfo object
 631                 $normalFields = [ 'id', 'comment', 'type', 'action', 'timestamp',
 632                                         'logtitle', 'params' ];
 633
 634                 while ( $this->reader->read() ) {
 635                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 636                                         $this->reader->localName == 'logitem' ) {
 637                                 break;
 638                         }
 639
 640                         $tag = $this->reader->localName;
 641
 642                         if ( !Hooks::run( 'ImportHandleLogItemXMLTag', [
 643                                 $this, $logInfo
 644                         ] ) ) {
 645                                 // Do nothing
 646                         } elseif ( in_array( $tag, $normalFields ) ) {
 647                                 $logInfo[$tag] = $this->nodeContents();
 648                         } elseif ( $tag == 'contributor' ) {
 649                                 $logInfo['contributor'] = $this->handleContributor();
 650                         } elseif ( $tag != '#text' ) {
 651                                 $this->warn( "Unhandled log-item XML tag $tag" );
 652                         }
 653                 }
 654
 655                 $this->processLogItem( $logInfo );
 656         }
 657
 658         /**
 659          * @param array $logInfo
 660          * @return bool|mixed
 661          */
 662         private function processLogItem( $logInfo ) {
 663
 664                 $revision = new WikiRevision( $this->config );
 665
 666                 if ( isset( $logInfo['id'] ) ) {
 667                         $revision->setID( $logInfo['id'] );
 668                 }
 669                 $revision->setType( $logInfo['type'] );
 670                 $revision->setAction( $logInfo['action'] );
 671                 if ( isset( $logInfo['timestamp'] ) ) {
 672                         $revision->setTimestamp( $logInfo['timestamp'] );
 673                 }
 674                 if ( isset( $logInfo['params'] ) ) {
 675                         $revision->setParams( $logInfo['params'] );
 676                 }
 677                 if ( isset( $logInfo['logtitle'] ) ) {
 678                         // @todo Using Title for non-local titles is a recipe for disaster.
 679                         // We should use ForeignTitle here instead.
 680                         $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 681                 }
 682
 683                 $revision->setNoUpdates( $this->mNoUpdates );
 684
 685                 if ( isset( $logInfo['comment'] ) ) {
 686                         $revision->setComment( $logInfo['comment'] );
 687                 }
 688
 689                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 690                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 691                 }
 692
 693                 if ( !isset( $logInfo['contributor']['username'] ) ) {
 694                         $revision->setUsername( 'Unknown user' );
 695                 } else {
 696                         $revision->setUsername( $logInfo['contributor']['username'] );
 697                 }
 698
 699                 return $this->logItemCallback( $revision );
 700         }
 701
 702         private function handlePage() {
 703                 // Handle page data.
 704                 $this->debug( "Enter page handler." );
 705                 $pageInfo = [ 'revisionCount' => 0, 'successfulRevisionCount' => 0 ];
 706
 707                 // Fields that can just be stuffed in the pageInfo object
 708                 $normalFields = [ 'title', 'ns', 'id', 'redirect', 'restrictions' ];
 709
 710                 $skip = false;
 711                 $badTitle = false;
 712
 713                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 714                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 715                                         $this->reader->localName == 'page' ) {
 716                                 break;
 717                         }
 718
 719                         $skip = false;
 720
 721                         $tag = $this->reader->localName;
 722
 723                         if ( $badTitle ) {
 724                                 // The title is invalid, bail out of this page
 725                                 $skip = true;
 726                         } elseif ( !Hooks::run( 'ImportHandlePageXMLTag', [ $this,
 727                                                 &$pageInfo ] ) ) {
 728                                 // Do nothing
 729                         } elseif ( in_array( $tag, $normalFields ) ) {
 730                                 // An XML snippet:
 731                                 // <page>
 732                                 //     <id>123</id>
 733                                 //     <title>Page</title>
 734                                 //     <redirect title="NewTitle"/>
 735                                 //     ...
 736                                 // Because the redirect tag is built differently, we need special handling for that case.
 737                                 if ( $tag == 'redirect' ) {
 738                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 739                                 } else {
 740                                         $pageInfo[$tag] = $this->nodeContents();
 741                                 }
 742                         } elseif ( $tag == 'revision' || $tag == 'upload' ) {
 743                                 if ( !isset( $title ) ) {
 744                                         $title = $this->processTitle( $pageInfo['title'],
 745                                                 isset( $pageInfo['ns'] ) ? $pageInfo['ns'] : null );
 746
 747                                         // $title is either an array of two titles or false.
 748                                         if ( is_array( $title ) ) {
 749                                                 $this->pageCallback( $title );
 750                                                 list( $pageInfo['_title'], $foreignTitle ) = $title;
 751                                         } else {
 752                                                 $badTitle = true;
 753                                                 $skip = true;
 754                                         }
 755                                 }
 756
 757                                 if ( $title ) {
 758                                         if ( $tag == 'revision' ) {
 759                                                 $this->handleRevision( $pageInfo );
 760                                         } else {
 761                                                 $this->handleUpload( $pageInfo );
 762                                         }
 763                                 }
 764                         } elseif ( $tag != '#text' ) {
 765                                 $this->warn( "Unhandled page XML tag $tag" );
 766                                 $skip = true;
 767                         }
 768                 }
 769
 770                 // @note $pageInfo is only set if a valid $title is processed above with
 771                 //       no error. If we have a valid $title, then pageCallback is called
 772                 //       above, $pageInfo['title'] is set and we do pageOutCallback here.
 773                 //       If $pageInfo['_title'] is not set, then $foreignTitle is also not
 774                 //       set since they both come from $title above.
 775                 if ( array_key_exists( '_title', $pageInfo ) ) {
 776                         $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
 777                                         $pageInfo['revisionCount'],
 778                                         $pageInfo['successfulRevisionCount'],
 779                                         $pageInfo );
 780                 }
 781         }
 782
 783         /**
 784          * @param array $pageInfo
 785          */
 786         private function handleRevision( &$pageInfo ) {
 787                 $this->debug( "Enter revision handler" );
 788                 $revisionInfo = [];
 789
 790                 $normalFields = [ 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' ];
 791
 792                 $skip = false;
 793
 794                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 795                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 796                                         $this->reader->localName == 'revision' ) {
 797                                 break;
 798                         }
 799
 800                         $tag = $this->reader->localName;
 801
 802                         if ( !Hooks::run( 'ImportHandleRevisionXMLTag', [
 803                                 $this, $pageInfo, $revisionInfo
 804                         ] ) ) {
 805                                 // Do nothing
 806                         } elseif ( in_array( $tag, $normalFields ) ) {
 807                                 $revisionInfo[$tag] = $this->nodeContents();
 808                         } elseif ( $tag == 'contributor' ) {
 809                                 $revisionInfo['contributor'] = $this->handleContributor();
 810                         } elseif ( $tag != '#text' ) {
 811                                 $this->warn( "Unhandled revision XML tag $tag" );
 812                                 $skip = true;
 813                         }
 814                 }
 815
 816                 $pageInfo['revisionCount']++;
 817                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 818                         $pageInfo['successfulRevisionCount']++;
 819                 }
 820         }
 821
 822         /**
 823          * @param array $pageInfo
 824          * @param array $revisionInfo
 825          * @return bool|mixed
 826          */
 827         private function processRevision( $pageInfo, $revisionInfo ) {
 828                 global $wgMaxArticleSize;
 829
 830                 // Make sure revisions won't violate $wgMaxArticleSize, which could lead to
 831                 // database errors and instability. Testing for revisions with only listed
 832                 // content models, as other content models might use serialization formats
 833                 // which aren't checked against $wgMaxArticleSize.
 834                 if ( ( !isset( $revisionInfo['model'] ) ||
 835                         in_array( $revisionInfo['model'], [
 836                                 'wikitext',
 837                                 'css',
 838                                 'json',
 839                                 'javascript',
 840                                 'text',
 841                                 ''
 842                         ] ) ) &&
 843                         (int)( strlen( $revisionInfo['text'] ) / 1024 ) > $wgMaxArticleSize
 844                 ) {
 845                         throw new MWException( 'The text of ' .
 846                                 ( isset( $revisionInfo['id'] ) ?
 847                                         "the revision with ID $revisionInfo[id]" :
 848                                         'a revision'
 849                                 ) . " exceeds the maximum allowable size ($wgMaxArticleSize KB)" );
 850                 }
 851
 852                 $revision = new WikiRevision( $this->config );
 853
 854                 if ( isset( $revisionInfo['id'] ) ) {
 855                         $revision->setID( $revisionInfo['id'] );
 856                 }
 857                 if ( isset( $revisionInfo['model'] ) ) {
 858                         $revision->setModel( $revisionInfo['model'] );
 859                 }
 860                 if ( isset( $revisionInfo['format'] ) ) {
 861                         $revision->setFormat( $revisionInfo['format'] );
 862                 }
 863                 $revision->setTitle( $pageInfo['_title'] );
 864
 865                 if ( isset( $revisionInfo['text'] ) ) {
 866                         $handler = $revision->getContentHandler();
 867                         $text = $handler->importTransform(
 868                                 $revisionInfo['text'],
 869                                 $revision->getFormat() );
 870
 871                         $revision->setText( $text );
 872                 }
 873                 if ( isset( $revisionInfo['timestamp'] ) ) {
 874                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 875                 } else {
 876                         $revision->setTimestamp( wfTimestampNow() );
 877                 }
 878
 879                 if ( isset( $revisionInfo['comment'] ) ) {
 880                         $revision->setComment( $revisionInfo['comment'] );
 881                 }
 882
 883                 if ( isset( $revisionInfo['minor'] ) ) {
 884                         $revision->setMinor( true );
 885                 }
 886                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 887                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 888                 } elseif ( isset( $revisionInfo['contributor']['username'] ) ) {
 889                         $revision->setUsername( $revisionInfo['contributor']['username'] );
 890                 } else {
 891                         $revision->setUsername( 'Unknown user' );
 892                 }
 893                 $revision->setNoUpdates( $this->mNoUpdates );
 894
 895                 return $this->revisionCallback( $revision );
 896         }
 897
 898         /**
 899          * @param array $pageInfo
 900          * @return mixed
 901          */
 902         private function handleUpload( &$pageInfo ) {
 903                 $this->debug( "Enter upload handler" );
 904                 $uploadInfo = [];
 905
 906                 $normalFields = [ 'timestamp', 'comment', 'filename', 'text',
 907                                         'src', 'size', 'sha1base36', 'archivename', 'rel' ];
 908
 909                 $skip = false;
 910
 911                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 912                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 913                                         $this->reader->localName == 'upload' ) {
 914                                 break;
 915                         }
 916
 917                         $tag = $this->reader->localName;
 918
 919                         if ( !Hooks::run( 'ImportHandleUploadXMLTag', [
 920                                 $this, $pageInfo
 921                         ] ) ) {
 922                                 // Do nothing
 923                         } elseif ( in_array( $tag, $normalFields ) ) {
 924                                 $uploadInfo[$tag] = $this->nodeContents();
 925                         } elseif ( $tag == 'contributor' ) {
 926                                 $uploadInfo['contributor'] = $this->handleContributor();
 927                         } elseif ( $tag == 'contents' ) {
 928                                 $contents = $this->nodeContents();
 929                                 $encoding = $this->reader->getAttribute( 'encoding' );
 930                                 if ( $encoding === 'base64' ) {
 931                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 932                                         $uploadInfo['isTempSrc'] = true;
 933                                 }
 934                         } elseif ( $tag != '#text' ) {
 935                                 $this->warn( "Unhandled upload XML tag $tag" );
 936                                 $skip = true;
 937                         }
 938                 }
 939
 940                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 941                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 942                         if ( file_exists( $path ) ) {
 943                                 $uploadInfo['fileSrc'] = $path;
 944                                 $uploadInfo['isTempSrc'] = false;
 945                         }
 946                 }
 947
 948                 if ( $this->mImportUploads ) {
 949                         return $this->processUpload( $pageInfo, $uploadInfo );
 950                 }
 951         }
 952
 953         /**
 954          * @param string $contents
 955          * @return string
 956          */
 957         private function dumpTemp( $contents ) {
 958                 $filename = tempnam( wfTempDir(), 'importupload' );
 959                 file_put_contents( $filename, $contents );
 960                 return $filename;
 961         }
 962
 963         /**
 964          * @param array $pageInfo
 965          * @param array $uploadInfo
 966          * @return mixed
 967          */
 968         private function processUpload( $pageInfo, $uploadInfo ) {
 969                 $revision = new WikiRevision( $this->config );
 970                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 971
 972                 $revision->setTitle( $pageInfo['_title'] );
 973                 $revision->setID( $pageInfo['id'] );
 974                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 975                 $revision->setText( $text );
 976                 $revision->setFilename( $uploadInfo['filename'] );
 977                 if ( isset( $uploadInfo['archivename'] ) ) {
 978                         $revision->setArchiveName( $uploadInfo['archivename'] );
 979                 }
 980                 $revision->setSrc( $uploadInfo['src'] );
 981                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 982                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 983                                 !empty( $uploadInfo['isTempSrc'] ) );
 984                 }
 985                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 986                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 987                 }
 988                 $revision->setSize( intval( $uploadInfo['size'] ) );
 989                 $revision->setComment( $uploadInfo['comment'] );
 990
 991                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 992                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 993                 }
 994                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 995                         $revision->setUsername( $uploadInfo['contributor']['username'] );
 996                 }
 997                 $revision->setNoUpdates( $this->mNoUpdates );
 998
 999                 return call_user_func( $this->mUploadCallback, $revision );
1000         }
1001
1002         /**
1003          * @return array
1004          */
1005         private function handleContributor() {
1006                 $fields = [ 'id', 'ip', 'username' ];
1007                 $info = [];
1008
1009                 if ( $this->reader->isEmptyElement ) {
1010                         return $info;
1011                 }
1012                 while ( $this->reader->read() ) {
1013                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
1014                                         $this->reader->localName == 'contributor' ) {
1015                                 break;
1016                         }
1017
1018                         $tag = $this->reader->localName;
1019
1020                         if ( in_array( $tag, $fields ) ) {
1021                                 $info[$tag] = $this->nodeContents();
1022                         }
1023                 }
1024
1025                 return $info;
1026         }
1027
1028         /**
1029          * @param string $text
1030          * @param string|null $ns
1031          * @return array|bool
1032          */
1033         private function processTitle( $text, $ns = null ) {
1034                 if ( is_null( $this->foreignNamespaces ) ) {
1035                         $foreignTitleFactory = new NaiveForeignTitleFactory();
1036                 } else {
1037                         $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
1038                                 $this->foreignNamespaces );
1039                 }
1040
1041                 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
1042                         intval( $ns ) );
1043
1044                 $title = $this->importTitleFactory->createTitleFromForeignTitle(
1045                         $foreignTitle );
1046
1047                 $commandLineMode = $this->config->get( 'CommandLineMode' );
1048                 if ( is_null( $title ) ) {
1049                         # Invalid page title? Ignore the page
1050                         $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
1051                         return false;
1052                 } elseif ( $title->isExternal() ) {
1053                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
1054                         return false;
1055                 } elseif ( !$title->canExist() ) {
1056                         $this->notice( 'import-error-special', $title->getPrefixedText() );
1057                         return false;
1058                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
1059                         # Do not import if the importing wiki user cannot edit this page
1060                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
1061                         return false;
1062                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
1063                         # Do not import if the importing wiki user cannot create this page
1064                         $this->notice( 'import-error-create', $title->getPrefixedText() );
1065                         return false;
1066                 }
1067
1068                 return [ $title, $foreignTitle ];
1069         }
1070 }