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