includes/import/WikiImporter.php

   1 <?php
   2 /**
   3  * MediaWiki page data importer.
   4  *
   5  * Copyright © 2003,2005 Brion Vibber <brion@pobox.com>
   6  * https://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup SpecialPage
  25  */
  26
  27 use MediaWiki\MediaWikiServices;
  28
  29 /**
  30  * XML file reader for the page data importer.
  31  *
  32  * implements Special:Import
  33  * @ingroup SpecialPage
  34  */
  35 class WikiImporter {
  36         private $reader = null;
  37         private $foreignNamespaces = null;
  38         private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
  39         private $mSiteInfoCallback, $mPageOutCallback;
  40         private $mNoticeCallback, $mDebug;
  41         private $mImportUploads, $mImageBasePath;
  42         private $mNoUpdates = false;
  43         private $pageOffset = 0;
  44         /** @var Config */
  45         private $config;
  46         /** @var ImportTitleFactory */
  47         private $importTitleFactory;
  48         /** @var array */
  49         private $countableCache = [];
  50         /** @var bool */
  51         private $disableStatisticsUpdate = false;
  52         /** @var ExternalUserNames */
  53         private $externalUserNames;
  54
  55         /**
  56          * Creates an ImportXMLReader drawing from the source provided
  57          * @param ImportSource $source
  58          * @param Config $config
  59          * @throws Exception
  60          */
  61         function __construct( ImportSource $source, Config $config ) {
  62                 if ( !class_exists( 'XMLReader' ) ) {
  63                         throw new Exception( 'Import requires PHP to have been compiled with libxml support' );
  64                 }
  65
  66                 $this->reader = new XMLReader();
  67                 $this->config = $config;
  68
  69                 if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) {
  70                         stream_wrapper_register( 'uploadsource', UploadSourceAdapter::class );
  71                 }
  72                 $id = UploadSourceAdapter::registerSource( $source );
  73
  74                 // Enable the entity loader, as it is needed for loading external URLs via
  75                 // XMLReader::open (T86036)
  76                 $oldDisable = libxml_disable_entity_loader( false );
  77                 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
  78                         $status = $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  79                 } else {
  80                         $status = $this->reader->open( "uploadsource://$id" );
  81                 }
  82                 if ( !$status ) {
  83                         $error = libxml_get_last_error();
  84                         libxml_disable_entity_loader( $oldDisable );
  85                         throw new MWException( 'Encountered an internal error while initializing WikiImporter object: ' .
  86                                 $error->message );
  87                 }
  88                 libxml_disable_entity_loader( $oldDisable );
  89
  90                 // Default callbacks
  91                 $this->setPageCallback( [ $this, 'beforeImportPage' ] );
  92                 $this->setRevisionCallback( [ $this, "importRevision" ] );
  93                 $this->setUploadCallback( [ $this, 'importUpload' ] );
  94                 $this->setLogItemCallback( [ $this, 'importLogItem' ] );
  95                 $this->setPageOutCallback( [ $this, 'finishImportPage' ] );
  96
  97                 $this->importTitleFactory = new NaiveImportTitleFactory();
  98                 $this->externalUserNames = new ExternalUserNames( 'imported', false );
  99         }
 100
 101         /**
 102          * @return null|XMLReader
 103          */
 104         public function getReader() {
 105                 return $this->reader;
 106         }
 107
 108         public function throwXmlError( $err ) {
 109                 $this->debug( "FAILURE: $err" );
 110                 wfDebug( "WikiImporter XML error: $err\n" );
 111         }
 112
 113         public function debug( $data ) {
 114                 if ( $this->mDebug ) {
 115                         wfDebug( "IMPORT: $data\n" );
 116                 }
 117         }
 118
 119         public function warn( $data ) {
 120                 wfDebug( "IMPORT: $data\n" );
 121         }
 122
 123         public function notice( $msg, ...$params ) {
 124                 if ( is_callable( $this->mNoticeCallback ) ) {
 125                         call_user_func( $this->mNoticeCallback, $msg, $params );
 126                 } else { # No ImportReporter -> CLI
 127                         // T177997: the command line importers should call setNoticeCallback()
 128                         // for their own custom callback to echo the notice
 129                         wfDebug( wfMessage( $msg, $params )->text() . "\n" );
 130                 }
 131         }
 132
 133         /**
 134          * Set debug mode...
 135          * @param bool $debug
 136          */
 137         function setDebug( $debug ) {
 138                 $this->mDebug = $debug;
 139         }
 140
 141         /**
 142          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
 143          * @param bool $noupdates
 144          */
 145         function setNoUpdates( $noupdates ) {
 146                 $this->mNoUpdates = $noupdates;
 147         }
 148
 149         /**
 150          * Sets 'pageOffset' value. So it will skip the first n-1 pages
 151          * and start from the nth page. It's 1-based indexing.
 152          * @param int $nthPage
 153          * @since 1.29
 154          */
 155         function setPageOffset( $nthPage ) {
 156                 $this->pageOffset = $nthPage;
 157         }
 158
 159         /**
 160          * Set a callback that displays notice messages
 161          *
 162          * @param callable $callback
 163          * @return callable
 164          */
 165         public function setNoticeCallback( $callback ) {
 166                 return wfSetVar( $this->mNoticeCallback, $callback );
 167         }
 168
 169         /**
 170          * Sets the action to perform as each new page in the stream is reached.
 171          * @param callable $callback
 172          * @return callable
 173          */
 174         public function setPageCallback( $callback ) {
 175                 $previous = $this->mPageCallback;
 176                 $this->mPageCallback = $callback;
 177                 return $previous;
 178         }
 179
 180         /**
 181          * Sets the action to perform as each page in the stream is completed.
 182          * Callback accepts the page title (as a Title object), a second object
 183          * with the original title form (in case it's been overridden into a
 184          * local namespace), and a count of revisions.
 185          *
 186          * @param callable $callback
 187          * @return callable
 188          */
 189         public function setPageOutCallback( $callback ) {
 190                 $previous = $this->mPageOutCallback;
 191                 $this->mPageOutCallback = $callback;
 192                 return $previous;
 193         }
 194
 195         /**
 196          * Sets the action to perform as each page revision is reached.
 197          * @param callable $callback
 198          * @return callable
 199          */
 200         public function setRevisionCallback( $callback ) {
 201                 $previous = $this->mRevisionCallback;
 202                 $this->mRevisionCallback = $callback;
 203                 return $previous;
 204         }
 205
 206         /**
 207          * Sets the action to perform as each file upload version is reached.
 208          * @param callable $callback
 209          * @return callable
 210          */
 211         public function setUploadCallback( $callback ) {
 212                 $previous = $this->mUploadCallback;
 213                 $this->mUploadCallback = $callback;
 214                 return $previous;
 215         }
 216
 217         /**
 218          * Sets the action to perform as each log item reached.
 219          * @param callable $callback
 220          * @return callable
 221          */
 222         public function setLogItemCallback( $callback ) {
 223                 $previous = $this->mLogItemCallback;
 224                 $this->mLogItemCallback = $callback;
 225                 return $previous;
 226         }
 227
 228         /**
 229          * Sets the action to perform when site info is encountered
 230          * @param callable $callback
 231          * @return callable
 232          */
 233         public function setSiteInfoCallback( $callback ) {
 234                 $previous = $this->mSiteInfoCallback;
 235                 $this->mSiteInfoCallback = $callback;
 236                 return $previous;
 237         }
 238
 239         /**
 240          * Sets the factory object to use to convert ForeignTitle objects into local
 241          * Title objects
 242          * @param ImportTitleFactory $factory
 243          */
 244         public function setImportTitleFactory( $factory ) {
 245                 $this->importTitleFactory = $factory;
 246         }
 247
 248         /**
 249          * Set a target namespace to override the defaults
 250          * @param null|int $namespace
 251          * @return bool
 252          */
 253         public function setTargetNamespace( $namespace ) {
 254                 if ( is_null( $namespace ) ) {
 255                         // Don't override namespaces
 256                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 257                         return true;
 258                 } elseif (
 259                         $namespace >= 0 &&
 260                         MediaWikiServices::getInstance()->getNamespaceInfo()->exists( intval( $namespace ) )
 261                 ) {
 262                         $namespace = intval( $namespace );
 263                         $this->setImportTitleFactory( new NamespaceImportTitleFactory( $namespace ) );
 264                         return true;
 265                 } else {
 266                         return false;
 267                 }
 268         }
 269
 270         /**
 271          * Set a target root page under which all pages are imported
 272          * @param null|string $rootpage
 273          * @return Status
 274          */
 275         public function setTargetRootPage( $rootpage ) {
 276                 $status = Status::newGood();
 277                 if ( is_null( $rootpage ) ) {
 278                         // No rootpage
 279                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 280                 } elseif ( $rootpage !== '' ) {
 281                         $rootpage = rtrim( $rootpage, '/' ); // avoid double slashes
 282                         $title = Title::newFromText( $rootpage );
 283
 284                         if ( !$title || $title->isExternal() ) {
 285                                 $status->fatal( 'import-rootpage-invalid' );
 286                         } elseif (
 287                                 !MediaWikiServices::getInstance()->getNamespaceInfo()->
 288                                 hasSubpages( $title->getNamespace() )
 289                         ) {
 290                                 $displayNSText = $title->getNamespace() == NS_MAIN
 291                                         ? wfMessage( 'blanknamespace' )->text()
 292                                         : MediaWikiServices::getInstance()->getContentLanguage()->
 293                                                 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                 return $status;
 302         }
 303
 304         /**
 305          * @param string $dir
 306          */
 307         public function setImageBasePath( $dir ) {
 308                 $this->mImageBasePath = $dir;
 309         }
 310
 311         /**
 312          * @param bool $import
 313          */
 314         public function setImportUploads( $import ) {
 315                 $this->mImportUploads = $import;
 316         }
 317
 318         /**
 319          * @since 1.31
 320          * @param string $usernamePrefix Prefix to apply to unknown (and possibly also known) usernames
 321          * @param bool $assignKnownUsers Whether to apply the prefix to usernames that exist locally
 322          */
 323         public function setUsernamePrefix( $usernamePrefix, $assignKnownUsers ) {
 324                 $this->externalUserNames = new ExternalUserNames( $usernamePrefix, $assignKnownUsers );
 325         }
 326
 327         /**
 328          * Statistics update can cause a lot of time
 329          * @since 1.29
 330          */
 331         public function disableStatisticsUpdate() {
 332                 $this->disableStatisticsUpdate = true;
 333         }
 334
 335         /**
 336          * Default per-page callback. Sets up some things related to site statistics
 337          * @param array $titleAndForeignTitle Two-element array, with Title object at
 338          * index 0 and ForeignTitle object at index 1
 339          * @return bool
 340          */
 341         public function beforeImportPage( $titleAndForeignTitle ) {
 342                 $title = $titleAndForeignTitle[0];
 343                 $page = WikiPage::factory( $title );
 344                 $this->countableCache['title_' . $title->getPrefixedText()] = $page->isCountable();
 345                 return true;
 346         }
 347
 348         /**
 349          * Default per-revision callback, performs the import.
 350          * @param WikiRevision $revision
 351          * @return bool
 352          */
 353         public function importRevision( $revision ) {
 354                 if ( !$revision->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
 355                         $this->notice( 'import-error-bad-location',
 356                                 $revision->getTitle()->getPrefixedText(),
 357                                 $revision->getID(),
 358                                 $revision->getModel(),
 359                                 $revision->getFormat() );
 360
 361                         return false;
 362                 }
 363
 364                 try {
 365                         return $revision->importOldRevision();
 366                 } catch ( MWContentSerializationException $ex ) {
 367                         $this->notice( 'import-error-unserialize',
 368                                 $revision->getTitle()->getPrefixedText(),
 369                                 $revision->getID(),
 370                                 $revision->getModel(),
 371                                 $revision->getFormat() );
 372                 }
 373
 374                 return false;
 375         }
 376
 377         /**
 378          * Default per-revision callback, performs the import.
 379          * @param WikiRevision $revision
 380          * @return bool
 381          */
 382         public function importLogItem( $revision ) {
 383                 return $revision->importLogItem();
 384         }
 385
 386         /**
 387          * Dummy for now...
 388          * @param WikiRevision $revision
 389          * @return bool
 390          */
 391         public function importUpload( $revision ) {
 392                 return $revision->importUpload();
 393         }
 394
 395         /**
 396          * Mostly for hook use
 397          * @param Title $title
 398          * @param ForeignTitle $foreignTitle
 399          * @param int $revCount
 400          * @param int $sRevCount
 401          * @param array $pageInfo
 402          * @return bool
 403          */
 404         public function finishImportPage( $title, $foreignTitle, $revCount,
 405                 $sRevCount, $pageInfo
 406         ) {
 407                 // Update article count statistics (T42009)
 408                 // The normal counting logic in WikiPage->doEditUpdates() is designed for
 409                 // one-revision-at-a-time editing, not bulk imports. In this situation it
 410                 // suffers from issues of replica DB lag. We let WikiPage handle the total page
 411                 // and revision count, and we implement our own custom logic for the
 412                 // article (content page) count.
 413                 if ( !$this->disableStatisticsUpdate ) {
 414                         $page = WikiPage::factory( $title );
 415                         $page->loadPageData( 'fromdbmaster' );
 416                         $content = $page->getContent();
 417                         if ( $content === null ) {
 418                                 wfDebug( __METHOD__ . ': Skipping article count adjustment for ' . $title .
 419                                         ' because WikiPage::getContent() returned null' );
 420                         } else {
 421                                 $editInfo = $page->prepareContentForEdit( $content );
 422                                 $countKey = 'title_' . $title->getPrefixedText();
 423                                 $countable = $page->isCountable( $editInfo );
 424                                 if ( array_key_exists( $countKey, $this->countableCache ) &&
 425                                         $countable != $this->countableCache[$countKey] ) {
 426                                         DeferredUpdates::addUpdate( SiteStatsUpdate::factory( [
 427                                                 'articles' => ( (int)$countable - (int)$this->countableCache[$countKey] )
 428                                         ] ) );
 429                                 }
 430                         }
 431                 }
 432
 433                 return Hooks::run( 'AfterImportPage', func_get_args() );
 434         }
 435
 436         /**
 437          * Alternate per-revision callback, for debugging.
 438          * @param WikiRevision &$revision
 439          */
 440         public function debugRevisionHandler( &$revision ) {
 441                 $this->debug( "Got revision:" );
 442                 if ( is_object( $revision->title ) ) {
 443                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 444                 } else {
 445                         $this->debug( "-- Title: <invalid>" );
 446                 }
 447                 $this->debug( "-- User: " . $revision->user_text );
 448                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 449                 $this->debug( "-- Comment: " . $revision->comment );
 450                 $this->debug( "-- Text: " . $revision->text );
 451         }
 452
 453         /**
 454          * Notify the callback function of site info
 455          * @param array $siteInfo
 456          * @return bool|mixed
 457          */
 458         private function siteInfoCallback( $siteInfo ) {
 459                 if ( isset( $this->mSiteInfoCallback ) ) {
 460                         return call_user_func_array( $this->mSiteInfoCallback,
 461                                         [ $siteInfo, $this ] );
 462                 } else {
 463                         return false;
 464                 }
 465         }
 466
 467         /**
 468          * Notify the callback function when a new "<page>" is reached.
 469          * @param Title $title
 470          */
 471         function pageCallback( $title ) {
 472                 if ( isset( $this->mPageCallback ) ) {
 473                         call_user_func( $this->mPageCallback, $title );
 474                 }
 475         }
 476
 477         /**
 478          * Notify the callback function when a "</page>" is closed.
 479          * @param Title $title
 480          * @param ForeignTitle $foreignTitle
 481          * @param int $revCount
 482          * @param int $sucCount Number of revisions for which callback returned true
 483          * @param array $pageInfo Associative array of page information
 484          */
 485         private function pageOutCallback( $title, $foreignTitle, $revCount,
 486                         $sucCount, $pageInfo ) {
 487                 if ( isset( $this->mPageOutCallback ) ) {
 488                         call_user_func_array( $this->mPageOutCallback, func_get_args() );
 489                 }
 490         }
 491
 492         /**
 493          * Notify the callback function of a revision
 494          * @param WikiRevision $revision
 495          * @return bool|mixed
 496          */
 497         private function revisionCallback( $revision ) {
 498                 if ( isset( $this->mRevisionCallback ) ) {
 499                         return call_user_func_array( $this->mRevisionCallback,
 500                                         [ $revision, $this ] );
 501                 } else {
 502                         return false;
 503                 }
 504         }
 505
 506         /**
 507          * Notify the callback function of a new log item
 508          * @param WikiRevision $revision
 509          * @return bool|mixed
 510          */
 511         private function logItemCallback( $revision ) {
 512                 if ( isset( $this->mLogItemCallback ) ) {
 513                         return call_user_func_array( $this->mLogItemCallback,
 514                                         [ $revision, $this ] );
 515                 } else {
 516                         return false;
 517                 }
 518         }
 519
 520         /**
 521          * Retrieves the contents of the named attribute of the current element.
 522          * @param string $attr The name of the attribute
 523          * @return string The value of the attribute or an empty string if it is not set in the current
 524          * element.
 525          */
 526         public function nodeAttribute( $attr ) {
 527                 return $this->reader->getAttribute( $attr );
 528         }
 529
 530         /**
 531          * Shouldn't something like this be built-in to XMLReader?
 532          * Fetches text contents of the current element, assuming
 533          * no sub-elements or such scary things.
 534          * @return string
 535          * @private
 536          */
 537         public function nodeContents() {
 538                 if ( $this->reader->isEmptyElement ) {
 539                         return "";
 540                 }
 541                 $buffer = "";
 542                 while ( $this->reader->read() ) {
 543                         switch ( $this->reader->nodeType ) {
 544                                 case XMLReader::TEXT:
 545                                 case XMLReader::CDATA:
 546                                 case XMLReader::SIGNIFICANT_WHITESPACE:
 547                                         $buffer .= $this->reader->value;
 548                                         break;
 549                                 case XMLReader::END_ELEMENT:
 550                                         return $buffer;
 551                         }
 552                 }
 553
 554                 $this->reader->close();
 555                 return '';
 556         }
 557
 558         /**
 559          * Primary entry point
 560          * @throws Exception
 561          * @throws MWException
 562          * @return bool
 563          */
 564         public function doImport() {
 565                 // Calls to reader->read need to be wrapped in calls to
 566                 // libxml_disable_entity_loader() to avoid local file
 567                 // inclusion attacks (T48932).
 568                 $oldDisable = libxml_disable_entity_loader( true );
 569                 $this->reader->read();
 570
 571                 if ( $this->reader->localName != 'mediawiki' ) {
 572                         libxml_disable_entity_loader( $oldDisable );
 573                         throw new MWException( "Expected <mediawiki> tag, got " .
 574                                 $this->reader->localName );
 575                 }
 576                 $this->debug( "<mediawiki> tag is correct." );
 577
 578                 $this->debug( "Starting primary dump processing loop." );
 579
 580                 $keepReading = $this->reader->read();
 581                 $skip = false;
 582                 $rethrow = null;
 583                 $pageCount = 0;
 584                 try {
 585                         while ( $keepReading ) {
 586                                 $tag = $this->reader->localName;
 587                                 if ( $this->pageOffset ) {
 588                                         if ( $tag === 'page' ) {
 589                                                 $pageCount++;
 590                                         }
 591                                         if ( $pageCount < $this->pageOffset ) {
 592                                                 $keepReading = $this->reader->next();
 593                                                 continue;
 594                                         }
 595                                 }
 596                                 $type = $this->reader->nodeType;
 597
 598                                 if ( !Hooks::run( 'ImportHandleToplevelXMLTag', [ $this ] ) ) {
 599                                         // Do nothing
 600                                 } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) {
 601                                         break;
 602                                 } elseif ( $tag == 'siteinfo' ) {
 603                                         $this->handleSiteInfo();
 604                                 } elseif ( $tag == 'page' ) {
 605                                         $this->handlePage();
 606                                 } elseif ( $tag == 'logitem' ) {
 607                                         $this->handleLogItem();
 608                                 } elseif ( $tag != '#text' ) {
 609                                         $this->warn( "Unhandled top-level XML tag $tag" );
 610
 611                                         $skip = true;
 612                                 }
 613
 614                                 if ( $skip ) {
 615                                         $keepReading = $this->reader->next();
 616                                         $skip = false;
 617                                         $this->debug( "Skip" );
 618                                 } else {
 619                                         $keepReading = $this->reader->read();
 620                                 }
 621                         }
 622                 } catch ( Exception $ex ) {
 623                         $rethrow = $ex;
 624                 }
 625
 626                 // finally
 627                 libxml_disable_entity_loader( $oldDisable );
 628                 $this->reader->close();
 629
 630                 if ( $rethrow ) {
 631                         throw $rethrow;
 632                 }
 633
 634                 return true;
 635         }
 636
 637         private function handleSiteInfo() {
 638                 $this->debug( "Enter site info handler." );
 639                 $siteInfo = [];
 640
 641                 // Fields that can just be stuffed in the siteInfo object
 642                 $normalFields = [ 'sitename', 'base', 'generator', 'case' ];
 643
 644                 while ( $this->reader->read() ) {
 645                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 646                                         $this->reader->localName == 'siteinfo' ) {
 647                                 break;
 648                         }
 649
 650                         $tag = $this->reader->localName;
 651
 652                         if ( $tag == 'namespace' ) {
 653                                 $this->foreignNamespaces[$this->nodeAttribute( 'key' )] =
 654                                         $this->nodeContents();
 655                         } elseif ( in_array( $tag, $normalFields ) ) {
 656                                 $siteInfo[$tag] = $this->nodeContents();
 657                         }
 658                 }
 659
 660                 $siteInfo['_namespaces'] = $this->foreignNamespaces;
 661                 $this->siteInfoCallback( $siteInfo );
 662         }
 663
 664         private function handleLogItem() {
 665                 $this->debug( "Enter log item handler." );
 666                 $logInfo = [];
 667
 668                 // Fields that can just be stuffed in the pageInfo object
 669                 $normalFields = [ 'id', 'comment', 'type', 'action', 'timestamp',
 670                                         'logtitle', 'params' ];
 671
 672                 while ( $this->reader->read() ) {
 673                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 674                                         $this->reader->localName == 'logitem' ) {
 675                                 break;
 676                         }
 677
 678                         $tag = $this->reader->localName;
 679
 680                         if ( !Hooks::run( 'ImportHandleLogItemXMLTag', [
 681                                 $this, $logInfo
 682                         ] ) ) {
 683                                 // Do nothing
 684                         } elseif ( in_array( $tag, $normalFields ) ) {
 685                                 $logInfo[$tag] = $this->nodeContents();
 686                         } elseif ( $tag == 'contributor' ) {
 687                                 $logInfo['contributor'] = $this->handleContributor();
 688                         } elseif ( $tag != '#text' ) {
 689                                 $this->warn( "Unhandled log-item XML tag $tag" );
 690                         }
 691                 }
 692
 693                 $this->processLogItem( $logInfo );
 694         }
 695
 696         /**
 697          * @param array $logInfo
 698          * @return bool|mixed
 699          */
 700         private function processLogItem( $logInfo ) {
 701                 $revision = new WikiRevision( $this->config );
 702
 703                 if ( isset( $logInfo['id'] ) ) {
 704                         $revision->setID( $logInfo['id'] );
 705                 }
 706                 $revision->setType( $logInfo['type'] );
 707                 $revision->setAction( $logInfo['action'] );
 708                 if ( isset( $logInfo['timestamp'] ) ) {
 709                         $revision->setTimestamp( $logInfo['timestamp'] );
 710                 }
 711                 if ( isset( $logInfo['params'] ) ) {
 712                         $revision->setParams( $logInfo['params'] );
 713                 }
 714                 if ( isset( $logInfo['logtitle'] ) ) {
 715                         // @todo Using Title for non-local titles is a recipe for disaster.
 716                         // We should use ForeignTitle here instead.
 717                         $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 718                 }
 719
 720                 $revision->setNoUpdates( $this->mNoUpdates );
 721
 722                 if ( isset( $logInfo['comment'] ) ) {
 723                         $revision->setComment( $logInfo['comment'] );
 724                 }
 725
 726                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 727                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 728                 }
 729
 730                 if ( !isset( $logInfo['contributor']['username'] ) ) {
 731                         $revision->setUsername( $this->externalUserNames->addPrefix( 'Unknown user' ) );
 732                 } else {
 733                         $revision->setUsername(
 734                                 $this->externalUserNames->applyPrefix( $logInfo['contributor']['username'] )
 735                         );
 736                 }
 737
 738                 return $this->logItemCallback( $revision );
 739         }
 740
 741         private function handlePage() {
 742                 // Handle page data.
 743                 $this->debug( "Enter page handler." );
 744                 $pageInfo = [ 'revisionCount' => 0, 'successfulRevisionCount' => 0 ];
 745
 746                 // Fields that can just be stuffed in the pageInfo object
 747                 $normalFields = [ 'title', 'ns', 'id', 'redirect', 'restrictions' ];
 748
 749                 $skip = false;
 750                 $badTitle = false;
 751
 752                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 753                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 754                                         $this->reader->localName == 'page' ) {
 755                                 break;
 756                         }
 757
 758                         $skip = false;
 759
 760                         $tag = $this->reader->localName;
 761
 762                         if ( $badTitle ) {
 763                                 // The title is invalid, bail out of this page
 764                                 $skip = true;
 765                         } elseif ( !Hooks::run( 'ImportHandlePageXMLTag', [ $this,
 766                                                 &$pageInfo ] ) ) {
 767                                 // Do nothing
 768                         } elseif ( in_array( $tag, $normalFields ) ) {
 769                                 // An XML snippet:
 770                                 // <page>
 771                                 //     <id>123</id>
 772                                 //     <title>Page</title>
 773                                 //     <redirect title="NewTitle"/>
 774                                 //     ...
 775                                 // Because the redirect tag is built differently, we need special handling for that case.
 776                                 if ( $tag == 'redirect' ) {
 777                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 778                                 } else {
 779                                         $pageInfo[$tag] = $this->nodeContents();
 780                                 }
 781                         } elseif ( $tag == 'revision' || $tag == 'upload' ) {
 782                                 if ( !isset( $title ) ) {
 783                                         $title = $this->processTitle( $pageInfo['title'],
 784                                                 $pageInfo['ns'] ?? null );
 785
 786                                         // $title is either an array of two titles or false.
 787                                         if ( is_array( $title ) ) {
 788                                                 $this->pageCallback( $title );
 789                                                 list( $pageInfo['_title'], $foreignTitle ) = $title;
 790                                         } else {
 791                                                 $badTitle = true;
 792                                                 $skip = true;
 793                                         }
 794                                 }
 795
 796                                 if ( $title ) {
 797                                         if ( $tag == 'revision' ) {
 798                                                 $this->handleRevision( $pageInfo );
 799                                         } else {
 800                                                 $this->handleUpload( $pageInfo );
 801                                         }
 802                                 }
 803                         } elseif ( $tag != '#text' ) {
 804                                 $this->warn( "Unhandled page XML tag $tag" );
 805                                 $skip = true;
 806                         }
 807                 }
 808
 809                 // @note $pageInfo is only set if a valid $title is processed above with
 810                 //       no error. If we have a valid $title, then pageCallback is called
 811                 //       above, $pageInfo['title'] is set and we do pageOutCallback here.
 812                 //       If $pageInfo['_title'] is not set, then $foreignTitle is also not
 813                 //       set since they both come from $title above.
 814                 if ( array_key_exists( '_title', $pageInfo ) ) {
 815                         $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
 816                                         $pageInfo['revisionCount'],
 817                                         $pageInfo['successfulRevisionCount'],
 818                                         $pageInfo );
 819                 }
 820         }
 821
 822         /**
 823          * @param array $pageInfo
 824          */
 825         private function handleRevision( &$pageInfo ) {
 826                 $this->debug( "Enter revision handler" );
 827                 $revisionInfo = [];
 828
 829                 $normalFields = [ 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text', 'sha1' ];
 830
 831                 $skip = false;
 832
 833                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 834                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 835                                         $this->reader->localName == 'revision' ) {
 836                                 break;
 837                         }
 838
 839                         $tag = $this->reader->localName;
 840
 841                         if ( !Hooks::run( 'ImportHandleRevisionXMLTag', [
 842                                 $this, $pageInfo, $revisionInfo
 843                         ] ) ) {
 844                                 // Do nothing
 845                         } elseif ( in_array( $tag, $normalFields ) ) {
 846                                 $revisionInfo[$tag] = $this->nodeContents();
 847                         } elseif ( $tag == 'contributor' ) {
 848                                 $revisionInfo['contributor'] = $this->handleContributor();
 849                         } elseif ( $tag != '#text' ) {
 850                                 $this->warn( "Unhandled revision XML tag $tag" );
 851                                 $skip = true;
 852                         }
 853                 }
 854
 855                 $pageInfo['revisionCount']++;
 856                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 857                         $pageInfo['successfulRevisionCount']++;
 858                 }
 859         }
 860
 861         /**
 862          * @param array $pageInfo
 863          * @param array $revisionInfo
 864          * @throws MWException
 865          * @return bool|mixed
 866          */
 867         private function processRevision( $pageInfo, $revisionInfo ) {
 868                 global $wgMaxArticleSize;
 869
 870                 // Make sure revisions won't violate $wgMaxArticleSize, which could lead to
 871                 // database errors and instability. Testing for revisions with only listed
 872                 // content models, as other content models might use serialization formats
 873                 // which aren't checked against $wgMaxArticleSize.
 874                 if ( ( !isset( $revisionInfo['model'] ) ||
 875                         in_array( $revisionInfo['model'], [
 876                                 'wikitext',
 877                                 'css',
 878                                 'json',
 879                                 'javascript',
 880                                 'text',
 881                                 ''
 882                         ] ) ) &&
 883                         strlen( $revisionInfo['text'] ) > $wgMaxArticleSize * 1024
 884                 ) {
 885                         throw new MWException( 'The text of ' .
 886                                 ( isset( $revisionInfo['id'] ) ?
 887                                         "the revision with ID $revisionInfo[id]" :
 888                                         'a revision'
 889                                 ) . " exceeds the maximum allowable size ($wgMaxArticleSize KB)" );
 890                 }
 891
 892                 // FIXME: process schema version 11!
 893                 $revision = new WikiRevision( $this->config );
 894
 895                 if ( isset( $revisionInfo['id'] ) ) {
 896                         $revision->setID( $revisionInfo['id'] );
 897                 }
 898                 if ( isset( $revisionInfo['model'] ) ) {
 899                         $revision->setModel( $revisionInfo['model'] );
 900                 }
 901                 if ( isset( $revisionInfo['format'] ) ) {
 902                         $revision->setFormat( $revisionInfo['format'] );
 903                 }
 904                 $revision->setTitle( $pageInfo['_title'] );
 905
 906                 if ( isset( $revisionInfo['text'] ) ) {
 907                         $handler = $revision->getContentHandler();
 908                         $text = $handler->importTransform(
 909                                 $revisionInfo['text'],
 910                                 $revision->getFormat() );
 911
 912                         $revision->setText( $text );
 913                 }
 914                 $revision->setTimestamp( $revisionInfo['timestamp'] ?? wfTimestampNow() );
 915
 916                 if ( isset( $revisionInfo['comment'] ) ) {
 917                         $revision->setComment( $revisionInfo['comment'] );
 918                 }
 919
 920                 if ( isset( $revisionInfo['minor'] ) ) {
 921                         $revision->setMinor( true );
 922                 }
 923                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 924                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 925                 } elseif ( isset( $revisionInfo['contributor']['username'] ) ) {
 926                         $revision->setUsername(
 927                                 $this->externalUserNames->applyPrefix( $revisionInfo['contributor']['username'] )
 928                         );
 929                 } else {
 930                         $revision->setUsername( $this->externalUserNames->addPrefix( 'Unknown user' ) );
 931                 }
 932                 if ( isset( $revisionInfo['sha1'] ) ) {
 933                         $revision->setSha1Base36( $revisionInfo['sha1'] );
 934                 }
 935                 $revision->setNoUpdates( $this->mNoUpdates );
 936
 937                 return $this->revisionCallback( $revision );
 938         }
 939
 940         /**
 941          * @param array $pageInfo
 942          * @return mixed
 943          */
 944         private function handleUpload( &$pageInfo ) {
 945                 $this->debug( "Enter upload handler" );
 946                 $uploadInfo = [];
 947
 948                 $normalFields = [ 'timestamp', 'comment', 'filename', 'text',
 949                                         'src', 'size', 'sha1base36', 'archivename', 'rel' ];
 950
 951                 $skip = false;
 952
 953                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 954                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 955                                         $this->reader->localName == 'upload' ) {
 956                                 break;
 957                         }
 958
 959                         $tag = $this->reader->localName;
 960
 961                         if ( !Hooks::run( 'ImportHandleUploadXMLTag', [
 962                                 $this, $pageInfo
 963                         ] ) ) {
 964                                 // Do nothing
 965                         } elseif ( in_array( $tag, $normalFields ) ) {
 966                                 $uploadInfo[$tag] = $this->nodeContents();
 967                         } elseif ( $tag == 'contributor' ) {
 968                                 $uploadInfo['contributor'] = $this->handleContributor();
 969                         } elseif ( $tag == 'contents' ) {
 970                                 $contents = $this->nodeContents();
 971                                 $encoding = $this->reader->getAttribute( 'encoding' );
 972                                 if ( $encoding === 'base64' ) {
 973                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 974                                         $uploadInfo['isTempSrc'] = true;
 975                                 }
 976                         } elseif ( $tag != '#text' ) {
 977                                 $this->warn( "Unhandled upload XML tag $tag" );
 978                                 $skip = true;
 979                         }
 980                 }
 981
 982                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 983                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 984                         if ( file_exists( $path ) ) {
 985                                 $uploadInfo['fileSrc'] = $path;
 986                                 $uploadInfo['isTempSrc'] = false;
 987                         }
 988                 }
 989
 990                 if ( $this->mImportUploads ) {
 991                         return $this->processUpload( $pageInfo, $uploadInfo );
 992                 }
 993         }
 994
 995         /**
 996          * @param string $contents
 997          * @return string
 998          */
 999         private function dumpTemp( $contents ) {
1000                 $filename = tempnam( wfTempDir(), 'importupload' );
1001                 file_put_contents( $filename, $contents );
1002                 return $filename;
1003         }
1004
1005         /**
1006          * @param array $pageInfo
1007          * @param array $uploadInfo
1008          * @return mixed
1009          */
1010         private function processUpload( $pageInfo, $uploadInfo ) {
1011                 $revision = new WikiRevision( $this->config );
1012                 $text = $uploadInfo['text'] ?? '';
1013
1014                 $revision->setTitle( $pageInfo['_title'] );
1015                 $revision->setID( $pageInfo['id'] );
1016                 $revision->setTimestamp( $uploadInfo['timestamp'] );
1017                 $revision->setText( $text );
1018                 $revision->setFilename( $uploadInfo['filename'] );
1019                 if ( isset( $uploadInfo['archivename'] ) ) {
1020                         $revision->setArchiveName( $uploadInfo['archivename'] );
1021                 }
1022                 $revision->setSrc( $uploadInfo['src'] );
1023                 if ( isset( $uploadInfo['fileSrc'] ) ) {
1024                         $revision->setFileSrc( $uploadInfo['fileSrc'],
1025                                 !empty( $uploadInfo['isTempSrc'] ) );
1026                 }
1027                 if ( isset( $uploadInfo['sha1base36'] ) ) {
1028                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
1029                 }
1030                 $revision->setSize( intval( $uploadInfo['size'] ) );
1031                 $revision->setComment( $uploadInfo['comment'] );
1032
1033                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
1034                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
1035                 }
1036                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
1037                         $revision->setUsername(
1038                                 $this->externalUserNames->applyPrefix( $uploadInfo['contributor']['username'] )
1039                         );
1040                 }
1041                 $revision->setNoUpdates( $this->mNoUpdates );
1042
1043                 return call_user_func( $this->mUploadCallback, $revision );
1044         }
1045
1046         /**
1047          * @return array
1048          */
1049         private function handleContributor() {
1050                 $fields = [ 'id', 'ip', 'username' ];
1051                 $info = [];
1052
1053                 if ( $this->reader->isEmptyElement ) {
1054                         return $info;
1055                 }
1056                 while ( $this->reader->read() ) {
1057                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
1058                                         $this->reader->localName == 'contributor' ) {
1059                                 break;
1060                         }
1061
1062                         $tag = $this->reader->localName;
1063
1064                         if ( in_array( $tag, $fields ) ) {
1065                                 $info[$tag] = $this->nodeContents();
1066                         }
1067                 }
1068
1069                 return $info;
1070         }
1071
1072         /**
1073          * @param string $text
1074          * @param string|null $ns
1075          * @return array|bool
1076          */
1077         private function processTitle( $text, $ns = null ) {
1078                 if ( is_null( $this->foreignNamespaces ) ) {
1079                         $foreignTitleFactory = new NaiveForeignTitleFactory();
1080                 } else {
1081                         $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
1082                                 $this->foreignNamespaces );
1083                 }
1084
1085                 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
1086                         intval( $ns ) );
1087
1088                 $title = $this->importTitleFactory->createTitleFromForeignTitle(
1089                         $foreignTitle );
1090
1091                 $commandLineMode = $this->config->get( 'CommandLineMode' );
1092                 if ( is_null( $title ) ) {
1093                         # Invalid page title? Ignore the page
1094                         $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
1095                         return false;
1096                 } elseif ( $title->isExternal() ) {
1097                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
1098                         return false;
1099                 } elseif ( !$title->canExist() ) {
1100                         $this->notice( 'import-error-special', $title->getPrefixedText() );
1101                         return false;
1102                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
1103                         # Do not import if the importing wiki user cannot edit this page
1104                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
1105                         return false;
1106                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
1107                         # Do not import if the importing wiki user cannot create this page
1108                         $this->notice( 'import-error-create', $title->getPrefixedText() );
1109                         return false;
1110                 }
1111
1112                 return [ $title, $foreignTitle ];
1113         }
1114 }