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 /*, $param, ...*/ ) {
 124                 $params = func_get_args();
 125                 array_shift( $params );
 126
 127                 if ( is_callable( $this->mNoticeCallback ) ) {
 128                         call_user_func( $this->mNoticeCallback, $msg, $params );
 129                 } else { # No ImportReporter -> CLI
 130                         // T177997: the command line importers should call setNoticeCallback()
 131                         // for their own custom callback to echo the notice
 132                         wfDebug( wfMessage( $msg, $params )->text() . "\n" );
 133                 }
 134         }
 135
 136         /**
 137          * Set debug mode...
 138          * @param bool $debug
 139          */
 140         function setDebug( $debug ) {
 141                 $this->mDebug = $debug;
 142         }
 143
 144         /**
 145          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
 146          * @param bool $noupdates
 147          */
 148         function setNoUpdates( $noupdates ) {
 149                 $this->mNoUpdates = $noupdates;
 150         }
 151
 152         /**
 153          * Sets 'pageOffset' value. So it will skip the first n-1 pages
 154          * and start from the nth page. It's 1-based indexing.
 155          * @param int $nthPage
 156          * @since 1.29
 157          */
 158         function setPageOffset( $nthPage ) {
 159                 $this->pageOffset = $nthPage;
 160         }
 161
 162         /**
 163          * Set a callback that displays notice messages
 164          *
 165          * @param callable $callback
 166          * @return callable
 167          */
 168         public function setNoticeCallback( $callback ) {
 169                 return wfSetVar( $this->mNoticeCallback, $callback );
 170         }
 171
 172         /**
 173          * Sets the action to perform as each new page in the stream is reached.
 174          * @param callable $callback
 175          * @return callable
 176          */
 177         public function setPageCallback( $callback ) {
 178                 $previous = $this->mPageCallback;
 179                 $this->mPageCallback = $callback;
 180                 return $previous;
 181         }
 182
 183         /**
 184          * Sets the action to perform as each page in the stream is completed.
 185          * Callback accepts the page title (as a Title object), a second object
 186          * with the original title form (in case it's been overridden into a
 187          * local namespace), and a count of revisions.
 188          *
 189          * @param callable $callback
 190          * @return callable
 191          */
 192         public function setPageOutCallback( $callback ) {
 193                 $previous = $this->mPageOutCallback;
 194                 $this->mPageOutCallback = $callback;
 195                 return $previous;
 196         }
 197
 198         /**
 199          * Sets the action to perform as each page revision is reached.
 200          * @param callable $callback
 201          * @return callable
 202          */
 203         public function setRevisionCallback( $callback ) {
 204                 $previous = $this->mRevisionCallback;
 205                 $this->mRevisionCallback = $callback;
 206                 return $previous;
 207         }
 208
 209         /**
 210          * Sets the action to perform as each file upload version is reached.
 211          * @param callable $callback
 212          * @return callable
 213          */
 214         public function setUploadCallback( $callback ) {
 215                 $previous = $this->mUploadCallback;
 216                 $this->mUploadCallback = $callback;
 217                 return $previous;
 218         }
 219
 220         /**
 221          * Sets the action to perform as each log item reached.
 222          * @param callable $callback
 223          * @return callable
 224          */
 225         public function setLogItemCallback( $callback ) {
 226                 $previous = $this->mLogItemCallback;
 227                 $this->mLogItemCallback = $callback;
 228                 return $previous;
 229         }
 230
 231         /**
 232          * Sets the action to perform when site info is encountered
 233          * @param callable $callback
 234          * @return callable
 235          */
 236         public function setSiteInfoCallback( $callback ) {
 237                 $previous = $this->mSiteInfoCallback;
 238                 $this->mSiteInfoCallback = $callback;
 239                 return $previous;
 240         }
 241
 242         /**
 243          * Sets the factory object to use to convert ForeignTitle objects into local
 244          * Title objects
 245          * @param ImportTitleFactory $factory
 246          */
 247         public function setImportTitleFactory( $factory ) {
 248                 $this->importTitleFactory = $factory;
 249         }
 250
 251         /**
 252          * Set a target namespace to override the defaults
 253          * @param null|int $namespace
 254          * @return bool
 255          */
 256         public function setTargetNamespace( $namespace ) {
 257                 if ( is_null( $namespace ) ) {
 258                         // Don't override namespaces
 259                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 260                         return true;
 261                 } elseif (
 262                         $namespace >= 0 &&
 263                         MWNamespace::exists( intval( $namespace ) )
 264                 ) {
 265                         $namespace = intval( $namespace );
 266                         $this->setImportTitleFactory( new NamespaceImportTitleFactory( $namespace ) );
 267                         return true;
 268                 } else {
 269                         return false;
 270                 }
 271         }
 272
 273         /**
 274          * Set a target root page under which all pages are imported
 275          * @param null|string $rootpage
 276          * @return Status
 277          */
 278         public function setTargetRootPage( $rootpage ) {
 279                 $status = Status::newGood();
 280                 if ( is_null( $rootpage ) ) {
 281                         // No rootpage
 282                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 283                 } elseif ( $rootpage !== '' ) {
 284                         $rootpage = rtrim( $rootpage, '/' ); // avoid double slashes
 285                         $title = Title::newFromText( $rootpage );
 286
 287                         if ( !$title || $title->isExternal() ) {
 288                                 $status->fatal( 'import-rootpage-invalid' );
 289                         } elseif ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 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                 $args = func_get_args();
 434                 return Hooks::run( 'AfterImportPage', $args );
 435         }
 436
 437         /**
 438          * Alternate per-revision callback, for debugging.
 439          * @param WikiRevision &$revision
 440          */
 441         public function debugRevisionHandler( &$revision ) {
 442                 $this->debug( "Got revision:" );
 443                 if ( is_object( $revision->title ) ) {
 444                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 445                 } else {
 446                         $this->debug( "-- Title: <invalid>" );
 447                 }
 448                 $this->debug( "-- User: " . $revision->user_text );
 449                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 450                 $this->debug( "-- Comment: " . $revision->comment );
 451                 $this->debug( "-- Text: " . $revision->text );
 452         }
 453
 454         /**
 455          * Notify the callback function of site info
 456          * @param array $siteInfo
 457          * @return bool|mixed
 458          */
 459         private function siteInfoCallback( $siteInfo ) {
 460                 if ( isset( $this->mSiteInfoCallback ) ) {
 461                         return call_user_func_array( $this->mSiteInfoCallback,
 462                                         [ $siteInfo, $this ] );
 463                 } else {
 464                         return false;
 465                 }
 466         }
 467
 468         /**
 469          * Notify the callback function when a new "<page>" is reached.
 470          * @param Title $title
 471          */
 472         function pageCallback( $title ) {
 473                 if ( isset( $this->mPageCallback ) ) {
 474                         call_user_func( $this->mPageCallback, $title );
 475                 }
 476         }
 477
 478         /**
 479          * Notify the callback function when a "</page>" is closed.
 480          * @param Title $title
 481          * @param ForeignTitle $foreignTitle
 482          * @param int $revCount
 483          * @param int $sucCount Number of revisions for which callback returned true
 484          * @param array $pageInfo Associative array of page information
 485          */
 486         private function pageOutCallback( $title, $foreignTitle, $revCount,
 487                         $sucCount, $pageInfo ) {
 488                 if ( isset( $this->mPageOutCallback ) ) {
 489                         $args = func_get_args();
 490                         call_user_func_array( $this->mPageOutCallback, $args );
 491                 }
 492         }
 493
 494         /**
 495          * Notify the callback function of a revision
 496          * @param WikiRevision $revision
 497          * @return bool|mixed
 498          */
 499         private function revisionCallback( $revision ) {
 500                 if ( isset( $this->mRevisionCallback ) ) {
 501                         return call_user_func_array( $this->mRevisionCallback,
 502                                         [ $revision, $this ] );
 503                 } else {
 504                         return false;
 505                 }
 506         }
 507
 508         /**
 509          * Notify the callback function of a new log item
 510          * @param WikiRevision $revision
 511          * @return bool|mixed
 512          */
 513         private function logItemCallback( $revision ) {
 514                 if ( isset( $this->mLogItemCallback ) ) {
 515                         return call_user_func_array( $this->mLogItemCallback,
 516                                         [ $revision, $this ] );
 517                 } else {
 518                         return false;
 519                 }
 520         }
 521
 522         /**
 523          * Retrieves the contents of the named attribute of the current element.
 524          * @param string $attr The name of the attribute
 525          * @return string The value of the attribute or an empty string if it is not set in the current
 526          * element.
 527          */
 528         public function nodeAttribute( $attr ) {
 529                 return $this->reader->getAttribute( $attr );
 530         }
 531
 532         /**
 533          * Shouldn't something like this be built-in to XMLReader?
 534          * Fetches text contents of the current element, assuming
 535          * no sub-elements or such scary things.
 536          * @return string
 537          * @private
 538          */
 539         public function nodeContents() {
 540                 if ( $this->reader->isEmptyElement ) {
 541                         return "";
 542                 }
 543                 $buffer = "";
 544                 while ( $this->reader->read() ) {
 545                         switch ( $this->reader->nodeType ) {
 546                                 case XMLReader::TEXT:
 547                                 case XMLReader::CDATA:
 548                                 case XMLReader::SIGNIFICANT_WHITESPACE:
 549                                         $buffer .= $this->reader->value;
 550                                         break;
 551                                 case XMLReader::END_ELEMENT:
 552                                         return $buffer;
 553                         }
 554                 }
 555
 556                 $this->reader->close();
 557                 return '';
 558         }
 559
 560         /**
 561          * Primary entry point
 562          * @throws Exception
 563          * @throws MWException
 564          * @return bool
 565          */
 566         public function doImport() {
 567                 // Calls to reader->read need to be wrapped in calls to
 568                 // libxml_disable_entity_loader() to avoid local file
 569                 // inclusion attacks (T48932).
 570                 $oldDisable = libxml_disable_entity_loader( true );
 571                 $this->reader->read();
 572
 573                 if ( $this->reader->localName != 'mediawiki' ) {
 574                         libxml_disable_entity_loader( $oldDisable );
 575                         throw new MWException( "Expected <mediawiki> tag, got " .
 576                                 $this->reader->localName );
 577                 }
 578                 $this->debug( "<mediawiki> tag is correct." );
 579
 580                 $this->debug( "Starting primary dump processing loop." );
 581
 582                 $keepReading = $this->reader->read();
 583                 $skip = false;
 584                 $rethrow = null;
 585                 $pageCount = 0;
 586                 try {
 587                         while ( $keepReading ) {
 588                                 $tag = $this->reader->localName;
 589                                 if ( $this->pageOffset ) {
 590                                         if ( $tag === 'page' ) {
 591                                                 $pageCount++;
 592                                         }
 593                                         if ( $pageCount < $this->pageOffset ) {
 594                                                 $keepReading = $this->reader->next();
 595                                                 continue;
 596                                         }
 597                                 }
 598                                 $type = $this->reader->nodeType;
 599
 600                                 if ( !Hooks::run( 'ImportHandleToplevelXMLTag', [ $this ] ) ) {
 601                                         // Do nothing
 602                                 } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) {
 603                                         break;
 604                                 } elseif ( $tag == 'siteinfo' ) {
 605                                         $this->handleSiteInfo();
 606                                 } elseif ( $tag == 'page' ) {
 607                                         $this->handlePage();
 608                                 } elseif ( $tag == 'logitem' ) {
 609                                         $this->handleLogItem();
 610                                 } elseif ( $tag != '#text' ) {
 611                                         $this->warn( "Unhandled top-level XML tag $tag" );
 612
 613                                         $skip = true;
 614                                 }
 615
 616                                 if ( $skip ) {
 617                                         $keepReading = $this->reader->next();
 618                                         $skip = false;
 619                                         $this->debug( "Skip" );
 620                                 } else {
 621                                         $keepReading = $this->reader->read();
 622                                 }
 623                         }
 624                 } catch ( Exception $ex ) {
 625                         $rethrow = $ex;
 626                 }
 627
 628                 // finally
 629                 libxml_disable_entity_loader( $oldDisable );
 630                 $this->reader->close();
 631
 632                 if ( $rethrow ) {
 633                         throw $rethrow;
 634                 }
 635
 636                 return true;
 637         }
 638
 639         private function handleSiteInfo() {
 640                 $this->debug( "Enter site info handler." );
 641                 $siteInfo = [];
 642
 643                 // Fields that can just be stuffed in the siteInfo object
 644                 $normalFields = [ 'sitename', 'base', 'generator', 'case' ];
 645
 646                 while ( $this->reader->read() ) {
 647                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 648                                         $this->reader->localName == 'siteinfo' ) {
 649                                 break;
 650                         }
 651
 652                         $tag = $this->reader->localName;
 653
 654                         if ( $tag == 'namespace' ) {
 655                                 $this->foreignNamespaces[$this->nodeAttribute( 'key' )] =
 656                                         $this->nodeContents();
 657                         } elseif ( in_array( $tag, $normalFields ) ) {
 658                                 $siteInfo[$tag] = $this->nodeContents();
 659                         }
 660                 }
 661
 662                 $siteInfo['_namespaces'] = $this->foreignNamespaces;
 663                 $this->siteInfoCallback( $siteInfo );
 664         }
 665
 666         private function handleLogItem() {
 667                 $this->debug( "Enter log item handler." );
 668                 $logInfo = [];
 669
 670                 // Fields that can just be stuffed in the pageInfo object
 671                 $normalFields = [ 'id', 'comment', 'type', 'action', 'timestamp',
 672                                         'logtitle', 'params' ];
 673
 674                 while ( $this->reader->read() ) {
 675                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 676                                         $this->reader->localName == 'logitem' ) {
 677                                 break;
 678                         }
 679
 680                         $tag = $this->reader->localName;
 681
 682                         if ( !Hooks::run( 'ImportHandleLogItemXMLTag', [
 683                                 $this, $logInfo
 684                         ] ) ) {
 685                                 // Do nothing
 686                         } elseif ( in_array( $tag, $normalFields ) ) {
 687                                 $logInfo[$tag] = $this->nodeContents();
 688                         } elseif ( $tag == 'contributor' ) {
 689                                 $logInfo['contributor'] = $this->handleContributor();
 690                         } elseif ( $tag != '#text' ) {
 691                                 $this->warn( "Unhandled log-item XML tag $tag" );
 692                         }
 693                 }
 694
 695                 $this->processLogItem( $logInfo );
 696         }
 697
 698         /**
 699          * @param array $logInfo
 700          * @return bool|mixed
 701          */
 702         private function processLogItem( $logInfo ) {
 703                 $revision = new WikiRevision( $this->config );
 704
 705                 if ( isset( $logInfo['id'] ) ) {
 706                         $revision->setID( $logInfo['id'] );
 707                 }
 708                 $revision->setType( $logInfo['type'] );
 709                 $revision->setAction( $logInfo['action'] );
 710                 if ( isset( $logInfo['timestamp'] ) ) {
 711                         $revision->setTimestamp( $logInfo['timestamp'] );
 712                 }
 713                 if ( isset( $logInfo['params'] ) ) {
 714                         $revision->setParams( $logInfo['params'] );
 715                 }
 716                 if ( isset( $logInfo['logtitle'] ) ) {
 717                         // @todo Using Title for non-local titles is a recipe for disaster.
 718                         // We should use ForeignTitle here instead.
 719                         $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 720                 }
 721
 722                 $revision->setNoUpdates( $this->mNoUpdates );
 723
 724                 if ( isset( $logInfo['comment'] ) ) {
 725                         $revision->setComment( $logInfo['comment'] );
 726                 }
 727
 728                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 729                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 730                 }
 731
 732                 if ( !isset( $logInfo['contributor']['username'] ) ) {
 733                         $revision->setUsername( $this->externalUserNames->addPrefix( 'Unknown user' ) );
 734                 } else {
 735                         $revision->setUsername(
 736                                 $this->externalUserNames->applyPrefix( $logInfo['contributor']['username'] )
 737                         );
 738                 }
 739
 740                 return $this->logItemCallback( $revision );
 741         }
 742
 743         private function handlePage() {
 744                 // Handle page data.
 745                 $this->debug( "Enter page handler." );
 746                 $pageInfo = [ 'revisionCount' => 0, 'successfulRevisionCount' => 0 ];
 747
 748                 // Fields that can just be stuffed in the pageInfo object
 749                 $normalFields = [ 'title', 'ns', 'id', 'redirect', 'restrictions' ];
 750
 751                 $skip = false;
 752                 $badTitle = false;
 753
 754                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 755                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 756                                         $this->reader->localName == 'page' ) {
 757                                 break;
 758                         }
 759
 760                         $skip = false;
 761
 762                         $tag = $this->reader->localName;
 763
 764                         if ( $badTitle ) {
 765                                 // The title is invalid, bail out of this page
 766                                 $skip = true;
 767                         } elseif ( !Hooks::run( 'ImportHandlePageXMLTag', [ $this,
 768                                                 &$pageInfo ] ) ) {
 769                                 // Do nothing
 770                         } elseif ( in_array( $tag, $normalFields ) ) {
 771                                 // An XML snippet:
 772                                 // <page>
 773                                 //     <id>123</id>
 774                                 //     <title>Page</title>
 775                                 //     <redirect title="NewTitle"/>
 776                                 //     ...
 777                                 // Because the redirect tag is built differently, we need special handling for that case.
 778                                 if ( $tag == 'redirect' ) {
 779                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 780                                 } else {
 781                                         $pageInfo[$tag] = $this->nodeContents();
 782                                 }
 783                         } elseif ( $tag == 'revision' || $tag == 'upload' ) {
 784                                 if ( !isset( $title ) ) {
 785                                         $title = $this->processTitle( $pageInfo['title'],
 786                                                 $pageInfo['ns'] ?? null );
 787
 788                                         // $title is either an array of two titles or false.
 789                                         if ( is_array( $title ) ) {
 790                                                 $this->pageCallback( $title );
 791                                                 list( $pageInfo['_title'], $foreignTitle ) = $title;
 792                                         } else {
 793                                                 $badTitle = true;
 794                                                 $skip = true;
 795                                         }
 796                                 }
 797
 798                                 if ( $title ) {
 799                                         if ( $tag == 'revision' ) {
 800                                                 $this->handleRevision( $pageInfo );
 801                                         } else {
 802                                                 $this->handleUpload( $pageInfo );
 803                                         }
 804                                 }
 805                         } elseif ( $tag != '#text' ) {
 806                                 $this->warn( "Unhandled page XML tag $tag" );
 807                                 $skip = true;
 808                         }
 809                 }
 810
 811                 // @note $pageInfo is only set if a valid $title is processed above with
 812                 //       no error. If we have a valid $title, then pageCallback is called
 813                 //       above, $pageInfo['title'] is set and we do pageOutCallback here.
 814                 //       If $pageInfo['_title'] is not set, then $foreignTitle is also not
 815                 //       set since they both come from $title above.
 816                 if ( array_key_exists( '_title', $pageInfo ) ) {
 817                         $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
 818                                         $pageInfo['revisionCount'],
 819                                         $pageInfo['successfulRevisionCount'],
 820                                         $pageInfo );
 821                 }
 822         }
 823
 824         /**
 825          * @param array $pageInfo
 826          */
 827         private function handleRevision( &$pageInfo ) {
 828                 $this->debug( "Enter revision handler" );
 829                 $revisionInfo = [];
 830
 831                 $normalFields = [ 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text', 'sha1' ];
 832
 833                 $skip = false;
 834
 835                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 836                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 837                                         $this->reader->localName == 'revision' ) {
 838                                 break;
 839                         }
 840
 841                         $tag = $this->reader->localName;
 842
 843                         if ( !Hooks::run( 'ImportHandleRevisionXMLTag', [
 844                                 $this, $pageInfo, $revisionInfo
 845                         ] ) ) {
 846                                 // Do nothing
 847                         } elseif ( in_array( $tag, $normalFields ) ) {
 848                                 $revisionInfo[$tag] = $this->nodeContents();
 849                         } elseif ( $tag == 'contributor' ) {
 850                                 $revisionInfo['contributor'] = $this->handleContributor();
 851                         } elseif ( $tag != '#text' ) {
 852                                 $this->warn( "Unhandled revision XML tag $tag" );
 853                                 $skip = true;
 854                         }
 855                 }
 856
 857                 $pageInfo['revisionCount']++;
 858                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 859                         $pageInfo['successfulRevisionCount']++;
 860                 }
 861         }
 862
 863         /**
 864          * @param array $pageInfo
 865          * @param array $revisionInfo
 866          * @throws MWException
 867          * @return bool|mixed
 868          */
 869         private function processRevision( $pageInfo, $revisionInfo ) {
 870                 global $wgMaxArticleSize;
 871
 872                 // Make sure revisions won't violate $wgMaxArticleSize, which could lead to
 873                 // database errors and instability. Testing for revisions with only listed
 874                 // content models, as other content models might use serialization formats
 875                 // which aren't checked against $wgMaxArticleSize.
 876                 if ( ( !isset( $revisionInfo['model'] ) ||
 877                         in_array( $revisionInfo['model'], [
 878                                 'wikitext',
 879                                 'css',
 880                                 'json',
 881                                 'javascript',
 882                                 'text',
 883                                 ''
 884                         ] ) ) &&
 885                         strlen( $revisionInfo['text'] ) > $wgMaxArticleSize * 1024
 886                 ) {
 887                         throw new MWException( 'The text of ' .
 888                                 ( isset( $revisionInfo['id'] ) ?
 889                                         "the revision with ID $revisionInfo[id]" :
 890                                         'a revision'
 891                                 ) . " exceeds the maximum allowable size ($wgMaxArticleSize KB)" );
 892                 }
 893
 894                 // FIXME: process schema version 11!
 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                 $revision->setTimestamp( $revisionInfo['timestamp'] ?? wfTimestampNow() );
 917
 918                 if ( isset( $revisionInfo['comment'] ) ) {
 919                         $revision->setComment( $revisionInfo['comment'] );
 920                 }
 921
 922                 if ( isset( $revisionInfo['minor'] ) ) {
 923                         $revision->setMinor( true );
 924                 }
 925                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 926                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 927                 } elseif ( isset( $revisionInfo['contributor']['username'] ) ) {
 928                         $revision->setUsername(
 929                                 $this->externalUserNames->applyPrefix( $revisionInfo['contributor']['username'] )
 930                         );
 931                 } else {
 932                         $revision->setUsername( $this->externalUserNames->addPrefix( 'Unknown user' ) );
 933                 }
 934                 if ( isset( $revisionInfo['sha1'] ) ) {
 935                         $revision->setSha1Base36( $revisionInfo['sha1'] );
 936                 }
 937                 $revision->setNoUpdates( $this->mNoUpdates );
 938
 939                 return $this->revisionCallback( $revision );
 940         }
 941
 942         /**
 943          * @param array $pageInfo
 944          * @return mixed
 945          */
 946         private function handleUpload( &$pageInfo ) {
 947                 $this->debug( "Enter upload handler" );
 948                 $uploadInfo = [];
 949
 950                 $normalFields = [ 'timestamp', 'comment', 'filename', 'text',
 951                                         'src', 'size', 'sha1base36', 'archivename', 'rel' ];
 952
 953                 $skip = false;
 954
 955                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 956                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 957                                         $this->reader->localName == 'upload' ) {
 958                                 break;
 959                         }
 960
 961                         $tag = $this->reader->localName;
 962
 963                         if ( !Hooks::run( 'ImportHandleUploadXMLTag', [
 964                                 $this, $pageInfo
 965                         ] ) ) {
 966                                 // Do nothing
 967                         } elseif ( in_array( $tag, $normalFields ) ) {
 968                                 $uploadInfo[$tag] = $this->nodeContents();
 969                         } elseif ( $tag == 'contributor' ) {
 970                                 $uploadInfo['contributor'] = $this->handleContributor();
 971                         } elseif ( $tag == 'contents' ) {
 972                                 $contents = $this->nodeContents();
 973                                 $encoding = $this->reader->getAttribute( 'encoding' );
 974                                 if ( $encoding === 'base64' ) {
 975                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 976                                         $uploadInfo['isTempSrc'] = true;
 977                                 }
 978                         } elseif ( $tag != '#text' ) {
 979                                 $this->warn( "Unhandled upload XML tag $tag" );
 980                                 $skip = true;
 981                         }
 982                 }
 983
 984                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 985                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 986                         if ( file_exists( $path ) ) {
 987                                 $uploadInfo['fileSrc'] = $path;
 988                                 $uploadInfo['isTempSrc'] = false;
 989                         }
 990                 }
 991
 992                 if ( $this->mImportUploads ) {
 993                         return $this->processUpload( $pageInfo, $uploadInfo );
 994                 }
 995         }
 996
 997         /**
 998          * @param string $contents
 999          * @return string
1000          */
1001         private function dumpTemp( $contents ) {
1002                 $filename = tempnam( wfTempDir(), 'importupload' );
1003                 file_put_contents( $filename, $contents );
1004                 return $filename;
1005         }
1006
1007         /**
1008          * @param array $pageInfo
1009          * @param array $uploadInfo
1010          * @return mixed
1011          */
1012         private function processUpload( $pageInfo, $uploadInfo ) {
1013                 $revision = new WikiRevision( $this->config );
1014                 $text = $uploadInfo['text'] ?? '';
1015
1016                 $revision->setTitle( $pageInfo['_title'] );
1017                 $revision->setID( $pageInfo['id'] );
1018                 $revision->setTimestamp( $uploadInfo['timestamp'] );
1019                 $revision->setText( $text );
1020                 $revision->setFilename( $uploadInfo['filename'] );
1021                 if ( isset( $uploadInfo['archivename'] ) ) {
1022                         $revision->setArchiveName( $uploadInfo['archivename'] );
1023                 }
1024                 $revision->setSrc( $uploadInfo['src'] );
1025                 if ( isset( $uploadInfo['fileSrc'] ) ) {
1026                         $revision->setFileSrc( $uploadInfo['fileSrc'],
1027                                 !empty( $uploadInfo['isTempSrc'] ) );
1028                 }
1029                 if ( isset( $uploadInfo['sha1base36'] ) ) {
1030                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
1031                 }
1032                 $revision->setSize( intval( $uploadInfo['size'] ) );
1033                 $revision->setComment( $uploadInfo['comment'] );
1034
1035                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
1036                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
1037                 }
1038                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
1039                         $revision->setUsername(
1040                                 $this->externalUserNames->applyPrefix( $uploadInfo['contributor']['username'] )
1041                         );
1042                 }
1043                 $revision->setNoUpdates( $this->mNoUpdates );
1044
1045                 return call_user_func( $this->mUploadCallback, $revision );
1046         }
1047
1048         /**
1049          * @return array
1050          */
1051         private function handleContributor() {
1052                 $fields = [ 'id', 'ip', 'username' ];
1053                 $info = [];
1054
1055                 if ( $this->reader->isEmptyElement ) {
1056                         return $info;
1057                 }
1058                 while ( $this->reader->read() ) {
1059                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
1060                                         $this->reader->localName == 'contributor' ) {
1061                                 break;
1062                         }
1063
1064                         $tag = $this->reader->localName;
1065
1066                         if ( in_array( $tag, $fields ) ) {
1067                                 $info[$tag] = $this->nodeContents();
1068                         }
1069                 }
1070
1071                 return $info;
1072         }
1073
1074         /**
1075          * @param string $text
1076          * @param string|null $ns
1077          * @return array|bool
1078          */
1079         private function processTitle( $text, $ns = null ) {
1080                 if ( is_null( $this->foreignNamespaces ) ) {
1081                         $foreignTitleFactory = new NaiveForeignTitleFactory();
1082                 } else {
1083                         $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
1084                                 $this->foreignNamespaces );
1085                 }
1086
1087                 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
1088                         intval( $ns ) );
1089
1090                 $title = $this->importTitleFactory->createTitleFromForeignTitle(
1091                         $foreignTitle );
1092
1093                 $commandLineMode = $this->config->get( 'CommandLineMode' );
1094                 if ( is_null( $title ) ) {
1095                         # Invalid page title? Ignore the page
1096                         $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
1097                         return false;
1098                 } elseif ( $title->isExternal() ) {
1099                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
1100                         return false;
1101                 } elseif ( !$title->canExist() ) {
1102                         $this->notice( 'import-error-special', $title->getPrefixedText() );
1103                         return false;
1104                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
1105                         # Do not import if the importing wiki user cannot edit this page
1106                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
1107                         return false;
1108                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
1109                         # Do not import if the importing wiki user cannot create this page
1110                         $this->notice( 'import-error-create', $title->getPrefixedText() );
1111                         return false;
1112                 }
1113
1114                 return [ $title, $foreignTitle ];
1115         }
1116 }