includes/import/WikiImporter.php

   1 <?php
   2 /**
   3  * MediaWiki page data importer.
   4  *
   5  * Copyright © 2003,2005 Brion Vibber <brion@pobox.com>
   6  * https://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup SpecialPage
  25  */
  26 use MediaWiki\MediaWikiServices;
  27
  28 /**
  29  * XML file reader for the page data importer.
  30  *
  31  * implements Special:Import
  32  * @ingroup SpecialPage
  33  */
  34 class WikiImporter {
  35         private $reader = null;
  36         private $foreignNamespaces = null;
  37         private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
  38         private $mSiteInfoCallback, $mPageOutCallback;
  39         private $mNoticeCallback, $mDebug;
  40         private $mImportUploads, $mImageBasePath;
  41         private $mNoUpdates = false;
  42         private $pageOffset = 0;
  43         /** @var Config */
  44         private $config;
  45         /** @var ImportTitleFactory */
  46         private $importTitleFactory;
  47         /** @var array */
  48         private $countableCache = [];
  49         /** @var bool */
  50         private $disableStatisticsUpdate = false;
  51
  52         /**
  53          * Creates an ImportXMLReader drawing from the source provided
  54          * @param ImportSource $source
  55          * @param Config $config
  56          * @throws Exception
  57          */
  58         function __construct( ImportSource $source, Config $config = null ) {
  59                 if ( !class_exists( 'XMLReader' ) ) {
  60                         throw new Exception( 'Import requires PHP to have been compiled with libxml support' );
  61                 }
  62
  63                 $this->reader = new XMLReader();
  64                 if ( !$config ) {
  65                         wfDeprecated( __METHOD__ . ' without a Config instance', '1.25' );
  66                         $config = MediaWikiServices::getInstance()->getMainConfig();
  67                 }
  68                 $this->config = $config;
  69
  70                 if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) {
  71                         stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  72                 }
  73                 $id = UploadSourceAdapter::registerSource( $source );
  74
  75                 // Enable the entity loader, as it is needed for loading external URLs via
  76                 // XMLReader::open (T86036)
  77                 $oldDisable = libxml_disable_entity_loader( false );
  78                 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
  79                         $status = $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  80                 } else {
  81                         $status = $this->reader->open( "uploadsource://$id" );
  82                 }
  83                 if ( !$status ) {
  84                         $error = libxml_get_last_error();
  85                         libxml_disable_entity_loader( $oldDisable );
  86                         throw new MWException( 'Encountered an internal error while initializing WikiImporter object: ' .
  87                                 $error->message );
  88                 }
  89                 libxml_disable_entity_loader( $oldDisable );
  90
  91                 // Default callbacks
  92                 $this->setPageCallback( [ $this, 'beforeImportPage' ] );
  93                 $this->setRevisionCallback( [ $this, "importRevision" ] );
  94                 $this->setUploadCallback( [ $this, 'importUpload' ] );
  95                 $this->setLogItemCallback( [ $this, 'importLogItem' ] );
  96                 $this->setPageOutCallback( [ $this, 'finishImportPage' ] );
  97
  98                 $this->importTitleFactory = new NaiveImportTitleFactory();
  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                         echo wfMessage( $msg, $params )->text() . "\n";
 131                 }
 132         }
 133
 134         /**
 135          * Set debug mode...
 136          * @param bool $debug
 137          */
 138         function setDebug( $debug ) {
 139                 $this->mDebug = $debug;
 140         }
 141
 142         /**
 143          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
 144          * @param bool $noupdates
 145          */
 146         function setNoUpdates( $noupdates ) {
 147                 $this->mNoUpdates = $noupdates;
 148         }
 149
 150         /**
 151          * Sets 'pageOffset' value. So it will skip the first n-1 pages
 152          * and start from the nth page. It's 1-based indexing.
 153          * @param int $nthPage
 154          * @since 1.29
 155          */
 156         function setPageOffset( $nthPage ) {
 157                 $this->pageOffset = $nthPage;
 158         }
 159
 160         /**
 161          * Set a callback that displays notice messages
 162          *
 163          * @param callable $callback
 164          * @return callable
 165          */
 166         public function setNoticeCallback( $callback ) {
 167                 return wfSetVar( $this->mNoticeCallback, $callback );
 168         }
 169
 170         /**
 171          * Sets the action to perform as each new page in the stream is reached.
 172          * @param callable $callback
 173          * @return callable
 174          */
 175         public function setPageCallback( $callback ) {
 176                 $previous = $this->mPageCallback;
 177                 $this->mPageCallback = $callback;
 178                 return $previous;
 179         }
 180
 181         /**
 182          * Sets the action to perform as each page in the stream is completed.
 183          * Callback accepts the page title (as a Title object), a second object
 184          * with the original title form (in case it's been overridden into a
 185          * local namespace), and a count of revisions.
 186          *
 187          * @param callable $callback
 188          * @return callable
 189          */
 190         public function setPageOutCallback( $callback ) {
 191                 $previous = $this->mPageOutCallback;
 192                 $this->mPageOutCallback = $callback;
 193                 return $previous;
 194         }
 195
 196         /**
 197          * Sets the action to perform as each page revision is reached.
 198          * @param callable $callback
 199          * @return callable
 200          */
 201         public function setRevisionCallback( $callback ) {
 202                 $previous = $this->mRevisionCallback;
 203                 $this->mRevisionCallback = $callback;
 204                 return $previous;
 205         }
 206
 207         /**
 208          * Sets the action to perform as each file upload version is reached.
 209          * @param callable $callback
 210          * @return callable
 211          */
 212         public function setUploadCallback( $callback ) {
 213                 $previous = $this->mUploadCallback;
 214                 $this->mUploadCallback = $callback;
 215                 return $previous;
 216         }
 217
 218         /**
 219          * Sets the action to perform as each log item reached.
 220          * @param callable $callback
 221          * @return callable
 222          */
 223         public function setLogItemCallback( $callback ) {
 224                 $previous = $this->mLogItemCallback;
 225                 $this->mLogItemCallback = $callback;
 226                 return $previous;
 227         }
 228
 229         /**
 230          * Sets the action to perform when site info is encountered
 231          * @param callable $callback
 232          * @return callable
 233          */
 234         public function setSiteInfoCallback( $callback ) {
 235                 $previous = $this->mSiteInfoCallback;
 236                 $this->mSiteInfoCallback = $callback;
 237                 return $previous;
 238         }
 239
 240         /**
 241          * Sets the factory object to use to convert ForeignTitle objects into local
 242          * Title objects
 243          * @param ImportTitleFactory $factory
 244          */
 245         public function setImportTitleFactory( $factory ) {
 246                 $this->importTitleFactory = $factory;
 247         }
 248
 249         /**
 250          * Set a target namespace to override the defaults
 251          * @param null|int $namespace
 252          * @return bool
 253          */
 254         public function setTargetNamespace( $namespace ) {
 255                 if ( is_null( $namespace ) ) {
 256                         // Don't override namespaces
 257                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 258                         return true;
 259                 } elseif (
 260                         $namespace >= 0 &&
 261                         MWNamespace::exists( intval( $namespace ) )
 262                 ) {
 263                         $namespace = intval( $namespace );
 264                         $this->setImportTitleFactory( new NamespaceImportTitleFactory( $namespace ) );
 265                         return true;
 266                 } else {
 267                         return false;
 268                 }
 269         }
 270
 271         /**
 272          * Set a target root page under which all pages are imported
 273          * @param null|string $rootpage
 274          * @return Status
 275          */
 276         public function setTargetRootPage( $rootpage ) {
 277                 $status = Status::newGood();
 278                 if ( is_null( $rootpage ) ) {
 279                         // No rootpage
 280                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 281                 } elseif ( $rootpage !== '' ) {
 282                         $rootpage = rtrim( $rootpage, '/' ); // avoid double slashes
 283                         $title = Title::newFromText( $rootpage );
 284
 285                         if ( !$title || $title->isExternal() ) {
 286                                 $status->fatal( 'import-rootpage-invalid' );
 287                         } else {
 288                                 if ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 289                                         global $wgContLang;
 290
 291                                         $displayNSText = $title->getNamespace() == NS_MAIN
 292                                                 ? wfMessage( 'blanknamespace' )->text()
 293                                                 : $wgContLang->getNsText( $title->getNamespace() );
 294                                         $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 295                                 } else {
 296                                         // set namespace to 'all', so the namespace check in processTitle() can pass
 297                                         $this->setTargetNamespace( null );
 298                                         $this->setImportTitleFactory( new SubpageImportTitleFactory( $title ) );
 299                                 }
 300                         }
 301                 }
 302                 return $status;
 303         }
 304
 305         /**
 306          * @param string $dir
 307          */
 308         public function setImageBasePath( $dir ) {
 309                 $this->mImageBasePath = $dir;
 310         }
 311
 312         /**
 313          * @param bool $import
 314          */
 315         public function setImportUploads( $import ) {
 316                 $this->mImportUploads = $import;
 317         }
 318
 319         /**
 320          * Statistics update can cause a lot of time
 321          * @since 1.29
 322          */
 323         public function disableStatisticsUpdate() {
 324                 $this->disableStatisticsUpdate = true;
 325         }
 326
 327         /**
 328          * Default per-page callback. Sets up some things related to site statistics
 329          * @param array $titleAndForeignTitle Two-element array, with Title object at
 330          * index 0 and ForeignTitle object at index 1
 331          * @return bool
 332          */
 333         public function beforeImportPage( $titleAndForeignTitle ) {
 334                 $title = $titleAndForeignTitle[0];
 335                 $page = WikiPage::factory( $title );
 336                 $this->countableCache['title_' . $title->getPrefixedText()] = $page->isCountable();
 337                 return true;
 338         }
 339
 340         /**
 341          * Default per-revision callback, performs the import.
 342          * @param WikiRevision $revision
 343          * @return bool
 344          */
 345         public function importRevision( $revision ) {
 346                 if ( !$revision->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
 347                         $this->notice( 'import-error-bad-location',
 348                                 $revision->getTitle()->getPrefixedText(),
 349                                 $revision->getID(),
 350                                 $revision->getModel(),
 351                                 $revision->getFormat() );
 352
 353                         return false;
 354                 }
 355
 356                 try {
 357                         return $revision->importOldRevision();
 358                 } catch ( MWContentSerializationException $ex ) {
 359                         $this->notice( 'import-error-unserialize',
 360                                 $revision->getTitle()->getPrefixedText(),
 361                                 $revision->getID(),
 362                                 $revision->getModel(),
 363                                 $revision->getFormat() );
 364                 }
 365
 366                 return false;
 367         }
 368
 369         /**
 370          * Default per-revision callback, performs the import.
 371          * @param WikiRevision $revision
 372          * @return bool
 373          */
 374         public function importLogItem( $revision ) {
 375                 return $revision->importLogItem();
 376         }
 377
 378         /**
 379          * Dummy for now...
 380          * @param WikiRevision $revision
 381          * @return bool
 382          */
 383         public function importUpload( $revision ) {
 384                 return $revision->importUpload();
 385         }
 386
 387         /**
 388          * Mostly for hook use
 389          * @param Title $title
 390          * @param ForeignTitle $foreignTitle
 391          * @param int $revCount
 392          * @param int $sRevCount
 393          * @param array $pageInfo
 394          * @return bool
 395          */
 396         public function finishImportPage( $title, $foreignTitle, $revCount,
 397                 $sRevCount, $pageInfo
 398         ) {
 399                 // Update article count statistics (T42009)
 400                 // The normal counting logic in WikiPage->doEditUpdates() is designed for
 401                 // one-revision-at-a-time editing, not bulk imports. In this situation it
 402                 // suffers from issues of replica DB lag. We let WikiPage handle the total page
 403                 // and revision count, and we implement our own custom logic for the
 404                 // article (content page) count.
 405                 if ( !$this->disableStatisticsUpdate ) {
 406                         $page = WikiPage::factory( $title );
 407                         $page->loadPageData( 'fromdbmaster' );
 408                         $content = $page->getContent();
 409                         if ( $content === null ) {
 410                                 wfDebug( __METHOD__ . ': Skipping article count adjustment for ' . $title .
 411                                         ' because WikiPage::getContent() returned null' );
 412                         } else {
 413                                 $editInfo = $page->prepareContentForEdit( $content );
 414                                 $countKey = 'title_' . $title->getPrefixedText();
 415                                 $countable = $page->isCountable( $editInfo );
 416                                 if ( array_key_exists( $countKey, $this->countableCache ) &&
 417                                         $countable != $this->countableCache[$countKey] ) {
 418                                         DeferredUpdates::addUpdate( SiteStatsUpdate::factory( [
 419                                                 'articles' => ( (int)$countable - (int)$this->countableCache[$countKey] )
 420                                         ] ) );
 421                                 }
 422                         }
 423                 }
 424
 425                 $args = func_get_args();
 426                 return Hooks::run( 'AfterImportPage', $args );
 427         }
 428
 429         /**
 430          * Alternate per-revision callback, for debugging.
 431          * @param WikiRevision $revision
 432          */
 433         public function debugRevisionHandler( &$revision ) {
 434                 $this->debug( "Got revision:" );
 435                 if ( is_object( $revision->title ) ) {
 436                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 437                 } else {
 438                         $this->debug( "-- Title: <invalid>" );
 439                 }
 440                 $this->debug( "-- User: " . $revision->user_text );
 441                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 442                 $this->debug( "-- Comment: " . $revision->comment );
 443                 $this->debug( "-- Text: " . $revision->text );
 444         }
 445
 446         /**
 447          * Notify the callback function of site info
 448          * @param array $siteInfo
 449          * @return bool|mixed
 450          */
 451         private function siteInfoCallback( $siteInfo ) {
 452                 if ( isset( $this->mSiteInfoCallback ) ) {
 453                         return call_user_func_array( $this->mSiteInfoCallback,
 454                                         [ $siteInfo, $this ] );
 455                 } else {
 456                         return false;
 457                 }
 458         }
 459
 460         /**
 461          * Notify the callback function when a new "<page>" is reached.
 462          * @param Title $title
 463          */
 464         function pageCallback( $title ) {
 465                 if ( isset( $this->mPageCallback ) ) {
 466                         call_user_func( $this->mPageCallback, $title );
 467                 }
 468         }
 469
 470         /**
 471          * Notify the callback function when a "</page>" is closed.
 472          * @param Title $title
 473          * @param ForeignTitle $foreignTitle
 474          * @param int $revCount
 475          * @param int $sucCount Number of revisions for which callback returned true
 476          * @param array $pageInfo Associative array of page information
 477          */
 478         private function pageOutCallback( $title, $foreignTitle, $revCount,
 479                         $sucCount, $pageInfo ) {
 480                 if ( isset( $this->mPageOutCallback ) ) {
 481                         $args = func_get_args();
 482                         call_user_func_array( $this->mPageOutCallback, $args );
 483                 }
 484         }
 485
 486         /**
 487          * Notify the callback function of a revision
 488          * @param WikiRevision $revision
 489          * @return bool|mixed
 490          */
 491         private function revisionCallback( $revision ) {
 492                 if ( isset( $this->mRevisionCallback ) ) {
 493                         return call_user_func_array( $this->mRevisionCallback,
 494                                         [ $revision, $this ] );
 495                 } else {
 496                         return false;
 497                 }
 498         }
 499
 500         /**
 501          * Notify the callback function of a new log item
 502          * @param WikiRevision $revision
 503          * @return bool|mixed
 504          */
 505         private function logItemCallback( $revision ) {
 506                 if ( isset( $this->mLogItemCallback ) ) {
 507                         return call_user_func_array( $this->mLogItemCallback,
 508                                         [ $revision, $this ] );
 509                 } else {
 510                         return false;
 511                 }
 512         }
 513
 514         /**
 515          * Retrieves the contents of the named attribute of the current element.
 516          * @param string $attr The name of the attribute
 517          * @return string The value of the attribute or an empty string if it is not set in the current
 518          * element.
 519          */
 520         public function nodeAttribute( $attr ) {
 521                 return $this->reader->getAttribute( $attr );
 522         }
 523
 524         /**
 525          * Shouldn't something like this be built-in to XMLReader?
 526          * Fetches text contents of the current element, assuming
 527          * no sub-elements or such scary things.
 528          * @return string
 529          * @access private
 530          */
 531         public function nodeContents() {
 532                 if ( $this->reader->isEmptyElement ) {
 533                         return "";
 534                 }
 535                 $buffer = "";
 536                 while ( $this->reader->read() ) {
 537                         switch ( $this->reader->nodeType ) {
 538                         case XMLReader::TEXT:
 539                         case XMLReader::CDATA:
 540                         case XMLReader::SIGNIFICANT_WHITESPACE:
 541                                 $buffer .= $this->reader->value;
 542                                 break;
 543                         case XMLReader::END_ELEMENT:
 544                                 return $buffer;
 545                         }
 546                 }
 547
 548                 $this->reader->close();
 549                 return '';
 550         }
 551
 552         /**
 553          * Primary entry point
 554          * @throws MWException
 555          * @return bool
 556          */
 557         public function doImport() {
 558                 // Calls to reader->read need to be wrapped in calls to
 559                 // libxml_disable_entity_loader() to avoid local file
 560                 // inclusion attacks (T48932).
 561                 $oldDisable = libxml_disable_entity_loader( true );
 562                 $this->reader->read();
 563
 564                 if ( $this->reader->localName != 'mediawiki' ) {
 565                         libxml_disable_entity_loader( $oldDisable );
 566                         throw new MWException( "Expected <mediawiki> tag, got " .
 567                                 $this->reader->localName );
 568                 }
 569                 $this->debug( "<mediawiki> tag is correct." );
 570
 571                 $this->debug( "Starting primary dump processing loop." );
 572
 573                 $keepReading = $this->reader->read();
 574                 $skip = false;
 575                 $rethrow = null;
 576                 $pageCount = 0;
 577                 try {
 578                         while ( $keepReading ) {
 579                                 $tag = $this->reader->localName;
 580                                 if ( $this->pageOffset ) {
 581                                         if ( $tag === 'page' ) {
 582                                                 $pageCount++;
 583                                         }
 584                                         if ( $pageCount < $this->pageOffset ) {
 585                                                 $keepReading = $this->reader->next();
 586                                                 continue;
 587                                         }
 588                                 }
 589                                 $type = $this->reader->nodeType;
 590
 591                                 if ( !Hooks::run( 'ImportHandleToplevelXMLTag', [ $this ] ) ) {
 592                                         // Do nothing
 593                                 } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) {
 594                                         break;
 595                                 } elseif ( $tag == 'siteinfo' ) {
 596                                         $this->handleSiteInfo();
 597                                 } elseif ( $tag == 'page' ) {
 598                                         $this->handlePage();
 599                                 } elseif ( $tag == 'logitem' ) {
 600                                         $this->handleLogItem();
 601                                 } elseif ( $tag != '#text' ) {
 602                                         $this->warn( "Unhandled top-level XML tag $tag" );
 603
 604                                         $skip = true;
 605                                 }
 606
 607                                 if ( $skip ) {
 608                                         $keepReading = $this->reader->next();
 609                                         $skip = false;
 610                                         $this->debug( "Skip" );
 611                                 } else {
 612                                         $keepReading = $this->reader->read();
 613                                 }
 614                         }
 615                 } catch ( Exception $ex ) {
 616                         $rethrow = $ex;
 617                 }
 618
 619                 // finally
 620                 libxml_disable_entity_loader( $oldDisable );
 621                 $this->reader->close();
 622
 623                 if ( $rethrow ) {
 624                         throw $rethrow;
 625                 }
 626
 627                 return true;
 628         }
 629
 630         private function handleSiteInfo() {
 631                 $this->debug( "Enter site info handler." );
 632                 $siteInfo = [];
 633
 634                 // Fields that can just be stuffed in the siteInfo object
 635                 $normalFields = [ 'sitename', 'base', 'generator', 'case' ];
 636
 637                 while ( $this->reader->read() ) {
 638                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 639                                         $this->reader->localName == 'siteinfo' ) {
 640                                 break;
 641                         }
 642
 643                         $tag = $this->reader->localName;
 644
 645                         if ( $tag == 'namespace' ) {
 646                                 $this->foreignNamespaces[$this->nodeAttribute( 'key' )] =
 647                                         $this->nodeContents();
 648                         } elseif ( in_array( $tag, $normalFields ) ) {
 649                                 $siteInfo[$tag] = $this->nodeContents();
 650                         }
 651                 }
 652
 653                 $siteInfo['_namespaces'] = $this->foreignNamespaces;
 654                 $this->siteInfoCallback( $siteInfo );
 655         }
 656
 657         private function handleLogItem() {
 658                 $this->debug( "Enter log item handler." );
 659                 $logInfo = [];
 660
 661                 // Fields that can just be stuffed in the pageInfo object
 662                 $normalFields = [ 'id', 'comment', 'type', 'action', 'timestamp',
 663                                         'logtitle', 'params' ];
 664
 665                 while ( $this->reader->read() ) {
 666                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 667                                         $this->reader->localName == 'logitem' ) {
 668                                 break;
 669                         }
 670
 671                         $tag = $this->reader->localName;
 672
 673                         if ( !Hooks::run( 'ImportHandleLogItemXMLTag', [
 674                                 $this, $logInfo
 675                         ] ) ) {
 676                                 // Do nothing
 677                         } elseif ( in_array( $tag, $normalFields ) ) {
 678                                 $logInfo[$tag] = $this->nodeContents();
 679                         } elseif ( $tag == 'contributor' ) {
 680                                 $logInfo['contributor'] = $this->handleContributor();
 681                         } elseif ( $tag != '#text' ) {
 682                                 $this->warn( "Unhandled log-item XML tag $tag" );
 683                         }
 684                 }
 685
 686                 $this->processLogItem( $logInfo );
 687         }
 688
 689         /**
 690          * @param array $logInfo
 691          * @return bool|mixed
 692          */
 693         private function processLogItem( $logInfo ) {
 694                 $revision = new WikiRevision( $this->config );
 695
 696                 if ( isset( $logInfo['id'] ) ) {
 697                         $revision->setID( $logInfo['id'] );
 698                 }
 699                 $revision->setType( $logInfo['type'] );
 700                 $revision->setAction( $logInfo['action'] );
 701                 if ( isset( $logInfo['timestamp'] ) ) {
 702                         $revision->setTimestamp( $logInfo['timestamp'] );
 703                 }
 704                 if ( isset( $logInfo['params'] ) ) {
 705                         $revision->setParams( $logInfo['params'] );
 706                 }
 707                 if ( isset( $logInfo['logtitle'] ) ) {
 708                         // @todo Using Title for non-local titles is a recipe for disaster.
 709                         // We should use ForeignTitle here instead.
 710                         $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 711                 }
 712
 713                 $revision->setNoUpdates( $this->mNoUpdates );
 714
 715                 if ( isset( $logInfo['comment'] ) ) {
 716                         $revision->setComment( $logInfo['comment'] );
 717                 }
 718
 719                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 720                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 721                 }
 722
 723                 if ( !isset( $logInfo['contributor']['username'] ) ) {
 724                         $revision->setUsername( 'Unknown user' );
 725                 } else {
 726                         $revision->setUsername( $logInfo['contributor']['username'] );
 727                 }
 728
 729                 return $this->logItemCallback( $revision );
 730         }
 731
 732         private function handlePage() {
 733                 // Handle page data.
 734                 $this->debug( "Enter page handler." );
 735                 $pageInfo = [ 'revisionCount' => 0, 'successfulRevisionCount' => 0 ];
 736
 737                 // Fields that can just be stuffed in the pageInfo object
 738                 $normalFields = [ 'title', 'ns', 'id', 'redirect', 'restrictions' ];
 739
 740                 $skip = false;
 741                 $badTitle = false;
 742
 743                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 744                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 745                                         $this->reader->localName == 'page' ) {
 746                                 break;
 747                         }
 748
 749                         $skip = false;
 750
 751                         $tag = $this->reader->localName;
 752
 753                         if ( $badTitle ) {
 754                                 // The title is invalid, bail out of this page
 755                                 $skip = true;
 756                         } elseif ( !Hooks::run( 'ImportHandlePageXMLTag', [ $this,
 757                                                 &$pageInfo ] ) ) {
 758                                 // Do nothing
 759                         } elseif ( in_array( $tag, $normalFields ) ) {
 760                                 // An XML snippet:
 761                                 // <page>
 762                                 //     <id>123</id>
 763                                 //     <title>Page</title>
 764                                 //     <redirect title="NewTitle"/>
 765                                 //     ...
 766                                 // Because the redirect tag is built differently, we need special handling for that case.
 767                                 if ( $tag == 'redirect' ) {
 768                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 769                                 } else {
 770                                         $pageInfo[$tag] = $this->nodeContents();
 771                                 }
 772                         } elseif ( $tag == 'revision' || $tag == 'upload' ) {
 773                                 if ( !isset( $title ) ) {
 774                                         $title = $this->processTitle( $pageInfo['title'],
 775                                                 isset( $pageInfo['ns'] ) ? $pageInfo['ns'] : null );
 776
 777                                         // $title is either an array of two titles or false.
 778                                         if ( is_array( $title ) ) {
 779                                                 $this->pageCallback( $title );
 780                                                 list( $pageInfo['_title'], $foreignTitle ) = $title;
 781                                         } else {
 782                                                 $badTitle = true;
 783                                                 $skip = true;
 784                                         }
 785                                 }
 786
 787                                 if ( $title ) {
 788                                         if ( $tag == 'revision' ) {
 789                                                 $this->handleRevision( $pageInfo );
 790                                         } else {
 791                                                 $this->handleUpload( $pageInfo );
 792                                         }
 793                                 }
 794                         } elseif ( $tag != '#text' ) {
 795                                 $this->warn( "Unhandled page XML tag $tag" );
 796                                 $skip = true;
 797                         }
 798                 }
 799
 800                 // @note $pageInfo is only set if a valid $title is processed above with
 801                 //       no error. If we have a valid $title, then pageCallback is called
 802                 //       above, $pageInfo['title'] is set and we do pageOutCallback here.
 803                 //       If $pageInfo['_title'] is not set, then $foreignTitle is also not
 804                 //       set since they both come from $title above.
 805                 if ( array_key_exists( '_title', $pageInfo ) ) {
 806                         $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
 807                                         $pageInfo['revisionCount'],
 808                                         $pageInfo['successfulRevisionCount'],
 809                                         $pageInfo );
 810                 }
 811         }
 812
 813         /**
 814          * @param array $pageInfo
 815          */
 816         private function handleRevision( &$pageInfo ) {
 817                 $this->debug( "Enter revision handler" );
 818                 $revisionInfo = [];
 819
 820                 $normalFields = [ 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' ];
 821
 822                 $skip = false;
 823
 824                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 825                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 826                                         $this->reader->localName == 'revision' ) {
 827                                 break;
 828                         }
 829
 830                         $tag = $this->reader->localName;
 831
 832                         if ( !Hooks::run( 'ImportHandleRevisionXMLTag', [
 833                                 $this, $pageInfo, $revisionInfo
 834                         ] ) ) {
 835                                 // Do nothing
 836                         } elseif ( in_array( $tag, $normalFields ) ) {
 837                                 $revisionInfo[$tag] = $this->nodeContents();
 838                         } elseif ( $tag == 'contributor' ) {
 839                                 $revisionInfo['contributor'] = $this->handleContributor();
 840                         } elseif ( $tag != '#text' ) {
 841                                 $this->warn( "Unhandled revision XML tag $tag" );
 842                                 $skip = true;
 843                         }
 844                 }
 845
 846                 $pageInfo['revisionCount']++;
 847                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 848                         $pageInfo['successfulRevisionCount']++;
 849                 }
 850         }
 851
 852         /**
 853          * @param array $pageInfo
 854          * @param array $revisionInfo
 855          * @return bool|mixed
 856          */
 857         private function processRevision( $pageInfo, $revisionInfo ) {
 858                 global $wgMaxArticleSize;
 859
 860                 // Make sure revisions won't violate $wgMaxArticleSize, which could lead to
 861                 // database errors and instability. Testing for revisions with only listed
 862                 // content models, as other content models might use serialization formats
 863                 // which aren't checked against $wgMaxArticleSize.
 864                 if ( ( !isset( $revisionInfo['model'] ) ||
 865                         in_array( $revisionInfo['model'], [
 866                                 'wikitext',
 867                                 'css',
 868                                 'json',
 869                                 'javascript',
 870                                 'text',
 871                                 ''
 872                         ] ) ) &&
 873                         strlen( $revisionInfo['text'] ) > $wgMaxArticleSize * 1024
 874                 ) {
 875                         throw new MWException( 'The text of ' .
 876                                 ( isset( $revisionInfo['id'] ) ?
 877                                         "the revision with ID $revisionInfo[id]" :
 878                                         'a revision'
 879                                 ) . " exceeds the maximum allowable size ($wgMaxArticleSize KB)" );
 880                 }
 881
 882                 $revision = new WikiRevision( $this->config );
 883
 884                 if ( isset( $revisionInfo['id'] ) ) {
 885                         $revision->setID( $revisionInfo['id'] );
 886                 }
 887                 if ( isset( $revisionInfo['model'] ) ) {
 888                         $revision->setModel( $revisionInfo['model'] );
 889                 }
 890                 if ( isset( $revisionInfo['format'] ) ) {
 891                         $revision->setFormat( $revisionInfo['format'] );
 892                 }
 893                 $revision->setTitle( $pageInfo['_title'] );
 894
 895                 if ( isset( $revisionInfo['text'] ) ) {
 896                         $handler = $revision->getContentHandler();
 897                         $text = $handler->importTransform(
 898                                 $revisionInfo['text'],
 899                                 $revision->getFormat() );
 900
 901                         $revision->setText( $text );
 902                 }
 903                 if ( isset( $revisionInfo['timestamp'] ) ) {
 904                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 905                 } else {
 906                         $revision->setTimestamp( wfTimestampNow() );
 907                 }
 908
 909                 if ( isset( $revisionInfo['comment'] ) ) {
 910                         $revision->setComment( $revisionInfo['comment'] );
 911                 }
 912
 913                 if ( isset( $revisionInfo['minor'] ) ) {
 914                         $revision->setMinor( true );
 915                 }
 916                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 917                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 918                 } elseif ( isset( $revisionInfo['contributor']['username'] ) ) {
 919                         $revision->setUsername( $revisionInfo['contributor']['username'] );
 920                 } else {
 921                         $revision->setUsername( 'Unknown user' );
 922                 }
 923                 $revision->setNoUpdates( $this->mNoUpdates );
 924
 925                 return $this->revisionCallback( $revision );
 926         }
 927
 928         /**
 929          * @param array $pageInfo
 930          * @return mixed
 931          */
 932         private function handleUpload( &$pageInfo ) {
 933                 $this->debug( "Enter upload handler" );
 934                 $uploadInfo = [];
 935
 936                 $normalFields = [ 'timestamp', 'comment', 'filename', 'text',
 937                                         'src', 'size', 'sha1base36', 'archivename', 'rel' ];
 938
 939                 $skip = false;
 940
 941                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 942                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 943                                         $this->reader->localName == 'upload' ) {
 944                                 break;
 945                         }
 946
 947                         $tag = $this->reader->localName;
 948
 949                         if ( !Hooks::run( 'ImportHandleUploadXMLTag', [
 950                                 $this, $pageInfo
 951                         ] ) ) {
 952                                 // Do nothing
 953                         } elseif ( in_array( $tag, $normalFields ) ) {
 954                                 $uploadInfo[$tag] = $this->nodeContents();
 955                         } elseif ( $tag == 'contributor' ) {
 956                                 $uploadInfo['contributor'] = $this->handleContributor();
 957                         } elseif ( $tag == 'contents' ) {
 958                                 $contents = $this->nodeContents();
 959                                 $encoding = $this->reader->getAttribute( 'encoding' );
 960                                 if ( $encoding === 'base64' ) {
 961                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 962                                         $uploadInfo['isTempSrc'] = true;
 963                                 }
 964                         } elseif ( $tag != '#text' ) {
 965                                 $this->warn( "Unhandled upload XML tag $tag" );
 966                                 $skip = true;
 967                         }
 968                 }
 969
 970                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 971                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 972                         if ( file_exists( $path ) ) {
 973                                 $uploadInfo['fileSrc'] = $path;
 974                                 $uploadInfo['isTempSrc'] = false;
 975                         }
 976                 }
 977
 978                 if ( $this->mImportUploads ) {
 979                         return $this->processUpload( $pageInfo, $uploadInfo );
 980                 }
 981         }
 982
 983         /**
 984          * @param string $contents
 985          * @return string
 986          */
 987         private function dumpTemp( $contents ) {
 988                 $filename = tempnam( wfTempDir(), 'importupload' );
 989                 file_put_contents( $filename, $contents );
 990                 return $filename;
 991         }
 992
 993         /**
 994          * @param array $pageInfo
 995          * @param array $uploadInfo
 996          * @return mixed
 997          */
 998         private function processUpload( $pageInfo, $uploadInfo ) {
 999                 $revision = new WikiRevision( $this->config );
1000                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
1001
1002                 $revision->setTitle( $pageInfo['_title'] );
1003                 $revision->setID( $pageInfo['id'] );
1004                 $revision->setTimestamp( $uploadInfo['timestamp'] );
1005                 $revision->setText( $text );
1006                 $revision->setFilename( $uploadInfo['filename'] );
1007                 if ( isset( $uploadInfo['archivename'] ) ) {
1008                         $revision->setArchiveName( $uploadInfo['archivename'] );
1009                 }
1010                 $revision->setSrc( $uploadInfo['src'] );
1011                 if ( isset( $uploadInfo['fileSrc'] ) ) {
1012                         $revision->setFileSrc( $uploadInfo['fileSrc'],
1013                                 !empty( $uploadInfo['isTempSrc'] ) );
1014                 }
1015                 if ( isset( $uploadInfo['sha1base36'] ) ) {
1016                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
1017                 }
1018                 $revision->setSize( intval( $uploadInfo['size'] ) );
1019                 $revision->setComment( $uploadInfo['comment'] );
1020
1021                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
1022                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
1023                 }
1024                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
1025                         $revision->setUsername( $uploadInfo['contributor']['username'] );
1026                 }
1027                 $revision->setNoUpdates( $this->mNoUpdates );
1028
1029                 return call_user_func( $this->mUploadCallback, $revision );
1030         }
1031
1032         /**
1033          * @return array
1034          */
1035         private function handleContributor() {
1036                 $fields = [ 'id', 'ip', 'username' ];
1037                 $info = [];
1038
1039                 if ( $this->reader->isEmptyElement ) {
1040                         return $info;
1041                 }
1042                 while ( $this->reader->read() ) {
1043                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
1044                                         $this->reader->localName == 'contributor' ) {
1045                                 break;
1046                         }
1047
1048                         $tag = $this->reader->localName;
1049
1050                         if ( in_array( $tag, $fields ) ) {
1051                                 $info[$tag] = $this->nodeContents();
1052                         }
1053                 }
1054
1055                 return $info;
1056         }
1057
1058         /**
1059          * @param string $text
1060          * @param string|null $ns
1061          * @return array|bool
1062          */
1063         private function processTitle( $text, $ns = null ) {
1064                 if ( is_null( $this->foreignNamespaces ) ) {
1065                         $foreignTitleFactory = new NaiveForeignTitleFactory();
1066                 } else {
1067                         $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
1068                                 $this->foreignNamespaces );
1069                 }
1070
1071                 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
1072                         intval( $ns ) );
1073
1074                 $title = $this->importTitleFactory->createTitleFromForeignTitle(
1075                         $foreignTitle );
1076
1077                 $commandLineMode = $this->config->get( 'CommandLineMode' );
1078                 if ( is_null( $title ) ) {
1079                         # Invalid page title? Ignore the page
1080                         $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
1081                         return false;
1082                 } elseif ( $title->isExternal() ) {
1083                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
1084                         return false;
1085                 } elseif ( !$title->canExist() ) {
1086                         $this->notice( 'import-error-special', $title->getPrefixedText() );
1087                         return false;
1088                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
1089                         # Do not import if the importing wiki user cannot edit this page
1090                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
1091                         return false;
1092                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
1093                         # Do not import if the importing wiki user cannot create this page
1094                         $this->notice( 'import-error-create', $title->getPrefixedText() );
1095                         return false;
1096                 }
1097
1098                 return [ $title, $foreignTitle ];
1099         }
1100 }