includes/import/WikiImporter.php

   1 <?php
   2 /**
   3  * MediaWiki page data importer.
   4  *
   5  * Copyright © 2003,2005 Brion Vibber <brion@pobox.com>
   6  * https://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup SpecialPage
  25  */
  26
  27 /**
  28  * XML file reader for the page data importer.
  29  *
  30  * implements Special:Import
  31  * @ingroup SpecialPage
  32  */
  33 class WikiImporter {
  34         private $reader = null;
  35         private $foreignNamespaces = null;
  36         private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
  37         private $mSiteInfoCallback, $mPageOutCallback;
  38         private $mNoticeCallback, $mDebug;
  39         private $mImportUploads, $mImageBasePath;
  40         private $mNoUpdates = false;
  41         private $pageOffset = 0;
  42         /** @var Config */
  43         private $config;
  44         /** @var ImportTitleFactory */
  45         private $importTitleFactory;
  46         /** @var array */
  47         private $countableCache = [];
  48         /** @var bool */
  49         private $disableStatisticsUpdate = false;
  50         private $usernamePrefix = 'imported';
  51         private $assignKnownUsers = false;
  52         private $triedCreations = [];
  53
  54         /**
  55          * Creates an ImportXMLReader drawing from the source provided
  56          * @param ImportSource $source
  57          * @param Config $config
  58          * @throws Exception
  59          */
  60         function __construct( ImportSource $source, Config $config ) {
  61                 if ( !class_exists( 'XMLReader' ) ) {
  62                         throw new Exception( 'Import requires PHP to have been compiled with libxml support' );
  63                 }
  64
  65                 $this->reader = new XMLReader();
  66                 $this->config = $config;
  67
  68                 if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) {
  69                         stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  70                 }
  71                 $id = UploadSourceAdapter::registerSource( $source );
  72
  73                 // Enable the entity loader, as it is needed for loading external URLs via
  74                 // XMLReader::open (T86036)
  75                 $oldDisable = libxml_disable_entity_loader( false );
  76                 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
  77                         $status = $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  78                 } else {
  79                         $status = $this->reader->open( "uploadsource://$id" );
  80                 }
  81                 if ( !$status ) {
  82                         $error = libxml_get_last_error();
  83                         libxml_disable_entity_loader( $oldDisable );
  84                         throw new MWException( 'Encountered an internal error while initializing WikiImporter object: ' .
  85                                 $error->message );
  86                 }
  87                 libxml_disable_entity_loader( $oldDisable );
  88
  89                 // Default callbacks
  90                 $this->setPageCallback( [ $this, 'beforeImportPage' ] );
  91                 $this->setRevisionCallback( [ $this, "importRevision" ] );
  92                 $this->setUploadCallback( [ $this, 'importUpload' ] );
  93                 $this->setLogItemCallback( [ $this, 'importLogItem' ] );
  94                 $this->setPageOutCallback( [ $this, 'finishImportPage' ] );
  95
  96                 $this->importTitleFactory = new NaiveImportTitleFactory();
  97         }
  98
  99         /**
 100          * @return null|XMLReader
 101          */
 102         public function getReader() {
 103                 return $this->reader;
 104         }
 105
 106         public function throwXmlError( $err ) {
 107                 $this->debug( "FAILURE: $err" );
 108                 wfDebug( "WikiImporter XML error: $err\n" );
 109         }
 110
 111         public function debug( $data ) {
 112                 if ( $this->mDebug ) {
 113                         wfDebug( "IMPORT: $data\n" );
 114                 }
 115         }
 116
 117         public function warn( $data ) {
 118                 wfDebug( "IMPORT: $data\n" );
 119         }
 120
 121         public function notice( $msg /*, $param, ...*/ ) {
 122                 $params = func_get_args();
 123                 array_shift( $params );
 124
 125                 if ( is_callable( $this->mNoticeCallback ) ) {
 126                         call_user_func( $this->mNoticeCallback, $msg, $params );
 127                 } else { # No ImportReporter -> CLI
 128                         echo wfMessage( $msg, $params )->text() . "\n";
 129                 }
 130         }
 131
 132         /**
 133          * Set debug mode...
 134          * @param bool $debug
 135          */
 136         function setDebug( $debug ) {
 137                 $this->mDebug = $debug;
 138         }
 139
 140         /**
 141          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
 142          * @param bool $noupdates
 143          */
 144         function setNoUpdates( $noupdates ) {
 145                 $this->mNoUpdates = $noupdates;
 146         }
 147
 148         /**
 149          * Sets 'pageOffset' value. So it will skip the first n-1 pages
 150          * and start from the nth page. It's 1-based indexing.
 151          * @param int $nthPage
 152          * @since 1.29
 153          */
 154         function setPageOffset( $nthPage ) {
 155                 $this->pageOffset = $nthPage;
 156         }
 157
 158         /**
 159          * Set a callback that displays notice messages
 160          *
 161          * @param callable $callback
 162          * @return callable
 163          */
 164         public function setNoticeCallback( $callback ) {
 165                 return wfSetVar( $this->mNoticeCallback, $callback );
 166         }
 167
 168         /**
 169          * Sets the action to perform as each new page in the stream is reached.
 170          * @param callable $callback
 171          * @return callable
 172          */
 173         public function setPageCallback( $callback ) {
 174                 $previous = $this->mPageCallback;
 175                 $this->mPageCallback = $callback;
 176                 return $previous;
 177         }
 178
 179         /**
 180          * Sets the action to perform as each page in the stream is completed.
 181          * Callback accepts the page title (as a Title object), a second object
 182          * with the original title form (in case it's been overridden into a
 183          * local namespace), and a count of revisions.
 184          *
 185          * @param callable $callback
 186          * @return callable
 187          */
 188         public function setPageOutCallback( $callback ) {
 189                 $previous = $this->mPageOutCallback;
 190                 $this->mPageOutCallback = $callback;
 191                 return $previous;
 192         }
 193
 194         /**
 195          * Sets the action to perform as each page revision is reached.
 196          * @param callable $callback
 197          * @return callable
 198          */
 199         public function setRevisionCallback( $callback ) {
 200                 $previous = $this->mRevisionCallback;
 201                 $this->mRevisionCallback = $callback;
 202                 return $previous;
 203         }
 204
 205         /**
 206          * Sets the action to perform as each file upload version is reached.
 207          * @param callable $callback
 208          * @return callable
 209          */
 210         public function setUploadCallback( $callback ) {
 211                 $previous = $this->mUploadCallback;
 212                 $this->mUploadCallback = $callback;
 213                 return $previous;
 214         }
 215
 216         /**
 217          * Sets the action to perform as each log item reached.
 218          * @param callable $callback
 219          * @return callable
 220          */
 221         public function setLogItemCallback( $callback ) {
 222                 $previous = $this->mLogItemCallback;
 223                 $this->mLogItemCallback = $callback;
 224                 return $previous;
 225         }
 226
 227         /**
 228          * Sets the action to perform when site info is encountered
 229          * @param callable $callback
 230          * @return callable
 231          */
 232         public function setSiteInfoCallback( $callback ) {
 233                 $previous = $this->mSiteInfoCallback;
 234                 $this->mSiteInfoCallback = $callback;
 235                 return $previous;
 236         }
 237
 238         /**
 239          * Sets the factory object to use to convert ForeignTitle objects into local
 240          * Title objects
 241          * @param ImportTitleFactory $factory
 242          */
 243         public function setImportTitleFactory( $factory ) {
 244                 $this->importTitleFactory = $factory;
 245         }
 246
 247         /**
 248          * Set a target namespace to override the defaults
 249          * @param null|int $namespace
 250          * @return bool
 251          */
 252         public function setTargetNamespace( $namespace ) {
 253                 if ( is_null( $namespace ) ) {
 254                         // Don't override namespaces
 255                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 256                         return true;
 257                 } elseif (
 258                         $namespace >= 0 &&
 259                         MWNamespace::exists( intval( $namespace ) )
 260                 ) {
 261                         $namespace = intval( $namespace );
 262                         $this->setImportTitleFactory( new NamespaceImportTitleFactory( $namespace ) );
 263                         return true;
 264                 } else {
 265                         return false;
 266                 }
 267         }
 268
 269         /**
 270          * Set a target root page under which all pages are imported
 271          * @param null|string $rootpage
 272          * @return Status
 273          */
 274         public function setTargetRootPage( $rootpage ) {
 275                 $status = Status::newGood();
 276                 if ( is_null( $rootpage ) ) {
 277                         // No rootpage
 278                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 279                 } elseif ( $rootpage !== '' ) {
 280                         $rootpage = rtrim( $rootpage, '/' ); // avoid double slashes
 281                         $title = Title::newFromText( $rootpage );
 282
 283                         if ( !$title || $title->isExternal() ) {
 284                                 $status->fatal( 'import-rootpage-invalid' );
 285                         } else {
 286                                 if ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 287                                         global $wgContLang;
 288
 289                                         $displayNSText = $title->getNamespace() == NS_MAIN
 290                                                 ? wfMessage( 'blanknamespace' )->text()
 291                                                 : $wgContLang->getNsText( $title->getNamespace() );
 292                                         $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 293                                 } else {
 294                                         // set namespace to 'all', so the namespace check in processTitle() can pass
 295                                         $this->setTargetNamespace( null );
 296                                         $this->setImportTitleFactory( new SubpageImportTitleFactory( $title ) );
 297                                 }
 298                         }
 299                 }
 300                 return $status;
 301         }
 302
 303         /**
 304          * @param string $dir
 305          */
 306         public function setImageBasePath( $dir ) {
 307                 $this->mImageBasePath = $dir;
 308         }
 309
 310         /**
 311          * @param bool $import
 312          */
 313         public function setImportUploads( $import ) {
 314                 $this->mImportUploads = $import;
 315         }
 316
 317         /**
 318          * @since 1.31
 319          * @param string $usernamePrefix Prefix to apply to unknown (and possibly also known) usernames
 320          * @param bool $assignKnownUsers Whether to apply the prefix to usernames that exist locally
 321          */
 322         public function setUsernamePrefix( $usernamePrefix, $assignKnownUsers ) {
 323                 $this->usernamePrefix = rtrim( (string)$usernamePrefix, ':>' );
 324                 $this->assignKnownUsers = (bool)$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          * @access 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->usernamePrefix . '>Unknown user' );
 734                 } else {
 735                         $revision->setUsername( $this->prefixUsername( $logInfo['contributor']['username'] ) );
 736                 }
 737
 738                 return $this->logItemCallback( $revision );
 739         }
 740
 741         private function handlePage() {
 742                 // Handle page data.
 743                 $this->debug( "Enter page handler." );
 744                 $pageInfo = [ 'revisionCount' => 0, 'successfulRevisionCount' => 0 ];
 745
 746                 // Fields that can just be stuffed in the pageInfo object
 747                 $normalFields = [ 'title', 'ns', 'id', 'redirect', 'restrictions' ];
 748
 749                 $skip = false;
 750                 $badTitle = false;
 751
 752                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 753                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 754                                         $this->reader->localName == 'page' ) {
 755                                 break;
 756                         }
 757
 758                         $skip = false;
 759
 760                         $tag = $this->reader->localName;
 761
 762                         if ( $badTitle ) {
 763                                 // The title is invalid, bail out of this page
 764                                 $skip = true;
 765                         } elseif ( !Hooks::run( 'ImportHandlePageXMLTag', [ $this,
 766                                                 &$pageInfo ] ) ) {
 767                                 // Do nothing
 768                         } elseif ( in_array( $tag, $normalFields ) ) {
 769                                 // An XML snippet:
 770                                 // <page>
 771                                 //     <id>123</id>
 772                                 //     <title>Page</title>
 773                                 //     <redirect title="NewTitle"/>
 774                                 //     ...
 775                                 // Because the redirect tag is built differently, we need special handling for that case.
 776                                 if ( $tag == 'redirect' ) {
 777                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 778                                 } else {
 779                                         $pageInfo[$tag] = $this->nodeContents();
 780                                 }
 781                         } elseif ( $tag == 'revision' || $tag == 'upload' ) {
 782                                 if ( !isset( $title ) ) {
 783                                         $title = $this->processTitle( $pageInfo['title'],
 784                                                 isset( $pageInfo['ns'] ) ? $pageInfo['ns'] : null );
 785
 786                                         // $title is either an array of two titles or false.
 787                                         if ( is_array( $title ) ) {
 788                                                 $this->pageCallback( $title );
 789                                                 list( $pageInfo['_title'], $foreignTitle ) = $title;
 790                                         } else {
 791                                                 $badTitle = true;
 792                                                 $skip = true;
 793                                         }
 794                                 }
 795
 796                                 if ( $title ) {
 797                                         if ( $tag == 'revision' ) {
 798                                                 $this->handleRevision( $pageInfo );
 799                                         } else {
 800                                                 $this->handleUpload( $pageInfo );
 801                                         }
 802                                 }
 803                         } elseif ( $tag != '#text' ) {
 804                                 $this->warn( "Unhandled page XML tag $tag" );
 805                                 $skip = true;
 806                         }
 807                 }
 808
 809                 // @note $pageInfo is only set if a valid $title is processed above with
 810                 //       no error. If we have a valid $title, then pageCallback is called
 811                 //       above, $pageInfo['title'] is set and we do pageOutCallback here.
 812                 //       If $pageInfo['_title'] is not set, then $foreignTitle is also not
 813                 //       set since they both come from $title above.
 814                 if ( array_key_exists( '_title', $pageInfo ) ) {
 815                         $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
 816                                         $pageInfo['revisionCount'],
 817                                         $pageInfo['successfulRevisionCount'],
 818                                         $pageInfo );
 819                 }
 820         }
 821
 822         /**
 823          * @param array $pageInfo
 824          */
 825         private function handleRevision( &$pageInfo ) {
 826                 $this->debug( "Enter revision handler" );
 827                 $revisionInfo = [];
 828
 829                 $normalFields = [ 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text', 'sha1' ];
 830
 831                 $skip = false;
 832
 833                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 834                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 835                                         $this->reader->localName == 'revision' ) {
 836                                 break;
 837                         }
 838
 839                         $tag = $this->reader->localName;
 840
 841                         if ( !Hooks::run( 'ImportHandleRevisionXMLTag', [
 842                                 $this, $pageInfo, $revisionInfo
 843                         ] ) ) {
 844                                 // Do nothing
 845                         } elseif ( in_array( $tag, $normalFields ) ) {
 846                                 $revisionInfo[$tag] = $this->nodeContents();
 847                         } elseif ( $tag == 'contributor' ) {
 848                                 $revisionInfo['contributor'] = $this->handleContributor();
 849                         } elseif ( $tag != '#text' ) {
 850                                 $this->warn( "Unhandled revision XML tag $tag" );
 851                                 $skip = true;
 852                         }
 853                 }
 854
 855                 $pageInfo['revisionCount']++;
 856                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 857                         $pageInfo['successfulRevisionCount']++;
 858                 }
 859         }
 860
 861         /**
 862          * @param array $pageInfo
 863          * @param array $revisionInfo
 864          * @throws MWException
 865          * @return bool|mixed
 866          */
 867         private function processRevision( $pageInfo, $revisionInfo ) {
 868                 global $wgMaxArticleSize;
 869
 870                 // Make sure revisions won't violate $wgMaxArticleSize, which could lead to
 871                 // database errors and instability. Testing for revisions with only listed
 872                 // content models, as other content models might use serialization formats
 873                 // which aren't checked against $wgMaxArticleSize.
 874                 if ( ( !isset( $revisionInfo['model'] ) ||
 875                         in_array( $revisionInfo['model'], [
 876                                 'wikitext',
 877                                 'css',
 878                                 'json',
 879                                 'javascript',
 880                                 'text',
 881                                 ''
 882                         ] ) ) &&
 883                         strlen( $revisionInfo['text'] ) > $wgMaxArticleSize * 1024
 884                 ) {
 885                         throw new MWException( 'The text of ' .
 886                                 ( isset( $revisionInfo['id'] ) ?
 887                                         "the revision with ID $revisionInfo[id]" :
 888                                         'a revision'
 889                                 ) . " exceeds the maximum allowable size ($wgMaxArticleSize KB)" );
 890                 }
 891
 892                 $revision = new WikiRevision( $this->config );
 893
 894                 if ( isset( $revisionInfo['id'] ) ) {
 895                         $revision->setID( $revisionInfo['id'] );
 896                 }
 897                 if ( isset( $revisionInfo['model'] ) ) {
 898                         $revision->setModel( $revisionInfo['model'] );
 899                 }
 900                 if ( isset( $revisionInfo['format'] ) ) {
 901                         $revision->setFormat( $revisionInfo['format'] );
 902                 }
 903                 $revision->setTitle( $pageInfo['_title'] );
 904
 905                 if ( isset( $revisionInfo['text'] ) ) {
 906                         $handler = $revision->getContentHandler();
 907                         $text = $handler->importTransform(
 908                                 $revisionInfo['text'],
 909                                 $revision->getFormat() );
 910
 911                         $revision->setText( $text );
 912                 }
 913                 if ( isset( $revisionInfo['timestamp'] ) ) {
 914                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 915                 } else {
 916                         $revision->setTimestamp( wfTimestampNow() );
 917                 }
 918
 919                 if ( isset( $revisionInfo['comment'] ) ) {
 920                         $revision->setComment( $revisionInfo['comment'] );
 921                 }
 922
 923                 if ( isset( $revisionInfo['minor'] ) ) {
 924                         $revision->setMinor( true );
 925                 }
 926                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 927                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 928                 } elseif ( isset( $revisionInfo['contributor']['username'] ) ) {
 929                         $revision->setUsername( $this->prefixUsername( $revisionInfo['contributor']['username'] ) );
 930                 } else {
 931                         $revision->setUsername( $this->usernamePrefix . '>Unknown user' );
 932                 }
 933                 if ( isset( $revisionInfo['sha1'] ) ) {
 934                         $revision->setSha1Base36( $revisionInfo['sha1'] );
 935                 }
 936                 $revision->setNoUpdates( $this->mNoUpdates );
 937
 938                 return $this->revisionCallback( $revision );
 939         }
 940
 941         /**
 942          * @param array $pageInfo
 943          * @return mixed
 944          */
 945         private function handleUpload( &$pageInfo ) {
 946                 $this->debug( "Enter upload handler" );
 947                 $uploadInfo = [];
 948
 949                 $normalFields = [ 'timestamp', 'comment', 'filename', 'text',
 950                                         'src', 'size', 'sha1base36', 'archivename', 'rel' ];
 951
 952                 $skip = false;
 953
 954                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 955                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 956                                         $this->reader->localName == 'upload' ) {
 957                                 break;
 958                         }
 959
 960                         $tag = $this->reader->localName;
 961
 962                         if ( !Hooks::run( 'ImportHandleUploadXMLTag', [
 963                                 $this, $pageInfo
 964                         ] ) ) {
 965                                 // Do nothing
 966                         } elseif ( in_array( $tag, $normalFields ) ) {
 967                                 $uploadInfo[$tag] = $this->nodeContents();
 968                         } elseif ( $tag == 'contributor' ) {
 969                                 $uploadInfo['contributor'] = $this->handleContributor();
 970                         } elseif ( $tag == 'contents' ) {
 971                                 $contents = $this->nodeContents();
 972                                 $encoding = $this->reader->getAttribute( 'encoding' );
 973                                 if ( $encoding === 'base64' ) {
 974                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 975                                         $uploadInfo['isTempSrc'] = true;
 976                                 }
 977                         } elseif ( $tag != '#text' ) {
 978                                 $this->warn( "Unhandled upload XML tag $tag" );
 979                                 $skip = true;
 980                         }
 981                 }
 982
 983                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 984                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 985                         if ( file_exists( $path ) ) {
 986                                 $uploadInfo['fileSrc'] = $path;
 987                                 $uploadInfo['isTempSrc'] = false;
 988                         }
 989                 }
 990
 991                 if ( $this->mImportUploads ) {
 992                         return $this->processUpload( $pageInfo, $uploadInfo );
 993                 }
 994         }
 995
 996         /**
 997          * @param string $contents
 998          * @return string
 999          */
1000         private function dumpTemp( $contents ) {
1001                 $filename = tempnam( wfTempDir(), 'importupload' );
1002                 file_put_contents( $filename, $contents );
1003                 return $filename;
1004         }
1005
1006         /**
1007          * @param array $pageInfo
1008          * @param array $uploadInfo
1009          * @return mixed
1010          */
1011         private function processUpload( $pageInfo, $uploadInfo ) {
1012                 $revision = new WikiRevision( $this->config );
1013                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
1014
1015                 $revision->setTitle( $pageInfo['_title'] );
1016                 $revision->setID( $pageInfo['id'] );
1017                 $revision->setTimestamp( $uploadInfo['timestamp'] );
1018                 $revision->setText( $text );
1019                 $revision->setFilename( $uploadInfo['filename'] );
1020                 if ( isset( $uploadInfo['archivename'] ) ) {
1021                         $revision->setArchiveName( $uploadInfo['archivename'] );
1022                 }
1023                 $revision->setSrc( $uploadInfo['src'] );
1024                 if ( isset( $uploadInfo['fileSrc'] ) ) {
1025                         $revision->setFileSrc( $uploadInfo['fileSrc'],
1026                                 !empty( $uploadInfo['isTempSrc'] ) );
1027                 }
1028                 if ( isset( $uploadInfo['sha1base36'] ) ) {
1029                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
1030                 }
1031                 $revision->setSize( intval( $uploadInfo['size'] ) );
1032                 $revision->setComment( $uploadInfo['comment'] );
1033
1034                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
1035                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
1036                 }
1037                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
1038                         $revision->setUsername( $this->prefixUsername( $uploadInfo['contributor']['username'] ) );
1039                 }
1040                 $revision->setNoUpdates( $this->mNoUpdates );
1041
1042                 return call_user_func( $this->mUploadCallback, $revision );
1043         }
1044
1045         /**
1046          * Add an interwiki prefix to the username, if appropriate
1047          * @since 1.31
1048          * @param string $name Name being imported
1049          * @return string Name, possibly with the prefix prepended.
1050          */
1051         protected function prefixUsername( $name ) {
1052                 if ( !User::isUsableName( $name ) ) {
1053                         return $name;
1054                 }
1055
1056                 if ( $this->assignKnownUsers ) {
1057                         if ( User::idFromName( $name ) ) {
1058                                 return $name;
1059                         }
1060
1061                         // See if any extension wants to create it.
1062                         if ( !isset( $this->triedCreations[$name] ) ) {
1063                                 $this->triedCreations[$name] = true;
1064                                 if ( !Hooks::run( 'ImportHandleUnknownUser', [ $name ] ) &&
1065                                         User::idFromName( $name, User::READ_LATEST )
1066                                 ) {
1067                                         return $name;
1068                                 }
1069                         }
1070                 }
1071
1072                 return substr( $this->usernamePrefix . '>' . $name, 0, 255 );
1073         }
1074
1075         /**
1076          * @return array
1077          */
1078         private function handleContributor() {
1079                 $fields = [ 'id', 'ip', 'username' ];
1080                 $info = [];
1081
1082                 if ( $this->reader->isEmptyElement ) {
1083                         return $info;
1084                 }
1085                 while ( $this->reader->read() ) {
1086                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
1087                                         $this->reader->localName == 'contributor' ) {
1088                                 break;
1089                         }
1090
1091                         $tag = $this->reader->localName;
1092
1093                         if ( in_array( $tag, $fields ) ) {
1094                                 $info[$tag] = $this->nodeContents();
1095                         }
1096                 }
1097
1098                 return $info;
1099         }
1100
1101         /**
1102          * @param string $text
1103          * @param string|null $ns
1104          * @return array|bool
1105          */
1106         private function processTitle( $text, $ns = null ) {
1107                 if ( is_null( $this->foreignNamespaces ) ) {
1108                         $foreignTitleFactory = new NaiveForeignTitleFactory();
1109                 } else {
1110                         $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
1111                                 $this->foreignNamespaces );
1112                 }
1113
1114                 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
1115                         intval( $ns ) );
1116
1117                 $title = $this->importTitleFactory->createTitleFromForeignTitle(
1118                         $foreignTitle );
1119
1120                 $commandLineMode = $this->config->get( 'CommandLineMode' );
1121                 if ( is_null( $title ) ) {
1122                         # Invalid page title? Ignore the page
1123                         $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
1124                         return false;
1125                 } elseif ( $title->isExternal() ) {
1126                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
1127                         return false;
1128                 } elseif ( !$title->canExist() ) {
1129                         $this->notice( 'import-error-special', $title->getPrefixedText() );
1130                         return false;
1131                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
1132                         # Do not import if the importing wiki user cannot edit this page
1133                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
1134                         return false;
1135                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
1136                         # Do not import if the importing wiki user cannot create this page
1137                         $this->notice( 'import-error-create', $title->getPrefixedText() );
1138                         return false;
1139                 }
1140
1141                 return [ $title, $foreignTitle ];
1142         }
1143 }