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 ) {
  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                 $this->config = $config;
  65
  66                 if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) {
  67                         stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  68                 }
  69                 $id = UploadSourceAdapter::registerSource( $source );
  70
  71                 // Enable the entity loader, as it is needed for loading external URLs via
  72                 // XMLReader::open (T86036)
  73                 $oldDisable = libxml_disable_entity_loader( false );
  74                 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
  75                         $status = $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  76                 } else {
  77                         $status = $this->reader->open( "uploadsource://$id" );
  78                 }
  79                 if ( !$status ) {
  80                         $error = libxml_get_last_error();
  81                         libxml_disable_entity_loader( $oldDisable );
  82                         throw new MWException( 'Encountered an internal error while initializing WikiImporter object: ' .
  83                                 $error->message );
  84                 }
  85                 libxml_disable_entity_loader( $oldDisable );
  86
  87                 // Default callbacks
  88                 $this->setPageCallback( [ $this, 'beforeImportPage' ] );
  89                 $this->setRevisionCallback( [ $this, "importRevision" ] );
  90                 $this->setUploadCallback( [ $this, 'importUpload' ] );
  91                 $this->setLogItemCallback( [ $this, 'importLogItem' ] );
  92                 $this->setPageOutCallback( [ $this, 'finishImportPage' ] );
  93
  94                 $this->importTitleFactory = new NaiveImportTitleFactory();
  95         }
  96
  97         /**
  98          * @return null|XMLReader
  99          */
 100         public function getReader() {
 101                 return $this->reader;
 102         }
 103
 104         public function throwXmlError( $err ) {
 105                 $this->debug( "FAILURE: $err" );
 106                 wfDebug( "WikiImporter XML error: $err\n" );
 107         }
 108
 109         public function debug( $data ) {
 110                 if ( $this->mDebug ) {
 111                         wfDebug( "IMPORT: $data\n" );
 112                 }
 113         }
 114
 115         public function warn( $data ) {
 116                 wfDebug( "IMPORT: $data\n" );
 117         }
 118
 119         public function notice( $msg /*, $param, ...*/ ) {
 120                 $params = func_get_args();
 121                 array_shift( $params );
 122
 123                 if ( is_callable( $this->mNoticeCallback ) ) {
 124                         call_user_func( $this->mNoticeCallback, $msg, $params );
 125                 } else { # No ImportReporter -> CLI
 126                         echo wfMessage( $msg, $params )->text() . "\n";
 127                 }
 128         }
 129
 130         /**
 131          * Set debug mode...
 132          * @param bool $debug
 133          */
 134         function setDebug( $debug ) {
 135                 $this->mDebug = $debug;
 136         }
 137
 138         /**
 139          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
 140          * @param bool $noupdates
 141          */
 142         function setNoUpdates( $noupdates ) {
 143                 $this->mNoUpdates = $noupdates;
 144         }
 145
 146         /**
 147          * Sets 'pageOffset' value. So it will skip the first n-1 pages
 148          * and start from the nth page. It's 1-based indexing.
 149          * @param int $nthPage
 150          * @since 1.29
 151          */
 152         function setPageOffset( $nthPage ) {
 153                 $this->pageOffset = $nthPage;
 154         }
 155
 156         /**
 157          * Set a callback that displays notice messages
 158          *
 159          * @param callable $callback
 160          * @return callable
 161          */
 162         public function setNoticeCallback( $callback ) {
 163                 return wfSetVar( $this->mNoticeCallback, $callback );
 164         }
 165
 166         /**
 167          * Sets the action to perform as each new page in the stream is reached.
 168          * @param callable $callback
 169          * @return callable
 170          */
 171         public function setPageCallback( $callback ) {
 172                 $previous = $this->mPageCallback;
 173                 $this->mPageCallback = $callback;
 174                 return $previous;
 175         }
 176
 177         /**
 178          * Sets the action to perform as each page in the stream is completed.
 179          * Callback accepts the page title (as a Title object), a second object
 180          * with the original title form (in case it's been overridden into a
 181          * local namespace), and a count of revisions.
 182          *
 183          * @param callable $callback
 184          * @return callable
 185          */
 186         public function setPageOutCallback( $callback ) {
 187                 $previous = $this->mPageOutCallback;
 188                 $this->mPageOutCallback = $callback;
 189                 return $previous;
 190         }
 191
 192         /**
 193          * Sets the action to perform as each page revision is reached.
 194          * @param callable $callback
 195          * @return callable
 196          */
 197         public function setRevisionCallback( $callback ) {
 198                 $previous = $this->mRevisionCallback;
 199                 $this->mRevisionCallback = $callback;
 200                 return $previous;
 201         }
 202
 203         /**
 204          * Sets the action to perform as each file upload version is reached.
 205          * @param callable $callback
 206          * @return callable
 207          */
 208         public function setUploadCallback( $callback ) {
 209                 $previous = $this->mUploadCallback;
 210                 $this->mUploadCallback = $callback;
 211                 return $previous;
 212         }
 213
 214         /**
 215          * Sets the action to perform as each log item reached.
 216          * @param callable $callback
 217          * @return callable
 218          */
 219         public function setLogItemCallback( $callback ) {
 220                 $previous = $this->mLogItemCallback;
 221                 $this->mLogItemCallback = $callback;
 222                 return $previous;
 223         }
 224
 225         /**
 226          * Sets the action to perform when site info is encountered
 227          * @param callable $callback
 228          * @return callable
 229          */
 230         public function setSiteInfoCallback( $callback ) {
 231                 $previous = $this->mSiteInfoCallback;
 232                 $this->mSiteInfoCallback = $callback;
 233                 return $previous;
 234         }
 235
 236         /**
 237          * Sets the factory object to use to convert ForeignTitle objects into local
 238          * Title objects
 239          * @param ImportTitleFactory $factory
 240          */
 241         public function setImportTitleFactory( $factory ) {
 242                 $this->importTitleFactory = $factory;
 243         }
 244
 245         /**
 246          * Set a target namespace to override the defaults
 247          * @param null|int $namespace
 248          * @return bool
 249          */
 250         public function setTargetNamespace( $namespace ) {
 251                 if ( is_null( $namespace ) ) {
 252                         // Don't override namespaces
 253                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 254                         return true;
 255                 } elseif (
 256                         $namespace >= 0 &&
 257                         MWNamespace::exists( intval( $namespace ) )
 258                 ) {
 259                         $namespace = intval( $namespace );
 260                         $this->setImportTitleFactory( new NamespaceImportTitleFactory( $namespace ) );
 261                         return true;
 262                 } else {
 263                         return false;
 264                 }
 265         }
 266
 267         /**
 268          * Set a target root page under which all pages are imported
 269          * @param null|string $rootpage
 270          * @return Status
 271          */
 272         public function setTargetRootPage( $rootpage ) {
 273                 $status = Status::newGood();
 274                 if ( is_null( $rootpage ) ) {
 275                         // No rootpage
 276                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 277                 } elseif ( $rootpage !== '' ) {
 278                         $rootpage = rtrim( $rootpage, '/' ); // avoid double slashes
 279                         $title = Title::newFromText( $rootpage );
 280
 281                         if ( !$title || $title->isExternal() ) {
 282                                 $status->fatal( 'import-rootpage-invalid' );
 283                         } else {
 284                                 if ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 285                                         global $wgContLang;
 286
 287                                         $displayNSText = $title->getNamespace() == NS_MAIN
 288                                                 ? wfMessage( 'blanknamespace' )->text()
 289                                                 : $wgContLang->getNsText( $title->getNamespace() );
 290                                         $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 291                                 } else {
 292                                         // set namespace to 'all', so the namespace check in processTitle() can pass
 293                                         $this->setTargetNamespace( null );
 294                                         $this->setImportTitleFactory( new SubpageImportTitleFactory( $title ) );
 295                                 }
 296                         }
 297                 }
 298                 return $status;
 299         }
 300
 301         /**
 302          * @param string $dir
 303          */
 304         public function setImageBasePath( $dir ) {
 305                 $this->mImageBasePath = $dir;
 306         }
 307
 308         /**
 309          * @param bool $import
 310          */
 311         public function setImportUploads( $import ) {
 312                 $this->mImportUploads = $import;
 313         }
 314
 315         /**
 316          * Statistics update can cause a lot of time
 317          * @since 1.29
 318          */
 319         public function disableStatisticsUpdate() {
 320                 $this->disableStatisticsUpdate = true;
 321         }
 322
 323         /**
 324          * Default per-page callback. Sets up some things related to site statistics
 325          * @param array $titleAndForeignTitle Two-element array, with Title object at
 326          * index 0 and ForeignTitle object at index 1
 327          * @return bool
 328          */
 329         public function beforeImportPage( $titleAndForeignTitle ) {
 330                 $title = $titleAndForeignTitle[0];
 331                 $page = WikiPage::factory( $title );
 332                 $this->countableCache['title_' . $title->getPrefixedText()] = $page->isCountable();
 333                 return true;
 334         }
 335
 336         /**
 337          * Default per-revision callback, performs the import.
 338          * @param WikiRevision $revision
 339          * @return bool
 340          */
 341         public function importRevision( $revision ) {
 342                 if ( !$revision->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
 343                         $this->notice( 'import-error-bad-location',
 344                                 $revision->getTitle()->getPrefixedText(),
 345                                 $revision->getID(),
 346                                 $revision->getModel(),
 347                                 $revision->getFormat() );
 348
 349                         return false;
 350                 }
 351
 352                 try {
 353                         return $revision->importOldRevision();
 354                 } catch ( MWContentSerializationException $ex ) {
 355                         $this->notice( 'import-error-unserialize',
 356                                 $revision->getTitle()->getPrefixedText(),
 357                                 $revision->getID(),
 358                                 $revision->getModel(),
 359                                 $revision->getFormat() );
 360                 }
 361
 362                 return false;
 363         }
 364
 365         /**
 366          * Default per-revision callback, performs the import.
 367          * @param WikiRevision $revision
 368          * @return bool
 369          */
 370         public function importLogItem( $revision ) {
 371                 return $revision->importLogItem();
 372         }
 373
 374         /**
 375          * Dummy for now...
 376          * @param WikiRevision $revision
 377          * @return bool
 378          */
 379         public function importUpload( $revision ) {
 380                 return $revision->importUpload();
 381         }
 382
 383         /**
 384          * Mostly for hook use
 385          * @param Title $title
 386          * @param ForeignTitle $foreignTitle
 387          * @param int $revCount
 388          * @param int $sRevCount
 389          * @param array $pageInfo
 390          * @return bool
 391          */
 392         public function finishImportPage( $title, $foreignTitle, $revCount,
 393                 $sRevCount, $pageInfo
 394         ) {
 395                 // Update article count statistics (T42009)
 396                 // The normal counting logic in WikiPage->doEditUpdates() is designed for
 397                 // one-revision-at-a-time editing, not bulk imports. In this situation it
 398                 // suffers from issues of replica DB lag. We let WikiPage handle the total page
 399                 // and revision count, and we implement our own custom logic for the
 400                 // article (content page) count.
 401                 if ( !$this->disableStatisticsUpdate ) {
 402                         $page = WikiPage::factory( $title );
 403                         $page->loadPageData( 'fromdbmaster' );
 404                         $content = $page->getContent();
 405                         if ( $content === null ) {
 406                                 wfDebug( __METHOD__ . ': Skipping article count adjustment for ' . $title .
 407                                         ' because WikiPage::getContent() returned null' );
 408                         } else {
 409                                 $editInfo = $page->prepareContentForEdit( $content );
 410                                 $countKey = 'title_' . $title->getPrefixedText();
 411                                 $countable = $page->isCountable( $editInfo );
 412                                 if ( array_key_exists( $countKey, $this->countableCache ) &&
 413                                         $countable != $this->countableCache[$countKey] ) {
 414                                         DeferredUpdates::addUpdate( SiteStatsUpdate::factory( [
 415                                                 'articles' => ( (int)$countable - (int)$this->countableCache[$countKey] )
 416                                         ] ) );
 417                                 }
 418                         }
 419                 }
 420
 421                 $args = func_get_args();
 422                 return Hooks::run( 'AfterImportPage', $args );
 423         }
 424
 425         /**
 426          * Alternate per-revision callback, for debugging.
 427          * @param WikiRevision &$revision
 428          */
 429         public function debugRevisionHandler( &$revision ) {
 430                 $this->debug( "Got revision:" );
 431                 if ( is_object( $revision->title ) ) {
 432                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 433                 } else {
 434                         $this->debug( "-- Title: <invalid>" );
 435                 }
 436                 $this->debug( "-- User: " . $revision->user_text );
 437                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 438                 $this->debug( "-- Comment: " . $revision->comment );
 439                 $this->debug( "-- Text: " . $revision->text );
 440         }
 441
 442         /**
 443          * Notify the callback function of site info
 444          * @param array $siteInfo
 445          * @return bool|mixed
 446          */
 447         private function siteInfoCallback( $siteInfo ) {
 448                 if ( isset( $this->mSiteInfoCallback ) ) {
 449                         return call_user_func_array( $this->mSiteInfoCallback,
 450                                         [ $siteInfo, $this ] );
 451                 } else {
 452                         return false;
 453                 }
 454         }
 455
 456         /**
 457          * Notify the callback function when a new "<page>" is reached.
 458          * @param Title $title
 459          */
 460         function pageCallback( $title ) {
 461                 if ( isset( $this->mPageCallback ) ) {
 462                         call_user_func( $this->mPageCallback, $title );
 463                 }
 464         }
 465
 466         /**
 467          * Notify the callback function when a "</page>" is closed.
 468          * @param Title $title
 469          * @param ForeignTitle $foreignTitle
 470          * @param int $revCount
 471          * @param int $sucCount Number of revisions for which callback returned true
 472          * @param array $pageInfo Associative array of page information
 473          */
 474         private function pageOutCallback( $title, $foreignTitle, $revCount,
 475                         $sucCount, $pageInfo ) {
 476                 if ( isset( $this->mPageOutCallback ) ) {
 477                         $args = func_get_args();
 478                         call_user_func_array( $this->mPageOutCallback, $args );
 479                 }
 480         }
 481
 482         /**
 483          * Notify the callback function of a revision
 484          * @param WikiRevision $revision
 485          * @return bool|mixed
 486          */
 487         private function revisionCallback( $revision ) {
 488                 if ( isset( $this->mRevisionCallback ) ) {
 489                         return call_user_func_array( $this->mRevisionCallback,
 490                                         [ $revision, $this ] );
 491                 } else {
 492                         return false;
 493                 }
 494         }
 495
 496         /**
 497          * Notify the callback function of a new log item
 498          * @param WikiRevision $revision
 499          * @return bool|mixed
 500          */
 501         private function logItemCallback( $revision ) {
 502                 if ( isset( $this->mLogItemCallback ) ) {
 503                         return call_user_func_array( $this->mLogItemCallback,
 504                                         [ $revision, $this ] );
 505                 } else {
 506                         return false;
 507                 }
 508         }
 509
 510         /**
 511          * Retrieves the contents of the named attribute of the current element.
 512          * @param string $attr The name of the attribute
 513          * @return string The value of the attribute or an empty string if it is not set in the current
 514          * element.
 515          */
 516         public function nodeAttribute( $attr ) {
 517                 return $this->reader->getAttribute( $attr );
 518         }
 519
 520         /**
 521          * Shouldn't something like this be built-in to XMLReader?
 522          * Fetches text contents of the current element, assuming
 523          * no sub-elements or such scary things.
 524          * @return string
 525          * @access private
 526          */
 527         public function nodeContents() {
 528                 if ( $this->reader->isEmptyElement ) {
 529                         return "";
 530                 }
 531                 $buffer = "";
 532                 while ( $this->reader->read() ) {
 533                         switch ( $this->reader->nodeType ) {
 534                         case XMLReader::TEXT:
 535                         case XMLReader::CDATA:
 536                         case XMLReader::SIGNIFICANT_WHITESPACE:
 537                                 $buffer .= $this->reader->value;
 538                                 break;
 539                         case XMLReader::END_ELEMENT:
 540                                 return $buffer;
 541                         }
 542                 }
 543
 544                 $this->reader->close();
 545                 return '';
 546         }
 547
 548         /**
 549          * Primary entry point
 550          * @throws MWException
 551          * @return bool
 552          */
 553         public function doImport() {
 554                 // Calls to reader->read need to be wrapped in calls to
 555                 // libxml_disable_entity_loader() to avoid local file
 556                 // inclusion attacks (T48932).
 557                 $oldDisable = libxml_disable_entity_loader( true );
 558                 $this->reader->read();
 559
 560                 if ( $this->reader->localName != 'mediawiki' ) {
 561                         libxml_disable_entity_loader( $oldDisable );
 562                         throw new MWException( "Expected <mediawiki> tag, got " .
 563                                 $this->reader->localName );
 564                 }
 565                 $this->debug( "<mediawiki> tag is correct." );
 566
 567                 $this->debug( "Starting primary dump processing loop." );
 568
 569                 $keepReading = $this->reader->read();
 570                 $skip = false;
 571                 $rethrow = null;
 572                 $pageCount = 0;
 573                 try {
 574                         while ( $keepReading ) {
 575                                 $tag = $this->reader->localName;
 576                                 if ( $this->pageOffset ) {
 577                                         if ( $tag === 'page' ) {
 578                                                 $pageCount++;
 579                                         }
 580                                         if ( $pageCount < $this->pageOffset ) {
 581                                                 $keepReading = $this->reader->next();
 582                                                 continue;
 583                                         }
 584                                 }
 585                                 $type = $this->reader->nodeType;
 586
 587                                 if ( !Hooks::run( 'ImportHandleToplevelXMLTag', [ $this ] ) ) {
 588                                         // Do nothing
 589                                 } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) {
 590                                         break;
 591                                 } elseif ( $tag == 'siteinfo' ) {
 592                                         $this->handleSiteInfo();
 593                                 } elseif ( $tag == 'page' ) {
 594                                         $this->handlePage();
 595                                 } elseif ( $tag == 'logitem' ) {
 596                                         $this->handleLogItem();
 597                                 } elseif ( $tag != '#text' ) {
 598                                         $this->warn( "Unhandled top-level XML tag $tag" );
 599
 600                                         $skip = true;
 601                                 }
 602
 603                                 if ( $skip ) {
 604                                         $keepReading = $this->reader->next();
 605                                         $skip = false;
 606                                         $this->debug( "Skip" );
 607                                 } else {
 608                                         $keepReading = $this->reader->read();
 609                                 }
 610                         }
 611                 } catch ( Exception $ex ) {
 612                         $rethrow = $ex;
 613                 }
 614
 615                 // finally
 616                 libxml_disable_entity_loader( $oldDisable );
 617                 $this->reader->close();
 618
 619                 if ( $rethrow ) {
 620                         throw $rethrow;
 621                 }
 622
 623                 return true;
 624         }
 625
 626         private function handleSiteInfo() {
 627                 $this->debug( "Enter site info handler." );
 628                 $siteInfo = [];
 629
 630                 // Fields that can just be stuffed in the siteInfo object
 631                 $normalFields = [ 'sitename', 'base', 'generator', 'case' ];
 632
 633                 while ( $this->reader->read() ) {
 634                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 635                                         $this->reader->localName == 'siteinfo' ) {
 636                                 break;
 637                         }
 638
 639                         $tag = $this->reader->localName;
 640
 641                         if ( $tag == 'namespace' ) {
 642                                 $this->foreignNamespaces[$this->nodeAttribute( 'key' )] =
 643                                         $this->nodeContents();
 644                         } elseif ( in_array( $tag, $normalFields ) ) {
 645                                 $siteInfo[$tag] = $this->nodeContents();
 646                         }
 647                 }
 648
 649                 $siteInfo['_namespaces'] = $this->foreignNamespaces;
 650                 $this->siteInfoCallback( $siteInfo );
 651         }
 652
 653         private function handleLogItem() {
 654                 $this->debug( "Enter log item handler." );
 655                 $logInfo = [];
 656
 657                 // Fields that can just be stuffed in the pageInfo object
 658                 $normalFields = [ 'id', 'comment', 'type', 'action', 'timestamp',
 659                                         'logtitle', 'params' ];
 660
 661                 while ( $this->reader->read() ) {
 662                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 663                                         $this->reader->localName == 'logitem' ) {
 664                                 break;
 665                         }
 666
 667                         $tag = $this->reader->localName;
 668
 669                         if ( !Hooks::run( 'ImportHandleLogItemXMLTag', [
 670                                 $this, $logInfo
 671                         ] ) ) {
 672                                 // Do nothing
 673                         } elseif ( in_array( $tag, $normalFields ) ) {
 674                                 $logInfo[$tag] = $this->nodeContents();
 675                         } elseif ( $tag == 'contributor' ) {
 676                                 $logInfo['contributor'] = $this->handleContributor();
 677                         } elseif ( $tag != '#text' ) {
 678                                 $this->warn( "Unhandled log-item XML tag $tag" );
 679                         }
 680                 }
 681
 682                 $this->processLogItem( $logInfo );
 683         }
 684
 685         /**
 686          * @param array $logInfo
 687          * @return bool|mixed
 688          */
 689         private function processLogItem( $logInfo ) {
 690                 $revision = new WikiRevision( $this->config );
 691
 692                 if ( isset( $logInfo['id'] ) ) {
 693                         $revision->setID( $logInfo['id'] );
 694                 }
 695                 $revision->setType( $logInfo['type'] );
 696                 $revision->setAction( $logInfo['action'] );
 697                 if ( isset( $logInfo['timestamp'] ) ) {
 698                         $revision->setTimestamp( $logInfo['timestamp'] );
 699                 }
 700                 if ( isset( $logInfo['params'] ) ) {
 701                         $revision->setParams( $logInfo['params'] );
 702                 }
 703                 if ( isset( $logInfo['logtitle'] ) ) {
 704                         // @todo Using Title for non-local titles is a recipe for disaster.
 705                         // We should use ForeignTitle here instead.
 706                         $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 707                 }
 708
 709                 $revision->setNoUpdates( $this->mNoUpdates );
 710
 711                 if ( isset( $logInfo['comment'] ) ) {
 712                         $revision->setComment( $logInfo['comment'] );
 713                 }
 714
 715                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 716                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 717                 }
 718
 719                 if ( !isset( $logInfo['contributor']['username'] ) ) {
 720                         $revision->setUsername( 'Unknown user' );
 721                 } else {
 722                         $revision->setUsername( $logInfo['contributor']['username'] );
 723                 }
 724
 725                 return $this->logItemCallback( $revision );
 726         }
 727
 728         private function handlePage() {
 729                 // Handle page data.
 730                 $this->debug( "Enter page handler." );
 731                 $pageInfo = [ 'revisionCount' => 0, 'successfulRevisionCount' => 0 ];
 732
 733                 // Fields that can just be stuffed in the pageInfo object
 734                 $normalFields = [ 'title', 'ns', 'id', 'redirect', 'restrictions' ];
 735
 736                 $skip = false;
 737                 $badTitle = false;
 738
 739                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 740                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 741                                         $this->reader->localName == 'page' ) {
 742                                 break;
 743                         }
 744
 745                         $skip = false;
 746
 747                         $tag = $this->reader->localName;
 748
 749                         if ( $badTitle ) {
 750                                 // The title is invalid, bail out of this page
 751                                 $skip = true;
 752                         } elseif ( !Hooks::run( 'ImportHandlePageXMLTag', [ $this,
 753                                                 &$pageInfo ] ) ) {
 754                                 // Do nothing
 755                         } elseif ( in_array( $tag, $normalFields ) ) {
 756                                 // An XML snippet:
 757                                 // <page>
 758                                 //     <id>123</id>
 759                                 //     <title>Page</title>
 760                                 //     <redirect title="NewTitle"/>
 761                                 //     ...
 762                                 // Because the redirect tag is built differently, we need special handling for that case.
 763                                 if ( $tag == 'redirect' ) {
 764                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 765                                 } else {
 766                                         $pageInfo[$tag] = $this->nodeContents();
 767                                 }
 768                         } elseif ( $tag == 'revision' || $tag == 'upload' ) {
 769                                 if ( !isset( $title ) ) {
 770                                         $title = $this->processTitle( $pageInfo['title'],
 771                                                 isset( $pageInfo['ns'] ) ? $pageInfo['ns'] : null );
 772
 773                                         // $title is either an array of two titles or false.
 774                                         if ( is_array( $title ) ) {
 775                                                 $this->pageCallback( $title );
 776                                                 list( $pageInfo['_title'], $foreignTitle ) = $title;
 777                                         } else {
 778                                                 $badTitle = true;
 779                                                 $skip = true;
 780                                         }
 781                                 }
 782
 783                                 if ( $title ) {
 784                                         if ( $tag == 'revision' ) {
 785                                                 $this->handleRevision( $pageInfo );
 786                                         } else {
 787                                                 $this->handleUpload( $pageInfo );
 788                                         }
 789                                 }
 790                         } elseif ( $tag != '#text' ) {
 791                                 $this->warn( "Unhandled page XML tag $tag" );
 792                                 $skip = true;
 793                         }
 794                 }
 795
 796                 // @note $pageInfo is only set if a valid $title is processed above with
 797                 //       no error. If we have a valid $title, then pageCallback is called
 798                 //       above, $pageInfo['title'] is set and we do pageOutCallback here.
 799                 //       If $pageInfo['_title'] is not set, then $foreignTitle is also not
 800                 //       set since they both come from $title above.
 801                 if ( array_key_exists( '_title', $pageInfo ) ) {
 802                         $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
 803                                         $pageInfo['revisionCount'],
 804                                         $pageInfo['successfulRevisionCount'],
 805                                         $pageInfo );
 806                 }
 807         }
 808
 809         /**
 810          * @param array $pageInfo
 811          */
 812         private function handleRevision( &$pageInfo ) {
 813                 $this->debug( "Enter revision handler" );
 814                 $revisionInfo = [];
 815
 816                 $normalFields = [ 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text', 'sha1' ];
 817
 818                 $skip = false;
 819
 820                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 821                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 822                                         $this->reader->localName == 'revision' ) {
 823                                 break;
 824                         }
 825
 826                         $tag = $this->reader->localName;
 827
 828                         if ( !Hooks::run( 'ImportHandleRevisionXMLTag', [
 829                                 $this, $pageInfo, $revisionInfo
 830                         ] ) ) {
 831                                 // Do nothing
 832                         } elseif ( in_array( $tag, $normalFields ) ) {
 833                                 $revisionInfo[$tag] = $this->nodeContents();
 834                         } elseif ( $tag == 'contributor' ) {
 835                                 $revisionInfo['contributor'] = $this->handleContributor();
 836                         } elseif ( $tag != '#text' ) {
 837                                 $this->warn( "Unhandled revision XML tag $tag" );
 838                                 $skip = true;
 839                         }
 840                 }
 841
 842                 $pageInfo['revisionCount']++;
 843                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 844                         $pageInfo['successfulRevisionCount']++;
 845                 }
 846         }
 847
 848         /**
 849          * @param array $pageInfo
 850          * @param array $revisionInfo
 851          * @return bool|mixed
 852          */
 853         private function processRevision( $pageInfo, $revisionInfo ) {
 854                 global $wgMaxArticleSize;
 855
 856                 // Make sure revisions won't violate $wgMaxArticleSize, which could lead to
 857                 // database errors and instability. Testing for revisions with only listed
 858                 // content models, as other content models might use serialization formats
 859                 // which aren't checked against $wgMaxArticleSize.
 860                 if ( ( !isset( $revisionInfo['model'] ) ||
 861                         in_array( $revisionInfo['model'], [
 862                                 'wikitext',
 863                                 'css',
 864                                 'json',
 865                                 'javascript',
 866                                 'text',
 867                                 ''
 868                         ] ) ) &&
 869                         strlen( $revisionInfo['text'] ) > $wgMaxArticleSize * 1024
 870                 ) {
 871                         throw new MWException( 'The text of ' .
 872                                 ( isset( $revisionInfo['id'] ) ?
 873                                         "the revision with ID $revisionInfo[id]" :
 874                                         'a revision'
 875                                 ) . " exceeds the maximum allowable size ($wgMaxArticleSize KB)" );
 876                 }
 877
 878                 $revision = new WikiRevision( $this->config );
 879
 880                 if ( isset( $revisionInfo['id'] ) ) {
 881                         $revision->setID( $revisionInfo['id'] );
 882                 }
 883                 if ( isset( $revisionInfo['model'] ) ) {
 884                         $revision->setModel( $revisionInfo['model'] );
 885                 }
 886                 if ( isset( $revisionInfo['format'] ) ) {
 887                         $revision->setFormat( $revisionInfo['format'] );
 888                 }
 889                 $revision->setTitle( $pageInfo['_title'] );
 890
 891                 if ( isset( $revisionInfo['text'] ) ) {
 892                         $handler = $revision->getContentHandler();
 893                         $text = $handler->importTransform(
 894                                 $revisionInfo['text'],
 895                                 $revision->getFormat() );
 896
 897                         $revision->setText( $text );
 898                 }
 899                 if ( isset( $revisionInfo['timestamp'] ) ) {
 900                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 901                 } else {
 902                         $revision->setTimestamp( wfTimestampNow() );
 903                 }
 904
 905                 if ( isset( $revisionInfo['comment'] ) ) {
 906                         $revision->setComment( $revisionInfo['comment'] );
 907                 }
 908
 909                 if ( isset( $revisionInfo['minor'] ) ) {
 910                         $revision->setMinor( true );
 911                 }
 912                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 913                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 914                 } elseif ( isset( $revisionInfo['contributor']['username'] ) ) {
 915                         $revision->setUsername( $revisionInfo['contributor']['username'] );
 916                 } else {
 917                         $revision->setUsername( 'Unknown user' );
 918                 }
 919                 if ( isset( $revisionInfo['sha1'] ) ) {
 920                         $revision->setSha1Base36( $revisionInfo['sha1'] );
 921                 }
 922                 $revision->setNoUpdates( $this->mNoUpdates );
 923
 924                 return $this->revisionCallback( $revision );
 925         }
 926
 927         /**
 928          * @param array $pageInfo
 929          * @return mixed
 930          */
 931         private function handleUpload( &$pageInfo ) {
 932                 $this->debug( "Enter upload handler" );
 933                 $uploadInfo = [];
 934
 935                 $normalFields = [ 'timestamp', 'comment', 'filename', 'text',
 936                                         'src', 'size', 'sha1base36', 'archivename', 'rel' ];
 937
 938                 $skip = false;
 939
 940                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 941                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 942                                         $this->reader->localName == 'upload' ) {
 943                                 break;
 944                         }
 945
 946                         $tag = $this->reader->localName;
 947
 948                         if ( !Hooks::run( 'ImportHandleUploadXMLTag', [
 949                                 $this, $pageInfo
 950                         ] ) ) {
 951                                 // Do nothing
 952                         } elseif ( in_array( $tag, $normalFields ) ) {
 953                                 $uploadInfo[$tag] = $this->nodeContents();
 954                         } elseif ( $tag == 'contributor' ) {
 955                                 $uploadInfo['contributor'] = $this->handleContributor();
 956                         } elseif ( $tag == 'contents' ) {
 957                                 $contents = $this->nodeContents();
 958                                 $encoding = $this->reader->getAttribute( 'encoding' );
 959                                 if ( $encoding === 'base64' ) {
 960                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 961                                         $uploadInfo['isTempSrc'] = true;
 962                                 }
 963                         } elseif ( $tag != '#text' ) {
 964                                 $this->warn( "Unhandled upload XML tag $tag" );
 965                                 $skip = true;
 966                         }
 967                 }
 968
 969                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 970                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 971                         if ( file_exists( $path ) ) {
 972                                 $uploadInfo['fileSrc'] = $path;
 973                                 $uploadInfo['isTempSrc'] = false;
 974                         }
 975                 }
 976
 977                 if ( $this->mImportUploads ) {
 978                         return $this->processUpload( $pageInfo, $uploadInfo );
 979                 }
 980         }
 981
 982         /**
 983          * @param string $contents
 984          * @return string
 985          */
 986         private function dumpTemp( $contents ) {
 987                 $filename = tempnam( wfTempDir(), 'importupload' );
 988                 file_put_contents( $filename, $contents );
 989                 return $filename;
 990         }
 991
 992         /**
 993          * @param array $pageInfo
 994          * @param array $uploadInfo
 995          * @return mixed
 996          */
 997         private function processUpload( $pageInfo, $uploadInfo ) {
 998                 $revision = new WikiRevision( $this->config );
 999                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
1000
1001                 $revision->setTitle( $pageInfo['_title'] );
1002                 $revision->setID( $pageInfo['id'] );
1003                 $revision->setTimestamp( $uploadInfo['timestamp'] );
1004                 $revision->setText( $text );
1005                 $revision->setFilename( $uploadInfo['filename'] );
1006                 if ( isset( $uploadInfo['archivename'] ) ) {
1007                         $revision->setArchiveName( $uploadInfo['archivename'] );
1008                 }
1009                 $revision->setSrc( $uploadInfo['src'] );
1010                 if ( isset( $uploadInfo['fileSrc'] ) ) {
1011                         $revision->setFileSrc( $uploadInfo['fileSrc'],
1012                                 !empty( $uploadInfo['isTempSrc'] ) );
1013                 }
1014                 if ( isset( $uploadInfo['sha1base36'] ) ) {
1015                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
1016                 }
1017                 $revision->setSize( intval( $uploadInfo['size'] ) );
1018                 $revision->setComment( $uploadInfo['comment'] );
1019
1020                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
1021                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
1022                 }
1023                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
1024                         $revision->setUsername( $uploadInfo['contributor']['username'] );
1025                 }
1026                 $revision->setNoUpdates( $this->mNoUpdates );
1027
1028                 return call_user_func( $this->mUploadCallback, $revision );
1029         }
1030
1031         /**
1032          * @return array
1033          */
1034         private function handleContributor() {
1035                 $fields = [ 'id', 'ip', 'username' ];
1036                 $info = [];
1037
1038                 if ( $this->reader->isEmptyElement ) {
1039                         return $info;
1040                 }
1041                 while ( $this->reader->read() ) {
1042                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
1043                                         $this->reader->localName == 'contributor' ) {
1044                                 break;
1045                         }
1046
1047                         $tag = $this->reader->localName;
1048
1049                         if ( in_array( $tag, $fields ) ) {
1050                                 $info[$tag] = $this->nodeContents();
1051                         }
1052                 }
1053
1054                 return $info;
1055         }
1056
1057         /**
1058          * @param string $text
1059          * @param string|null $ns
1060          * @return array|bool
1061          */
1062         private function processTitle( $text, $ns = null ) {
1063                 if ( is_null( $this->foreignNamespaces ) ) {
1064                         $foreignTitleFactory = new NaiveForeignTitleFactory();
1065                 } else {
1066                         $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
1067                                 $this->foreignNamespaces );
1068                 }
1069
1070                 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
1071                         intval( $ns ) );
1072
1073                 $title = $this->importTitleFactory->createTitleFromForeignTitle(
1074                         $foreignTitle );
1075
1076                 $commandLineMode = $this->config->get( 'CommandLineMode' );
1077                 if ( is_null( $title ) ) {
1078                         # Invalid page title? Ignore the page
1079                         $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
1080                         return false;
1081                 } elseif ( $title->isExternal() ) {
1082                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
1083                         return false;
1084                 } elseif ( !$title->canExist() ) {
1085                         $this->notice( 'import-error-special', $title->getPrefixedText() );
1086                         return false;
1087                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
1088                         # Do not import if the importing wiki user cannot edit this page
1089                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
1090                         return false;
1091                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
1092                         # Do not import if the importing wiki user cannot create this page
1093                         $this->notice( 'import-error-create', $title->getPrefixedText() );
1094                         return false;
1095                 }
1096
1097                 return [ $title, $foreignTitle ];
1098         }
1099 }