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                         // T177997: the command line importers should call setNoticeCallback()
 129                         // for their own custom callback to echo the notice
 130                         wfDebug( wfMessage( $msg, $params )->text() . "\n" );
 131                 }
 132         }
 133
 134         /**
 135          * Set debug mode...
 136          * @param bool $debug
 137          */
 138         function setDebug( $debug ) {
 139                 $this->mDebug = $debug;
 140         }
 141
 142         /**
 143          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
 144          * @param bool $noupdates
 145          */
 146         function setNoUpdates( $noupdates ) {
 147                 $this->mNoUpdates = $noupdates;
 148         }
 149
 150         /**
 151          * Sets 'pageOffset' value. So it will skip the first n-1 pages
 152          * and start from the nth page. It's 1-based indexing.
 153          * @param int $nthPage
 154          * @since 1.29
 155          */
 156         function setPageOffset( $nthPage ) {
 157                 $this->pageOffset = $nthPage;
 158         }
 159
 160         /**
 161          * Set a callback that displays notice messages
 162          *
 163          * @param callable $callback
 164          * @return callable
 165          */
 166         public function setNoticeCallback( $callback ) {
 167                 return wfSetVar( $this->mNoticeCallback, $callback );
 168         }
 169
 170         /**
 171          * Sets the action to perform as each new page in the stream is reached.
 172          * @param callable $callback
 173          * @return callable
 174          */
 175         public function setPageCallback( $callback ) {
 176                 $previous = $this->mPageCallback;
 177                 $this->mPageCallback = $callback;
 178                 return $previous;
 179         }
 180
 181         /**
 182          * Sets the action to perform as each page in the stream is completed.
 183          * Callback accepts the page title (as a Title object), a second object
 184          * with the original title form (in case it's been overridden into a
 185          * local namespace), and a count of revisions.
 186          *
 187          * @param callable $callback
 188          * @return callable
 189          */
 190         public function setPageOutCallback( $callback ) {
 191                 $previous = $this->mPageOutCallback;
 192                 $this->mPageOutCallback = $callback;
 193                 return $previous;
 194         }
 195
 196         /**
 197          * Sets the action to perform as each page revision is reached.
 198          * @param callable $callback
 199          * @return callable
 200          */
 201         public function setRevisionCallback( $callback ) {
 202                 $previous = $this->mRevisionCallback;
 203                 $this->mRevisionCallback = $callback;
 204                 return $previous;
 205         }
 206
 207         /**
 208          * Sets the action to perform as each file upload version is reached.
 209          * @param callable $callback
 210          * @return callable
 211          */
 212         public function setUploadCallback( $callback ) {
 213                 $previous = $this->mUploadCallback;
 214                 $this->mUploadCallback = $callback;
 215                 return $previous;
 216         }
 217
 218         /**
 219          * Sets the action to perform as each log item reached.
 220          * @param callable $callback
 221          * @return callable
 222          */
 223         public function setLogItemCallback( $callback ) {
 224                 $previous = $this->mLogItemCallback;
 225                 $this->mLogItemCallback = $callback;
 226                 return $previous;
 227         }
 228
 229         /**
 230          * Sets the action to perform when site info is encountered
 231          * @param callable $callback
 232          * @return callable
 233          */
 234         public function setSiteInfoCallback( $callback ) {
 235                 $previous = $this->mSiteInfoCallback;
 236                 $this->mSiteInfoCallback = $callback;
 237                 return $previous;
 238         }
 239
 240         /**
 241          * Sets the factory object to use to convert ForeignTitle objects into local
 242          * Title objects
 243          * @param ImportTitleFactory $factory
 244          */
 245         public function setImportTitleFactory( $factory ) {
 246                 $this->importTitleFactory = $factory;
 247         }
 248
 249         /**
 250          * Set a target namespace to override the defaults
 251          * @param null|int $namespace
 252          * @return bool
 253          */
 254         public function setTargetNamespace( $namespace ) {
 255                 if ( is_null( $namespace ) ) {
 256                         // Don't override namespaces
 257                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 258                         return true;
 259                 } elseif (
 260                         $namespace >= 0 &&
 261                         MWNamespace::exists( intval( $namespace ) )
 262                 ) {
 263                         $namespace = intval( $namespace );
 264                         $this->setImportTitleFactory( new NamespaceImportTitleFactory( $namespace ) );
 265                         return true;
 266                 } else {
 267                         return false;
 268                 }
 269         }
 270
 271         /**
 272          * Set a target root page under which all pages are imported
 273          * @param null|string $rootpage
 274          * @return Status
 275          */
 276         public function setTargetRootPage( $rootpage ) {
 277                 $status = Status::newGood();
 278                 if ( is_null( $rootpage ) ) {
 279                         // No rootpage
 280                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 281                 } elseif ( $rootpage !== '' ) {
 282                         $rootpage = rtrim( $rootpage, '/' ); // avoid double slashes
 283                         $title = Title::newFromText( $rootpage );
 284
 285                         if ( !$title || $title->isExternal() ) {
 286                                 $status->fatal( 'import-rootpage-invalid' );
 287                         } else {
 288                                 if ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 289                                         global $wgContLang;
 290
 291                                         $displayNSText = $title->getNamespace() == NS_MAIN
 292                                                 ? wfMessage( 'blanknamespace' )->text()
 293                                                 : $wgContLang->getNsText( $title->getNamespace() );
 294                                         $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 295                                 } else {
 296                                         // set namespace to 'all', so the namespace check in processTitle() can pass
 297                                         $this->setTargetNamespace( null );
 298                                         $this->setImportTitleFactory( new SubpageImportTitleFactory( $title ) );
 299                                 }
 300                         }
 301                 }
 302                 return $status;
 303         }
 304
 305         /**
 306          * @param string $dir
 307          */
 308         public function setImageBasePath( $dir ) {
 309                 $this->mImageBasePath = $dir;
 310         }
 311
 312         /**
 313          * @param bool $import
 314          */
 315         public function setImportUploads( $import ) {
 316                 $this->mImportUploads = $import;
 317         }
 318
 319         /**
 320          * @since 1.31
 321          * @param string $usernamePrefix Prefix to apply to unknown (and possibly also known) usernames
 322          * @param bool $assignKnownUsers Whether to apply the prefix to usernames that exist locally
 323          */
 324         public function setUsernamePrefix( $usernamePrefix, $assignKnownUsers ) {
 325                 $this->usernamePrefix = rtrim( (string)$usernamePrefix, ':>' );
 326                 $this->assignKnownUsers = (bool)$assignKnownUsers;
 327         }
 328
 329         /**
 330          * Statistics update can cause a lot of time
 331          * @since 1.29
 332          */
 333         public function disableStatisticsUpdate() {
 334                 $this->disableStatisticsUpdate = true;
 335         }
 336
 337         /**
 338          * Default per-page callback. Sets up some things related to site statistics
 339          * @param array $titleAndForeignTitle Two-element array, with Title object at
 340          * index 0 and ForeignTitle object at index 1
 341          * @return bool
 342          */
 343         public function beforeImportPage( $titleAndForeignTitle ) {
 344                 $title = $titleAndForeignTitle[0];
 345                 $page = WikiPage::factory( $title );
 346                 $this->countableCache['title_' . $title->getPrefixedText()] = $page->isCountable();
 347                 return true;
 348         }
 349
 350         /**
 351          * Default per-revision callback, performs the import.
 352          * @param WikiRevision $revision
 353          * @return bool
 354          */
 355         public function importRevision( $revision ) {
 356                 if ( !$revision->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
 357                         $this->notice( 'import-error-bad-location',
 358                                 $revision->getTitle()->getPrefixedText(),
 359                                 $revision->getID(),
 360                                 $revision->getModel(),
 361                                 $revision->getFormat() );
 362
 363                         return false;
 364                 }
 365
 366                 try {
 367                         return $revision->importOldRevision();
 368                 } catch ( MWContentSerializationException $ex ) {
 369                         $this->notice( 'import-error-unserialize',
 370                                 $revision->getTitle()->getPrefixedText(),
 371                                 $revision->getID(),
 372                                 $revision->getModel(),
 373                                 $revision->getFormat() );
 374                 }
 375
 376                 return false;
 377         }
 378
 379         /**
 380          * Default per-revision callback, performs the import.
 381          * @param WikiRevision $revision
 382          * @return bool
 383          */
 384         public function importLogItem( $revision ) {
 385                 return $revision->importLogItem();
 386         }
 387
 388         /**
 389          * Dummy for now...
 390          * @param WikiRevision $revision
 391          * @return bool
 392          */
 393         public function importUpload( $revision ) {
 394                 return $revision->importUpload();
 395         }
 396
 397         /**
 398          * Mostly for hook use
 399          * @param Title $title
 400          * @param ForeignTitle $foreignTitle
 401          * @param int $revCount
 402          * @param int $sRevCount
 403          * @param array $pageInfo
 404          * @return bool
 405          */
 406         public function finishImportPage( $title, $foreignTitle, $revCount,
 407                 $sRevCount, $pageInfo
 408         ) {
 409                 // Update article count statistics (T42009)
 410                 // The normal counting logic in WikiPage->doEditUpdates() is designed for
 411                 // one-revision-at-a-time editing, not bulk imports. In this situation it
 412                 // suffers from issues of replica DB lag. We let WikiPage handle the total page
 413                 // and revision count, and we implement our own custom logic for the
 414                 // article (content page) count.
 415                 if ( !$this->disableStatisticsUpdate ) {
 416                         $page = WikiPage::factory( $title );
 417                         $page->loadPageData( 'fromdbmaster' );
 418                         $content = $page->getContent();
 419                         if ( $content === null ) {
 420                                 wfDebug( __METHOD__ . ': Skipping article count adjustment for ' . $title .
 421                                         ' because WikiPage::getContent() returned null' );
 422                         } else {
 423                                 $editInfo = $page->prepareContentForEdit( $content );
 424                                 $countKey = 'title_' . $title->getPrefixedText();
 425                                 $countable = $page->isCountable( $editInfo );
 426                                 if ( array_key_exists( $countKey, $this->countableCache ) &&
 427                                         $countable != $this->countableCache[$countKey] ) {
 428                                         DeferredUpdates::addUpdate( SiteStatsUpdate::factory( [
 429                                                 'articles' => ( (int)$countable - (int)$this->countableCache[$countKey] )
 430                                         ] ) );
 431                                 }
 432                         }
 433                 }
 434
 435                 $args = func_get_args();
 436                 return Hooks::run( 'AfterImportPage', $args );
 437         }
 438
 439         /**
 440          * Alternate per-revision callback, for debugging.
 441          * @param WikiRevision &$revision
 442          */
 443         public function debugRevisionHandler( &$revision ) {
 444                 $this->debug( "Got revision:" );
 445                 if ( is_object( $revision->title ) ) {
 446                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 447                 } else {
 448                         $this->debug( "-- Title: <invalid>" );
 449                 }
 450                 $this->debug( "-- User: " . $revision->user_text );
 451                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 452                 $this->debug( "-- Comment: " . $revision->comment );
 453                 $this->debug( "-- Text: " . $revision->text );
 454         }
 455
 456         /**
 457          * Notify the callback function of site info
 458          * @param array $siteInfo
 459          * @return bool|mixed
 460          */
 461         private function siteInfoCallback( $siteInfo ) {
 462                 if ( isset( $this->mSiteInfoCallback ) ) {
 463                         return call_user_func_array( $this->mSiteInfoCallback,
 464                                         [ $siteInfo, $this ] );
 465                 } else {
 466                         return false;
 467                 }
 468         }
 469
 470         /**
 471          * Notify the callback function when a new "<page>" is reached.
 472          * @param Title $title
 473          */
 474         function pageCallback( $title ) {
 475                 if ( isset( $this->mPageCallback ) ) {
 476                         call_user_func( $this->mPageCallback, $title );
 477                 }
 478         }
 479
 480         /**
 481          * Notify the callback function when a "</page>" is closed.
 482          * @param Title $title
 483          * @param ForeignTitle $foreignTitle
 484          * @param int $revCount
 485          * @param int $sucCount Number of revisions for which callback returned true
 486          * @param array $pageInfo Associative array of page information
 487          */
 488         private function pageOutCallback( $title, $foreignTitle, $revCount,
 489                         $sucCount, $pageInfo ) {
 490                 if ( isset( $this->mPageOutCallback ) ) {
 491                         $args = func_get_args();
 492                         call_user_func_array( $this->mPageOutCallback, $args );
 493                 }
 494         }
 495
 496         /**
 497          * Notify the callback function of a revision
 498          * @param WikiRevision $revision
 499          * @return bool|mixed
 500          */
 501         private function revisionCallback( $revision ) {
 502                 if ( isset( $this->mRevisionCallback ) ) {
 503                         return call_user_func_array( $this->mRevisionCallback,
 504                                         [ $revision, $this ] );
 505                 } else {
 506                         return false;
 507                 }
 508         }
 509
 510         /**
 511          * Notify the callback function of a new log item
 512          * @param WikiRevision $revision
 513          * @return bool|mixed
 514          */
 515         private function logItemCallback( $revision ) {
 516                 if ( isset( $this->mLogItemCallback ) ) {
 517                         return call_user_func_array( $this->mLogItemCallback,
 518                                         [ $revision, $this ] );
 519                 } else {
 520                         return false;
 521                 }
 522         }
 523
 524         /**
 525          * Retrieves the contents of the named attribute of the current element.
 526          * @param string $attr The name of the attribute
 527          * @return string The value of the attribute or an empty string if it is not set in the current
 528          * element.
 529          */
 530         public function nodeAttribute( $attr ) {
 531                 return $this->reader->getAttribute( $attr );
 532         }
 533
 534         /**
 535          * Shouldn't something like this be built-in to XMLReader?
 536          * Fetches text contents of the current element, assuming
 537          * no sub-elements or such scary things.
 538          * @return string
 539          * @access private
 540          */
 541         public function nodeContents() {
 542                 if ( $this->reader->isEmptyElement ) {
 543                         return "";
 544                 }
 545                 $buffer = "";
 546                 while ( $this->reader->read() ) {
 547                         switch ( $this->reader->nodeType ) {
 548                                 case XMLReader::TEXT:
 549                                 case XMLReader::CDATA:
 550                                 case XMLReader::SIGNIFICANT_WHITESPACE:
 551                                         $buffer .= $this->reader->value;
 552                                         break;
 553                                 case XMLReader::END_ELEMENT:
 554                                         return $buffer;
 555                         }
 556                 }
 557
 558                 $this->reader->close();
 559                 return '';
 560         }
 561
 562         /**
 563          * Primary entry point
 564          * @throws Exception
 565          * @throws MWException
 566          * @return bool
 567          */
 568         public function doImport() {
 569                 // Calls to reader->read need to be wrapped in calls to
 570                 // libxml_disable_entity_loader() to avoid local file
 571                 // inclusion attacks (T48932).
 572                 $oldDisable = libxml_disable_entity_loader( true );
 573                 $this->reader->read();
 574
 575                 if ( $this->reader->localName != 'mediawiki' ) {
 576                         libxml_disable_entity_loader( $oldDisable );
 577                         throw new MWException( "Expected <mediawiki> tag, got " .
 578                                 $this->reader->localName );
 579                 }
 580                 $this->debug( "<mediawiki> tag is correct." );
 581
 582                 $this->debug( "Starting primary dump processing loop." );
 583
 584                 $keepReading = $this->reader->read();
 585                 $skip = false;
 586                 $rethrow = null;
 587                 $pageCount = 0;
 588                 try {
 589                         while ( $keepReading ) {
 590                                 $tag = $this->reader->localName;
 591                                 if ( $this->pageOffset ) {
 592                                         if ( $tag === 'page' ) {
 593                                                 $pageCount++;
 594                                         }
 595                                         if ( $pageCount < $this->pageOffset ) {
 596                                                 $keepReading = $this->reader->next();
 597                                                 continue;
 598                                         }
 599                                 }
 600                                 $type = $this->reader->nodeType;
 601
 602                                 if ( !Hooks::run( 'ImportHandleToplevelXMLTag', [ $this ] ) ) {
 603                                         // Do nothing
 604                                 } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) {
 605                                         break;
 606                                 } elseif ( $tag == 'siteinfo' ) {
 607                                         $this->handleSiteInfo();
 608                                 } elseif ( $tag == 'page' ) {
 609                                         $this->handlePage();
 610                                 } elseif ( $tag == 'logitem' ) {
 611                                         $this->handleLogItem();
 612                                 } elseif ( $tag != '#text' ) {
 613                                         $this->warn( "Unhandled top-level XML tag $tag" );
 614
 615                                         $skip = true;
 616                                 }
 617
 618                                 if ( $skip ) {
 619                                         $keepReading = $this->reader->next();
 620                                         $skip = false;
 621                                         $this->debug( "Skip" );
 622                                 } else {
 623                                         $keepReading = $this->reader->read();
 624                                 }
 625                         }
 626                 } catch ( Exception $ex ) {
 627                         $rethrow = $ex;
 628                 }
 629
 630                 // finally
 631                 libxml_disable_entity_loader( $oldDisable );
 632                 $this->reader->close();
 633
 634                 if ( $rethrow ) {
 635                         throw $rethrow;
 636                 }
 637
 638                 return true;
 639         }
 640
 641         private function handleSiteInfo() {
 642                 $this->debug( "Enter site info handler." );
 643                 $siteInfo = [];
 644
 645                 // Fields that can just be stuffed in the siteInfo object
 646                 $normalFields = [ 'sitename', 'base', 'generator', 'case' ];
 647
 648                 while ( $this->reader->read() ) {
 649                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 650                                         $this->reader->localName == 'siteinfo' ) {
 651                                 break;
 652                         }
 653
 654                         $tag = $this->reader->localName;
 655
 656                         if ( $tag == 'namespace' ) {
 657                                 $this->foreignNamespaces[$this->nodeAttribute( 'key' )] =
 658                                         $this->nodeContents();
 659                         } elseif ( in_array( $tag, $normalFields ) ) {
 660                                 $siteInfo[$tag] = $this->nodeContents();
 661                         }
 662                 }
 663
 664                 $siteInfo['_namespaces'] = $this->foreignNamespaces;
 665                 $this->siteInfoCallback( $siteInfo );
 666         }
 667
 668         private function handleLogItem() {
 669                 $this->debug( "Enter log item handler." );
 670                 $logInfo = [];
 671
 672                 // Fields that can just be stuffed in the pageInfo object
 673                 $normalFields = [ 'id', 'comment', 'type', 'action', 'timestamp',
 674                                         'logtitle', 'params' ];
 675
 676                 while ( $this->reader->read() ) {
 677                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 678                                         $this->reader->localName == 'logitem' ) {
 679                                 break;
 680                         }
 681
 682                         $tag = $this->reader->localName;
 683
 684                         if ( !Hooks::run( 'ImportHandleLogItemXMLTag', [
 685                                 $this, $logInfo
 686                         ] ) ) {
 687                                 // Do nothing
 688                         } elseif ( in_array( $tag, $normalFields ) ) {
 689                                 $logInfo[$tag] = $this->nodeContents();
 690                         } elseif ( $tag == 'contributor' ) {
 691                                 $logInfo['contributor'] = $this->handleContributor();
 692                         } elseif ( $tag != '#text' ) {
 693                                 $this->warn( "Unhandled log-item XML tag $tag" );
 694                         }
 695                 }
 696
 697                 $this->processLogItem( $logInfo );
 698         }
 699
 700         /**
 701          * @param array $logInfo
 702          * @return bool|mixed
 703          */
 704         private function processLogItem( $logInfo ) {
 705                 $revision = new WikiRevision( $this->config );
 706
 707                 if ( isset( $logInfo['id'] ) ) {
 708                         $revision->setID( $logInfo['id'] );
 709                 }
 710                 $revision->setType( $logInfo['type'] );
 711                 $revision->setAction( $logInfo['action'] );
 712                 if ( isset( $logInfo['timestamp'] ) ) {
 713                         $revision->setTimestamp( $logInfo['timestamp'] );
 714                 }
 715                 if ( isset( $logInfo['params'] ) ) {
 716                         $revision->setParams( $logInfo['params'] );
 717                 }
 718                 if ( isset( $logInfo['logtitle'] ) ) {
 719                         // @todo Using Title for non-local titles is a recipe for disaster.
 720                         // We should use ForeignTitle here instead.
 721                         $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 722                 }
 723
 724                 $revision->setNoUpdates( $this->mNoUpdates );
 725
 726                 if ( isset( $logInfo['comment'] ) ) {
 727                         $revision->setComment( $logInfo['comment'] );
 728                 }
 729
 730                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 731                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 732                 }
 733
 734                 if ( !isset( $logInfo['contributor']['username'] ) ) {
 735                         $revision->setUsername( $this->usernamePrefix . '>Unknown user' );
 736                 } else {
 737                         $revision->setUsername( $this->prefixUsername( $logInfo['contributor']['username'] ) );
 738                 }
 739
 740                 return $this->logItemCallback( $revision );
 741         }
 742
 743         private function handlePage() {
 744                 // Handle page data.
 745                 $this->debug( "Enter page handler." );
 746                 $pageInfo = [ 'revisionCount' => 0, 'successfulRevisionCount' => 0 ];
 747
 748                 // Fields that can just be stuffed in the pageInfo object
 749                 $normalFields = [ 'title', 'ns', 'id', 'redirect', 'restrictions' ];
 750
 751                 $skip = false;
 752                 $badTitle = false;
 753
 754                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 755                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 756                                         $this->reader->localName == 'page' ) {
 757                                 break;
 758                         }
 759
 760                         $skip = false;
 761
 762                         $tag = $this->reader->localName;
 763
 764                         if ( $badTitle ) {
 765                                 // The title is invalid, bail out of this page
 766                                 $skip = true;
 767                         } elseif ( !Hooks::run( 'ImportHandlePageXMLTag', [ $this,
 768                                                 &$pageInfo ] ) ) {
 769                                 // Do nothing
 770                         } elseif ( in_array( $tag, $normalFields ) ) {
 771                                 // An XML snippet:
 772                                 // <page>
 773                                 //     <id>123</id>
 774                                 //     <title>Page</title>
 775                                 //     <redirect title="NewTitle"/>
 776                                 //     ...
 777                                 // Because the redirect tag is built differently, we need special handling for that case.
 778                                 if ( $tag == 'redirect' ) {
 779                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 780                                 } else {
 781                                         $pageInfo[$tag] = $this->nodeContents();
 782                                 }
 783                         } elseif ( $tag == 'revision' || $tag == 'upload' ) {
 784                                 if ( !isset( $title ) ) {
 785                                         $title = $this->processTitle( $pageInfo['title'],
 786                                                 isset( $pageInfo['ns'] ) ? $pageInfo['ns'] : null );
 787
 788                                         // $title is either an array of two titles or false.
 789                                         if ( is_array( $title ) ) {
 790                                                 $this->pageCallback( $title );
 791                                                 list( $pageInfo['_title'], $foreignTitle ) = $title;
 792                                         } else {
 793                                                 $badTitle = true;
 794                                                 $skip = true;
 795                                         }
 796                                 }
 797
 798                                 if ( $title ) {
 799                                         if ( $tag == 'revision' ) {
 800                                                 $this->handleRevision( $pageInfo );
 801                                         } else {
 802                                                 $this->handleUpload( $pageInfo );
 803                                         }
 804                                 }
 805                         } elseif ( $tag != '#text' ) {
 806                                 $this->warn( "Unhandled page XML tag $tag" );
 807                                 $skip = true;
 808                         }
 809                 }
 810
 811                 // @note $pageInfo is only set if a valid $title is processed above with
 812                 //       no error. If we have a valid $title, then pageCallback is called
 813                 //       above, $pageInfo['title'] is set and we do pageOutCallback here.
 814                 //       If $pageInfo['_title'] is not set, then $foreignTitle is also not
 815                 //       set since they both come from $title above.
 816                 if ( array_key_exists( '_title', $pageInfo ) ) {
 817                         $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
 818                                         $pageInfo['revisionCount'],
 819                                         $pageInfo['successfulRevisionCount'],
 820                                         $pageInfo );
 821                 }
 822         }
 823
 824         /**
 825          * @param array $pageInfo
 826          */
 827         private function handleRevision( &$pageInfo ) {
 828                 $this->debug( "Enter revision handler" );
 829                 $revisionInfo = [];
 830
 831                 $normalFields = [ 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text', 'sha1' ];
 832
 833                 $skip = false;
 834
 835                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 836                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 837                                         $this->reader->localName == 'revision' ) {
 838                                 break;
 839                         }
 840
 841                         $tag = $this->reader->localName;
 842
 843                         if ( !Hooks::run( 'ImportHandleRevisionXMLTag', [
 844                                 $this, $pageInfo, $revisionInfo
 845                         ] ) ) {
 846                                 // Do nothing
 847                         } elseif ( in_array( $tag, $normalFields ) ) {
 848                                 $revisionInfo[$tag] = $this->nodeContents();
 849                         } elseif ( $tag == 'contributor' ) {
 850                                 $revisionInfo['contributor'] = $this->handleContributor();
 851                         } elseif ( $tag != '#text' ) {
 852                                 $this->warn( "Unhandled revision XML tag $tag" );
 853                                 $skip = true;
 854                         }
 855                 }
 856
 857                 $pageInfo['revisionCount']++;
 858                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 859                         $pageInfo['successfulRevisionCount']++;
 860                 }
 861         }
 862
 863         /**
 864          * @param array $pageInfo
 865          * @param array $revisionInfo
 866          * @throws MWException
 867          * @return bool|mixed
 868          */
 869         private function processRevision( $pageInfo, $revisionInfo ) {
 870                 global $wgMaxArticleSize;
 871
 872                 // Make sure revisions won't violate $wgMaxArticleSize, which could lead to
 873                 // database errors and instability. Testing for revisions with only listed
 874                 // content models, as other content models might use serialization formats
 875                 // which aren't checked against $wgMaxArticleSize.
 876                 if ( ( !isset( $revisionInfo['model'] ) ||
 877                         in_array( $revisionInfo['model'], [
 878                                 'wikitext',
 879                                 'css',
 880                                 'json',
 881                                 'javascript',
 882                                 'text',
 883                                 ''
 884                         ] ) ) &&
 885                         strlen( $revisionInfo['text'] ) > $wgMaxArticleSize * 1024
 886                 ) {
 887                         throw new MWException( 'The text of ' .
 888                                 ( isset( $revisionInfo['id'] ) ?
 889                                         "the revision with ID $revisionInfo[id]" :
 890                                         'a revision'
 891                                 ) . " exceeds the maximum allowable size ($wgMaxArticleSize KB)" );
 892                 }
 893
 894                 $revision = new WikiRevision( $this->config );
 895
 896                 if ( isset( $revisionInfo['id'] ) ) {
 897                         $revision->setID( $revisionInfo['id'] );
 898                 }
 899                 if ( isset( $revisionInfo['model'] ) ) {
 900                         $revision->setModel( $revisionInfo['model'] );
 901                 }
 902                 if ( isset( $revisionInfo['format'] ) ) {
 903                         $revision->setFormat( $revisionInfo['format'] );
 904                 }
 905                 $revision->setTitle( $pageInfo['_title'] );
 906
 907                 if ( isset( $revisionInfo['text'] ) ) {
 908                         $handler = $revision->getContentHandler();
 909                         $text = $handler->importTransform(
 910                                 $revisionInfo['text'],
 911                                 $revision->getFormat() );
 912
 913                         $revision->setText( $text );
 914                 }
 915                 if ( isset( $revisionInfo['timestamp'] ) ) {
 916                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 917                 } else {
 918                         $revision->setTimestamp( wfTimestampNow() );
 919                 }
 920
 921                 if ( isset( $revisionInfo['comment'] ) ) {
 922                         $revision->setComment( $revisionInfo['comment'] );
 923                 }
 924
 925                 if ( isset( $revisionInfo['minor'] ) ) {
 926                         $revision->setMinor( true );
 927                 }
 928                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 929                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 930                 } elseif ( isset( $revisionInfo['contributor']['username'] ) ) {
 931                         $revision->setUsername( $this->prefixUsername( $revisionInfo['contributor']['username'] ) );
 932                 } else {
 933                         $revision->setUsername( $this->usernamePrefix . '>Unknown user' );
 934                 }
 935                 if ( isset( $revisionInfo['sha1'] ) ) {
 936                         $revision->setSha1Base36( $revisionInfo['sha1'] );
 937                 }
 938                 $revision->setNoUpdates( $this->mNoUpdates );
 939
 940                 return $this->revisionCallback( $revision );
 941         }
 942
 943         /**
 944          * @param array $pageInfo
 945          * @return mixed
 946          */
 947         private function handleUpload( &$pageInfo ) {
 948                 $this->debug( "Enter upload handler" );
 949                 $uploadInfo = [];
 950
 951                 $normalFields = [ 'timestamp', 'comment', 'filename', 'text',
 952                                         'src', 'size', 'sha1base36', 'archivename', 'rel' ];
 953
 954                 $skip = false;
 955
 956                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 957                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 958                                         $this->reader->localName == 'upload' ) {
 959                                 break;
 960                         }
 961
 962                         $tag = $this->reader->localName;
 963
 964                         if ( !Hooks::run( 'ImportHandleUploadXMLTag', [
 965                                 $this, $pageInfo
 966                         ] ) ) {
 967                                 // Do nothing
 968                         } elseif ( in_array( $tag, $normalFields ) ) {
 969                                 $uploadInfo[$tag] = $this->nodeContents();
 970                         } elseif ( $tag == 'contributor' ) {
 971                                 $uploadInfo['contributor'] = $this->handleContributor();
 972                         } elseif ( $tag == 'contents' ) {
 973                                 $contents = $this->nodeContents();
 974                                 $encoding = $this->reader->getAttribute( 'encoding' );
 975                                 if ( $encoding === 'base64' ) {
 976                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 977                                         $uploadInfo['isTempSrc'] = true;
 978                                 }
 979                         } elseif ( $tag != '#text' ) {
 980                                 $this->warn( "Unhandled upload XML tag $tag" );
 981                                 $skip = true;
 982                         }
 983                 }
 984
 985                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 986                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 987                         if ( file_exists( $path ) ) {
 988                                 $uploadInfo['fileSrc'] = $path;
 989                                 $uploadInfo['isTempSrc'] = false;
 990                         }
 991                 }
 992
 993                 if ( $this->mImportUploads ) {
 994                         return $this->processUpload( $pageInfo, $uploadInfo );
 995                 }
 996         }
 997
 998         /**
 999          * @param string $contents
1000          * @return string
1001          */
1002         private function dumpTemp( $contents ) {
1003                 $filename = tempnam( wfTempDir(), 'importupload' );
1004                 file_put_contents( $filename, $contents );
1005                 return $filename;
1006         }
1007
1008         /**
1009          * @param array $pageInfo
1010          * @param array $uploadInfo
1011          * @return mixed
1012          */
1013         private function processUpload( $pageInfo, $uploadInfo ) {
1014                 $revision = new WikiRevision( $this->config );
1015                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
1016
1017                 $revision->setTitle( $pageInfo['_title'] );
1018                 $revision->setID( $pageInfo['id'] );
1019                 $revision->setTimestamp( $uploadInfo['timestamp'] );
1020                 $revision->setText( $text );
1021                 $revision->setFilename( $uploadInfo['filename'] );
1022                 if ( isset( $uploadInfo['archivename'] ) ) {
1023                         $revision->setArchiveName( $uploadInfo['archivename'] );
1024                 }
1025                 $revision->setSrc( $uploadInfo['src'] );
1026                 if ( isset( $uploadInfo['fileSrc'] ) ) {
1027                         $revision->setFileSrc( $uploadInfo['fileSrc'],
1028                                 !empty( $uploadInfo['isTempSrc'] ) );
1029                 }
1030                 if ( isset( $uploadInfo['sha1base36'] ) ) {
1031                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
1032                 }
1033                 $revision->setSize( intval( $uploadInfo['size'] ) );
1034                 $revision->setComment( $uploadInfo['comment'] );
1035
1036                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
1037                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
1038                 }
1039                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
1040                         $revision->setUsername( $this->prefixUsername( $uploadInfo['contributor']['username'] ) );
1041                 }
1042                 $revision->setNoUpdates( $this->mNoUpdates );
1043
1044                 return call_user_func( $this->mUploadCallback, $revision );
1045         }
1046
1047         /**
1048          * Add an interwiki prefix to the username, if appropriate
1049          * @since 1.31
1050          * @param string $name Name being imported
1051          * @return string Name, possibly with the prefix prepended.
1052          */
1053         protected function prefixUsername( $name ) {
1054                 if ( !User::isUsableName( $name ) ) {
1055                         return $name;
1056                 }
1057
1058                 if ( $this->assignKnownUsers ) {
1059                         if ( User::idFromName( $name ) ) {
1060                                 return $name;
1061                         }
1062
1063                         // See if any extension wants to create it.
1064                         if ( !isset( $this->triedCreations[$name] ) ) {
1065                                 $this->triedCreations[$name] = true;
1066                                 if ( !Hooks::run( 'ImportHandleUnknownUser', [ $name ] ) &&
1067                                         User::idFromName( $name, User::READ_LATEST )
1068                                 ) {
1069                                         return $name;
1070                                 }
1071                         }
1072                 }
1073
1074                 return substr( $this->usernamePrefix . '>' . $name, 0, 255 );
1075         }
1076
1077         /**
1078          * @return array
1079          */
1080         private function handleContributor() {
1081                 $fields = [ 'id', 'ip', 'username' ];
1082                 $info = [];
1083
1084                 if ( $this->reader->isEmptyElement ) {
1085                         return $info;
1086                 }
1087                 while ( $this->reader->read() ) {
1088                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
1089                                         $this->reader->localName == 'contributor' ) {
1090                                 break;
1091                         }
1092
1093                         $tag = $this->reader->localName;
1094
1095                         if ( in_array( $tag, $fields ) ) {
1096                                 $info[$tag] = $this->nodeContents();
1097                         }
1098                 }
1099
1100                 return $info;
1101         }
1102
1103         /**
1104          * @param string $text
1105          * @param string|null $ns
1106          * @return array|bool
1107          */
1108         private function processTitle( $text, $ns = null ) {
1109                 if ( is_null( $this->foreignNamespaces ) ) {
1110                         $foreignTitleFactory = new NaiveForeignTitleFactory();
1111                 } else {
1112                         $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
1113                                 $this->foreignNamespaces );
1114                 }
1115
1116                 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
1117                         intval( $ns ) );
1118
1119                 $title = $this->importTitleFactory->createTitleFromForeignTitle(
1120                         $foreignTitle );
1121
1122                 $commandLineMode = $this->config->get( 'CommandLineMode' );
1123                 if ( is_null( $title ) ) {
1124                         # Invalid page title? Ignore the page
1125                         $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
1126                         return false;
1127                 } elseif ( $title->isExternal() ) {
1128                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
1129                         return false;
1130                 } elseif ( !$title->canExist() ) {
1131                         $this->notice( 'import-error-special', $title->getPrefixedText() );
1132                         return false;
1133                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
1134                         # Do not import if the importing wiki user cannot edit this page
1135                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
1136                         return false;
1137                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
1138                         # Do not import if the importing wiki user cannot create this page
1139                         $this->notice( 'import-error-create', $title->getPrefixedText() );
1140                         return false;
1141                 }
1142
1143                 return [ $title, $foreignTitle ];
1144         }
1145 }