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
  51         /**
  52          * Creates an ImportXMLReader drawing from the source provided
  53          * @param ImportSource $source
  54          * @param Config $config
  55          * @throws Exception
  56          */
  57         function __construct( ImportSource $source, Config $config ) {
  58                 if ( !class_exists( 'XMLReader' ) ) {
  59                         throw new Exception( 'Import requires PHP to have been compiled with libxml support' );
  60                 }
  61
  62                 $this->reader = new XMLReader();
  63                 $this->config = $config;
  64
  65                 if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) {
  66                         stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  67                 }
  68                 $id = UploadSourceAdapter::registerSource( $source );
  69
  70                 // Enable the entity loader, as it is needed for loading external URLs via
  71                 // XMLReader::open (T86036)
  72                 $oldDisable = libxml_disable_entity_loader( false );
  73                 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
  74                         $status = $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  75                 } else {
  76                         $status = $this->reader->open( "uploadsource://$id" );
  77                 }
  78                 if ( !$status ) {
  79                         $error = libxml_get_last_error();
  80                         libxml_disable_entity_loader( $oldDisable );
  81                         throw new MWException( 'Encountered an internal error while initializing WikiImporter object: ' .
  82                                 $error->message );
  83                 }
  84                 libxml_disable_entity_loader( $oldDisable );
  85
  86                 // Default callbacks
  87                 $this->setPageCallback( [ $this, 'beforeImportPage' ] );
  88                 $this->setRevisionCallback( [ $this, "importRevision" ] );
  89                 $this->setUploadCallback( [ $this, 'importUpload' ] );
  90                 $this->setLogItemCallback( [ $this, 'importLogItem' ] );
  91                 $this->setPageOutCallback( [ $this, 'finishImportPage' ] );
  92
  93                 $this->importTitleFactory = new NaiveImportTitleFactory();
  94         }
  95
  96         /**
  97          * @return null|XMLReader
  98          */
  99         public function getReader() {
 100                 return $this->reader;
 101         }
 102
 103         public function throwXmlError( $err ) {
 104                 $this->debug( "FAILURE: $err" );
 105                 wfDebug( "WikiImporter XML error: $err\n" );
 106         }
 107
 108         public function debug( $data ) {
 109                 if ( $this->mDebug ) {
 110                         wfDebug( "IMPORT: $data\n" );
 111                 }
 112         }
 113
 114         public function warn( $data ) {
 115                 wfDebug( "IMPORT: $data\n" );
 116         }
 117
 118         public function notice( $msg /*, $param, ...*/ ) {
 119                 $params = func_get_args();
 120                 array_shift( $params );
 121
 122                 if ( is_callable( $this->mNoticeCallback ) ) {
 123                         call_user_func( $this->mNoticeCallback, $msg, $params );
 124                 } else { # No ImportReporter -> CLI
 125                         echo wfMessage( $msg, $params )->text() . "\n";
 126                 }
 127         }
 128
 129         /**
 130          * Set debug mode...
 131          * @param bool $debug
 132          */
 133         function setDebug( $debug ) {
 134                 $this->mDebug = $debug;
 135         }
 136
 137         /**
 138          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
 139          * @param bool $noupdates
 140          */
 141         function setNoUpdates( $noupdates ) {
 142                 $this->mNoUpdates = $noupdates;
 143         }
 144
 145         /**
 146          * Sets 'pageOffset' value. So it will skip the first n-1 pages
 147          * and start from the nth page. It's 1-based indexing.
 148          * @param int $nthPage
 149          * @since 1.29
 150          */
 151         function setPageOffset( $nthPage ) {
 152                 $this->pageOffset = $nthPage;
 153         }
 154
 155         /**
 156          * Set a callback that displays notice messages
 157          *
 158          * @param callable $callback
 159          * @return callable
 160          */
 161         public function setNoticeCallback( $callback ) {
 162                 return wfSetVar( $this->mNoticeCallback, $callback );
 163         }
 164
 165         /**
 166          * Sets the action to perform as each new page in the stream is reached.
 167          * @param callable $callback
 168          * @return callable
 169          */
 170         public function setPageCallback( $callback ) {
 171                 $previous = $this->mPageCallback;
 172                 $this->mPageCallback = $callback;
 173                 return $previous;
 174         }
 175
 176         /**
 177          * Sets the action to perform as each page in the stream is completed.
 178          * Callback accepts the page title (as a Title object), a second object
 179          * with the original title form (in case it's been overridden into a
 180          * local namespace), and a count of revisions.
 181          *
 182          * @param callable $callback
 183          * @return callable
 184          */
 185         public function setPageOutCallback( $callback ) {
 186                 $previous = $this->mPageOutCallback;
 187                 $this->mPageOutCallback = $callback;
 188                 return $previous;
 189         }
 190
 191         /**
 192          * Sets the action to perform as each page revision is reached.
 193          * @param callable $callback
 194          * @return callable
 195          */
 196         public function setRevisionCallback( $callback ) {
 197                 $previous = $this->mRevisionCallback;
 198                 $this->mRevisionCallback = $callback;
 199                 return $previous;
 200         }
 201
 202         /**
 203          * Sets the action to perform as each file upload version is reached.
 204          * @param callable $callback
 205          * @return callable
 206          */
 207         public function setUploadCallback( $callback ) {
 208                 $previous = $this->mUploadCallback;
 209                 $this->mUploadCallback = $callback;
 210                 return $previous;
 211         }
 212
 213         /**
 214          * Sets the action to perform as each log item reached.
 215          * @param callable $callback
 216          * @return callable
 217          */
 218         public function setLogItemCallback( $callback ) {
 219                 $previous = $this->mLogItemCallback;
 220                 $this->mLogItemCallback = $callback;
 221                 return $previous;
 222         }
 223
 224         /**
 225          * Sets the action to perform when site info is encountered
 226          * @param callable $callback
 227          * @return callable
 228          */
 229         public function setSiteInfoCallback( $callback ) {
 230                 $previous = $this->mSiteInfoCallback;
 231                 $this->mSiteInfoCallback = $callback;
 232                 return $previous;
 233         }
 234
 235         /**
 236          * Sets the factory object to use to convert ForeignTitle objects into local
 237          * Title objects
 238          * @param ImportTitleFactory $factory
 239          */
 240         public function setImportTitleFactory( $factory ) {
 241                 $this->importTitleFactory = $factory;
 242         }
 243
 244         /**
 245          * Set a target namespace to override the defaults
 246          * @param null|int $namespace
 247          * @return bool
 248          */
 249         public function setTargetNamespace( $namespace ) {
 250                 if ( is_null( $namespace ) ) {
 251                         // Don't override namespaces
 252                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 253                         return true;
 254                 } elseif (
 255                         $namespace >= 0 &&
 256                         MWNamespace::exists( intval( $namespace ) )
 257                 ) {
 258                         $namespace = intval( $namespace );
 259                         $this->setImportTitleFactory( new NamespaceImportTitleFactory( $namespace ) );
 260                         return true;
 261                 } else {
 262                         return false;
 263                 }
 264         }
 265
 266         /**
 267          * Set a target root page under which all pages are imported
 268          * @param null|string $rootpage
 269          * @return Status
 270          */
 271         public function setTargetRootPage( $rootpage ) {
 272                 $status = Status::newGood();
 273                 if ( is_null( $rootpage ) ) {
 274                         // No rootpage
 275                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 276                 } elseif ( $rootpage !== '' ) {
 277                         $rootpage = rtrim( $rootpage, '/' ); // avoid double slashes
 278                         $title = Title::newFromText( $rootpage );
 279
 280                         if ( !$title || $title->isExternal() ) {
 281                                 $status->fatal( 'import-rootpage-invalid' );
 282                         } else {
 283                                 if ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 284                                         global $wgContLang;
 285
 286                                         $displayNSText = $title->getNamespace() == NS_MAIN
 287                                                 ? wfMessage( 'blanknamespace' )->text()
 288                                                 : $wgContLang->getNsText( $title->getNamespace() );
 289                                         $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 290                                 } else {
 291                                         // set namespace to 'all', so the namespace check in processTitle() can pass
 292                                         $this->setTargetNamespace( null );
 293                                         $this->setImportTitleFactory( new SubpageImportTitleFactory( $title ) );
 294                                 }
 295                         }
 296                 }
 297                 return $status;
 298         }
 299
 300         /**
 301          * @param string $dir
 302          */
 303         public function setImageBasePath( $dir ) {
 304                 $this->mImageBasePath = $dir;
 305         }
 306
 307         /**
 308          * @param bool $import
 309          */
 310         public function setImportUploads( $import ) {
 311                 $this->mImportUploads = $import;
 312         }
 313
 314         /**
 315          * Statistics update can cause a lot of time
 316          * @since 1.29
 317          */
 318         public function disableStatisticsUpdate() {
 319                 $this->disableStatisticsUpdate = true;
 320         }
 321
 322         /**
 323          * Default per-page callback. Sets up some things related to site statistics
 324          * @param array $titleAndForeignTitle Two-element array, with Title object at
 325          * index 0 and ForeignTitle object at index 1
 326          * @return bool
 327          */
 328         public function beforeImportPage( $titleAndForeignTitle ) {
 329                 $title = $titleAndForeignTitle[0];
 330                 $page = WikiPage::factory( $title );
 331                 $this->countableCache['title_' . $title->getPrefixedText()] = $page->isCountable();
 332                 return true;
 333         }
 334
 335         /**
 336          * Default per-revision callback, performs the import.
 337          * @param WikiRevision $revision
 338          * @return bool
 339          */
 340         public function importRevision( $revision ) {
 341                 if ( !$revision->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
 342                         $this->notice( 'import-error-bad-location',
 343                                 $revision->getTitle()->getPrefixedText(),
 344                                 $revision->getID(),
 345                                 $revision->getModel(),
 346                                 $revision->getFormat() );
 347
 348                         return false;
 349                 }
 350
 351                 try {
 352                         return $revision->importOldRevision();
 353                 } catch ( MWContentSerializationException $ex ) {
 354                         $this->notice( 'import-error-unserialize',
 355                                 $revision->getTitle()->getPrefixedText(),
 356                                 $revision->getID(),
 357                                 $revision->getModel(),
 358                                 $revision->getFormat() );
 359                 }
 360
 361                 return false;
 362         }
 363
 364         /**
 365          * Default per-revision callback, performs the import.
 366          * @param WikiRevision $revision
 367          * @return bool
 368          */
 369         public function importLogItem( $revision ) {
 370                 return $revision->importLogItem();
 371         }
 372
 373         /**
 374          * Dummy for now...
 375          * @param WikiRevision $revision
 376          * @return bool
 377          */
 378         public function importUpload( $revision ) {
 379                 return $revision->importUpload();
 380         }
 381
 382         /**
 383          * Mostly for hook use
 384          * @param Title $title
 385          * @param ForeignTitle $foreignTitle
 386          * @param int $revCount
 387          * @param int $sRevCount
 388          * @param array $pageInfo
 389          * @return bool
 390          */
 391         public function finishImportPage( $title, $foreignTitle, $revCount,
 392                 $sRevCount, $pageInfo
 393         ) {
 394                 // Update article count statistics (T42009)
 395                 // The normal counting logic in WikiPage->doEditUpdates() is designed for
 396                 // one-revision-at-a-time editing, not bulk imports. In this situation it
 397                 // suffers from issues of replica DB lag. We let WikiPage handle the total page
 398                 // and revision count, and we implement our own custom logic for the
 399                 // article (content page) count.
 400                 if ( !$this->disableStatisticsUpdate ) {
 401                         $page = WikiPage::factory( $title );
 402                         $page->loadPageData( 'fromdbmaster' );
 403                         $content = $page->getContent();
 404                         if ( $content === null ) {
 405                                 wfDebug( __METHOD__ . ': Skipping article count adjustment for ' . $title .
 406                                         ' because WikiPage::getContent() returned null' );
 407                         } else {
 408                                 $editInfo = $page->prepareContentForEdit( $content );
 409                                 $countKey = 'title_' . $title->getPrefixedText();
 410                                 $countable = $page->isCountable( $editInfo );
 411                                 if ( array_key_exists( $countKey, $this->countableCache ) &&
 412                                         $countable != $this->countableCache[$countKey] ) {
 413                                         DeferredUpdates::addUpdate( SiteStatsUpdate::factory( [
 414                                                 'articles' => ( (int)$countable - (int)$this->countableCache[$countKey] )
 415                                         ] ) );
 416                                 }
 417                         }
 418                 }
 419
 420                 $args = func_get_args();
 421                 return Hooks::run( 'AfterImportPage', $args );
 422         }
 423
 424         /**
 425          * Alternate per-revision callback, for debugging.
 426          * @param WikiRevision &$revision
 427          */
 428         public function debugRevisionHandler( &$revision ) {
 429                 $this->debug( "Got revision:" );
 430                 if ( is_object( $revision->title ) ) {
 431                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 432                 } else {
 433                         $this->debug( "-- Title: <invalid>" );
 434                 }
 435                 $this->debug( "-- User: " . $revision->user_text );
 436                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 437                 $this->debug( "-- Comment: " . $revision->comment );
 438                 $this->debug( "-- Text: " . $revision->text );
 439         }
 440
 441         /**
 442          * Notify the callback function of site info
 443          * @param array $siteInfo
 444          * @return bool|mixed
 445          */
 446         private function siteInfoCallback( $siteInfo ) {
 447                 if ( isset( $this->mSiteInfoCallback ) ) {
 448                         return call_user_func_array( $this->mSiteInfoCallback,
 449                                         [ $siteInfo, $this ] );
 450                 } else {
 451                         return false;
 452                 }
 453         }
 454
 455         /**
 456          * Notify the callback function when a new "<page>" is reached.
 457          * @param Title $title
 458          */
 459         function pageCallback( $title ) {
 460                 if ( isset( $this->mPageCallback ) ) {
 461                         call_user_func( $this->mPageCallback, $title );
 462                 }
 463         }
 464
 465         /**
 466          * Notify the callback function when a "</page>" is closed.
 467          * @param Title $title
 468          * @param ForeignTitle $foreignTitle
 469          * @param int $revCount
 470          * @param int $sucCount Number of revisions for which callback returned true
 471          * @param array $pageInfo Associative array of page information
 472          */
 473         private function pageOutCallback( $title, $foreignTitle, $revCount,
 474                         $sucCount, $pageInfo ) {
 475                 if ( isset( $this->mPageOutCallback ) ) {
 476                         $args = func_get_args();
 477                         call_user_func_array( $this->mPageOutCallback, $args );
 478                 }
 479         }
 480
 481         /**
 482          * Notify the callback function of a revision
 483          * @param WikiRevision $revision
 484          * @return bool|mixed
 485          */
 486         private function revisionCallback( $revision ) {
 487                 if ( isset( $this->mRevisionCallback ) ) {
 488                         return call_user_func_array( $this->mRevisionCallback,
 489                                         [ $revision, $this ] );
 490                 } else {
 491                         return false;
 492                 }
 493         }
 494
 495         /**
 496          * Notify the callback function of a new log item
 497          * @param WikiRevision $revision
 498          * @return bool|mixed
 499          */
 500         private function logItemCallback( $revision ) {
 501                 if ( isset( $this->mLogItemCallback ) ) {
 502                         return call_user_func_array( $this->mLogItemCallback,
 503                                         [ $revision, $this ] );
 504                 } else {
 505                         return false;
 506                 }
 507         }
 508
 509         /**
 510          * Retrieves the contents of the named attribute of the current element.
 511          * @param string $attr The name of the attribute
 512          * @return string The value of the attribute or an empty string if it is not set in the current
 513          * element.
 514          */
 515         public function nodeAttribute( $attr ) {
 516                 return $this->reader->getAttribute( $attr );
 517         }
 518
 519         /**
 520          * Shouldn't something like this be built-in to XMLReader?
 521          * Fetches text contents of the current element, assuming
 522          * no sub-elements or such scary things.
 523          * @return string
 524          * @access private
 525          */
 526         public function nodeContents() {
 527                 if ( $this->reader->isEmptyElement ) {
 528                         return "";
 529                 }
 530                 $buffer = "";
 531                 while ( $this->reader->read() ) {
 532                         switch ( $this->reader->nodeType ) {
 533                         case XMLReader::TEXT:
 534                         case XMLReader::CDATA:
 535                         case XMLReader::SIGNIFICANT_WHITESPACE:
 536                                 $buffer .= $this->reader->value;
 537                                 break;
 538                         case XMLReader::END_ELEMENT:
 539                                 return $buffer;
 540                         }
 541                 }
 542
 543                 $this->reader->close();
 544                 return '';
 545         }
 546
 547         /**
 548          * Primary entry point
 549          * @throws MWException
 550          * @return bool
 551          */
 552         public function doImport() {
 553                 // Calls to reader->read need to be wrapped in calls to
 554                 // libxml_disable_entity_loader() to avoid local file
 555                 // inclusion attacks (T48932).
 556                 $oldDisable = libxml_disable_entity_loader( true );
 557                 $this->reader->read();
 558
 559                 if ( $this->reader->localName != 'mediawiki' ) {
 560                         libxml_disable_entity_loader( $oldDisable );
 561                         throw new MWException( "Expected <mediawiki> tag, got " .
 562                                 $this->reader->localName );
 563                 }
 564                 $this->debug( "<mediawiki> tag is correct." );
 565
 566                 $this->debug( "Starting primary dump processing loop." );
 567
 568                 $keepReading = $this->reader->read();
 569                 $skip = false;
 570                 $rethrow = null;
 571                 $pageCount = 0;
 572                 try {
 573                         while ( $keepReading ) {
 574                                 $tag = $this->reader->localName;
 575                                 if ( $this->pageOffset ) {
 576                                         if ( $tag === 'page' ) {
 577                                                 $pageCount++;
 578                                         }
 579                                         if ( $pageCount < $this->pageOffset ) {
 580                                                 $keepReading = $this->reader->next();
 581                                                 continue;
 582                                         }
 583                                 }
 584                                 $type = $this->reader->nodeType;
 585
 586                                 if ( !Hooks::run( 'ImportHandleToplevelXMLTag', [ $this ] ) ) {
 587                                         // Do nothing
 588                                 } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) {
 589                                         break;
 590                                 } elseif ( $tag == 'siteinfo' ) {
 591                                         $this->handleSiteInfo();
 592                                 } elseif ( $tag == 'page' ) {
 593                                         $this->handlePage();
 594                                 } elseif ( $tag == 'logitem' ) {
 595                                         $this->handleLogItem();
 596                                 } elseif ( $tag != '#text' ) {
 597                                         $this->warn( "Unhandled top-level XML tag $tag" );
 598
 599                                         $skip = true;
 600                                 }
 601
 602                                 if ( $skip ) {
 603                                         $keepReading = $this->reader->next();
 604                                         $skip = false;
 605                                         $this->debug( "Skip" );
 606                                 } else {
 607                                         $keepReading = $this->reader->read();
 608                                 }
 609                         }
 610                 } catch ( Exception $ex ) {
 611                         $rethrow = $ex;
 612                 }
 613
 614                 // finally
 615                 libxml_disable_entity_loader( $oldDisable );
 616                 $this->reader->close();
 617
 618                 if ( $rethrow ) {
 619                         throw $rethrow;
 620                 }
 621
 622                 return true;
 623         }
 624
 625         private function handleSiteInfo() {
 626                 $this->debug( "Enter site info handler." );
 627                 $siteInfo = [];
 628
 629                 // Fields that can just be stuffed in the siteInfo object
 630                 $normalFields = [ 'sitename', 'base', 'generator', 'case' ];
 631
 632                 while ( $this->reader->read() ) {
 633                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 634                                         $this->reader->localName == 'siteinfo' ) {
 635                                 break;
 636                         }
 637
 638                         $tag = $this->reader->localName;
 639
 640                         if ( $tag == 'namespace' ) {
 641                                 $this->foreignNamespaces[$this->nodeAttribute( 'key' )] =
 642                                         $this->nodeContents();
 643                         } elseif ( in_array( $tag, $normalFields ) ) {
 644                                 $siteInfo[$tag] = $this->nodeContents();
 645                         }
 646                 }
 647
 648                 $siteInfo['_namespaces'] = $this->foreignNamespaces;
 649                 $this->siteInfoCallback( $siteInfo );
 650         }
 651
 652         private function handleLogItem() {
 653                 $this->debug( "Enter log item handler." );
 654                 $logInfo = [];
 655
 656                 // Fields that can just be stuffed in the pageInfo object
 657                 $normalFields = [ 'id', 'comment', 'type', 'action', 'timestamp',
 658                                         'logtitle', 'params' ];
 659
 660                 while ( $this->reader->read() ) {
 661                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 662                                         $this->reader->localName == 'logitem' ) {
 663                                 break;
 664                         }
 665
 666                         $tag = $this->reader->localName;
 667
 668                         if ( !Hooks::run( 'ImportHandleLogItemXMLTag', [
 669                                 $this, $logInfo
 670                         ] ) ) {
 671                                 // Do nothing
 672                         } elseif ( in_array( $tag, $normalFields ) ) {
 673                                 $logInfo[$tag] = $this->nodeContents();
 674                         } elseif ( $tag == 'contributor' ) {
 675                                 $logInfo['contributor'] = $this->handleContributor();
 676                         } elseif ( $tag != '#text' ) {
 677                                 $this->warn( "Unhandled log-item XML tag $tag" );
 678                         }
 679                 }
 680
 681                 $this->processLogItem( $logInfo );
 682         }
 683
 684         /**
 685          * @param array $logInfo
 686          * @return bool|mixed
 687          */
 688         private function processLogItem( $logInfo ) {
 689                 $revision = new WikiRevision( $this->config );
 690
 691                 if ( isset( $logInfo['id'] ) ) {
 692                         $revision->setID( $logInfo['id'] );
 693                 }
 694                 $revision->setType( $logInfo['type'] );
 695                 $revision->setAction( $logInfo['action'] );
 696                 if ( isset( $logInfo['timestamp'] ) ) {
 697                         $revision->setTimestamp( $logInfo['timestamp'] );
 698                 }
 699                 if ( isset( $logInfo['params'] ) ) {
 700                         $revision->setParams( $logInfo['params'] );
 701                 }
 702                 if ( isset( $logInfo['logtitle'] ) ) {
 703                         // @todo Using Title for non-local titles is a recipe for disaster.
 704                         // We should use ForeignTitle here instead.
 705                         $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 706                 }
 707
 708                 $revision->setNoUpdates( $this->mNoUpdates );
 709
 710                 if ( isset( $logInfo['comment'] ) ) {
 711                         $revision->setComment( $logInfo['comment'] );
 712                 }
 713
 714                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 715                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 716                 }
 717
 718                 if ( !isset( $logInfo['contributor']['username'] ) ) {
 719                         $revision->setUsername( 'Unknown user' );
 720                 } else {
 721                         $revision->setUsername( $logInfo['contributor']['username'] );
 722                 }
 723
 724                 return $this->logItemCallback( $revision );
 725         }
 726
 727         private function handlePage() {
 728                 // Handle page data.
 729                 $this->debug( "Enter page handler." );
 730                 $pageInfo = [ 'revisionCount' => 0, 'successfulRevisionCount' => 0 ];
 731
 732                 // Fields that can just be stuffed in the pageInfo object
 733                 $normalFields = [ 'title', 'ns', 'id', 'redirect', 'restrictions' ];
 734
 735                 $skip = false;
 736                 $badTitle = false;
 737
 738                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 739                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 740                                         $this->reader->localName == 'page' ) {
 741                                 break;
 742                         }
 743
 744                         $skip = false;
 745
 746                         $tag = $this->reader->localName;
 747
 748                         if ( $badTitle ) {
 749                                 // The title is invalid, bail out of this page
 750                                 $skip = true;
 751                         } elseif ( !Hooks::run( 'ImportHandlePageXMLTag', [ $this,
 752                                                 &$pageInfo ] ) ) {
 753                                 // Do nothing
 754                         } elseif ( in_array( $tag, $normalFields ) ) {
 755                                 // An XML snippet:
 756                                 // <page>
 757                                 //     <id>123</id>
 758                                 //     <title>Page</title>
 759                                 //     <redirect title="NewTitle"/>
 760                                 //     ...
 761                                 // Because the redirect tag is built differently, we need special handling for that case.
 762                                 if ( $tag == 'redirect' ) {
 763                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 764                                 } else {
 765                                         $pageInfo[$tag] = $this->nodeContents();
 766                                 }
 767                         } elseif ( $tag == 'revision' || $tag == 'upload' ) {
 768                                 if ( !isset( $title ) ) {
 769                                         $title = $this->processTitle( $pageInfo['title'],
 770                                                 isset( $pageInfo['ns'] ) ? $pageInfo['ns'] : null );
 771
 772                                         // $title is either an array of two titles or false.
 773                                         if ( is_array( $title ) ) {
 774                                                 $this->pageCallback( $title );
 775                                                 list( $pageInfo['_title'], $foreignTitle ) = $title;
 776                                         } else {
 777                                                 $badTitle = true;
 778                                                 $skip = true;
 779                                         }
 780                                 }
 781
 782                                 if ( $title ) {
 783                                         if ( $tag == 'revision' ) {
 784                                                 $this->handleRevision( $pageInfo );
 785                                         } else {
 786                                                 $this->handleUpload( $pageInfo );
 787                                         }
 788                                 }
 789                         } elseif ( $tag != '#text' ) {
 790                                 $this->warn( "Unhandled page XML tag $tag" );
 791                                 $skip = true;
 792                         }
 793                 }
 794
 795                 // @note $pageInfo is only set if a valid $title is processed above with
 796                 //       no error. If we have a valid $title, then pageCallback is called
 797                 //       above, $pageInfo['title'] is set and we do pageOutCallback here.
 798                 //       If $pageInfo['_title'] is not set, then $foreignTitle is also not
 799                 //       set since they both come from $title above.
 800                 if ( array_key_exists( '_title', $pageInfo ) ) {
 801                         $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
 802                                         $pageInfo['revisionCount'],
 803                                         $pageInfo['successfulRevisionCount'],
 804                                         $pageInfo );
 805                 }
 806         }
 807
 808         /**
 809          * @param array $pageInfo
 810          */
 811         private function handleRevision( &$pageInfo ) {
 812                 $this->debug( "Enter revision handler" );
 813                 $revisionInfo = [];
 814
 815                 $normalFields = [ 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text', 'sha1' ];
 816
 817                 $skip = false;
 818
 819                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 820                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 821                                         $this->reader->localName == 'revision' ) {
 822                                 break;
 823                         }
 824
 825                         $tag = $this->reader->localName;
 826
 827                         if ( !Hooks::run( 'ImportHandleRevisionXMLTag', [
 828                                 $this, $pageInfo, $revisionInfo
 829                         ] ) ) {
 830                                 // Do nothing
 831                         } elseif ( in_array( $tag, $normalFields ) ) {
 832                                 $revisionInfo[$tag] = $this->nodeContents();
 833                         } elseif ( $tag == 'contributor' ) {
 834                                 $revisionInfo['contributor'] = $this->handleContributor();
 835                         } elseif ( $tag != '#text' ) {
 836                                 $this->warn( "Unhandled revision XML tag $tag" );
 837                                 $skip = true;
 838                         }
 839                 }
 840
 841                 $pageInfo['revisionCount']++;
 842                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 843                         $pageInfo['successfulRevisionCount']++;
 844                 }
 845         }
 846
 847         /**
 848          * @param array $pageInfo
 849          * @param array $revisionInfo
 850          * @return bool|mixed
 851          */
 852         private function processRevision( $pageInfo, $revisionInfo ) {
 853                 global $wgMaxArticleSize;
 854
 855                 // Make sure revisions won't violate $wgMaxArticleSize, which could lead to
 856                 // database errors and instability. Testing for revisions with only listed
 857                 // content models, as other content models might use serialization formats
 858                 // which aren't checked against $wgMaxArticleSize.
 859                 if ( ( !isset( $revisionInfo['model'] ) ||
 860                         in_array( $revisionInfo['model'], [
 861                                 'wikitext',
 862                                 'css',
 863                                 'json',
 864                                 'javascript',
 865                                 'text',
 866                                 ''
 867                         ] ) ) &&
 868                         strlen( $revisionInfo['text'] ) > $wgMaxArticleSize * 1024
 869                 ) {
 870                         throw new MWException( 'The text of ' .
 871                                 ( isset( $revisionInfo['id'] ) ?
 872                                         "the revision with ID $revisionInfo[id]" :
 873                                         'a revision'
 874                                 ) . " exceeds the maximum allowable size ($wgMaxArticleSize KB)" );
 875                 }
 876
 877                 $revision = new WikiRevision( $this->config );
 878
 879                 if ( isset( $revisionInfo['id'] ) ) {
 880                         $revision->setID( $revisionInfo['id'] );
 881                 }
 882                 if ( isset( $revisionInfo['model'] ) ) {
 883                         $revision->setModel( $revisionInfo['model'] );
 884                 }
 885                 if ( isset( $revisionInfo['format'] ) ) {
 886                         $revision->setFormat( $revisionInfo['format'] );
 887                 }
 888                 $revision->setTitle( $pageInfo['_title'] );
 889
 890                 if ( isset( $revisionInfo['text'] ) ) {
 891                         $handler = $revision->getContentHandler();
 892                         $text = $handler->importTransform(
 893                                 $revisionInfo['text'],
 894                                 $revision->getFormat() );
 895
 896                         $revision->setText( $text );
 897                 }
 898                 if ( isset( $revisionInfo['timestamp'] ) ) {
 899                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 900                 } else {
 901                         $revision->setTimestamp( wfTimestampNow() );
 902                 }
 903
 904                 if ( isset( $revisionInfo['comment'] ) ) {
 905                         $revision->setComment( $revisionInfo['comment'] );
 906                 }
 907
 908                 if ( isset( $revisionInfo['minor'] ) ) {
 909                         $revision->setMinor( true );
 910                 }
 911                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 912                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 913                 } elseif ( isset( $revisionInfo['contributor']['username'] ) ) {
 914                         $revision->setUsername( $revisionInfo['contributor']['username'] );
 915                 } else {
 916                         $revision->setUsername( 'Unknown user' );
 917                 }
 918                 if ( isset( $revisionInfo['sha1'] ) ) {
 919                         $revision->setSha1Base36( $revisionInfo['sha1'] );
 920                 }
 921                 $revision->setNoUpdates( $this->mNoUpdates );
 922
 923                 return $this->revisionCallback( $revision );
 924         }
 925
 926         /**
 927          * @param array $pageInfo
 928          * @return mixed
 929          */
 930         private function handleUpload( &$pageInfo ) {
 931                 $this->debug( "Enter upload handler" );
 932                 $uploadInfo = [];
 933
 934                 $normalFields = [ 'timestamp', 'comment', 'filename', 'text',
 935                                         'src', 'size', 'sha1base36', 'archivename', 'rel' ];
 936
 937                 $skip = false;
 938
 939                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 940                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 941                                         $this->reader->localName == 'upload' ) {
 942                                 break;
 943                         }
 944
 945                         $tag = $this->reader->localName;
 946
 947                         if ( !Hooks::run( 'ImportHandleUploadXMLTag', [
 948                                 $this, $pageInfo
 949                         ] ) ) {
 950                                 // Do nothing
 951                         } elseif ( in_array( $tag, $normalFields ) ) {
 952                                 $uploadInfo[$tag] = $this->nodeContents();
 953                         } elseif ( $tag == 'contributor' ) {
 954                                 $uploadInfo['contributor'] = $this->handleContributor();
 955                         } elseif ( $tag == 'contents' ) {
 956                                 $contents = $this->nodeContents();
 957                                 $encoding = $this->reader->getAttribute( 'encoding' );
 958                                 if ( $encoding === 'base64' ) {
 959                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 960                                         $uploadInfo['isTempSrc'] = true;
 961                                 }
 962                         } elseif ( $tag != '#text' ) {
 963                                 $this->warn( "Unhandled upload XML tag $tag" );
 964                                 $skip = true;
 965                         }
 966                 }
 967
 968                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 969                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 970                         if ( file_exists( $path ) ) {
 971                                 $uploadInfo['fileSrc'] = $path;
 972                                 $uploadInfo['isTempSrc'] = false;
 973                         }
 974                 }
 975
 976                 if ( $this->mImportUploads ) {
 977                         return $this->processUpload( $pageInfo, $uploadInfo );
 978                 }
 979         }
 980
 981         /**
 982          * @param string $contents
 983          * @return string
 984          */
 985         private function dumpTemp( $contents ) {
 986                 $filename = tempnam( wfTempDir(), 'importupload' );
 987                 file_put_contents( $filename, $contents );
 988                 return $filename;
 989         }
 990
 991         /**
 992          * @param array $pageInfo
 993          * @param array $uploadInfo
 994          * @return mixed
 995          */
 996         private function processUpload( $pageInfo, $uploadInfo ) {
 997                 $revision = new WikiRevision( $this->config );
 998                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 999
1000                 $revision->setTitle( $pageInfo['_title'] );
1001                 $revision->setID( $pageInfo['id'] );
1002                 $revision->setTimestamp( $uploadInfo['timestamp'] );
1003                 $revision->setText( $text );
1004                 $revision->setFilename( $uploadInfo['filename'] );
1005                 if ( isset( $uploadInfo['archivename'] ) ) {
1006                         $revision->setArchiveName( $uploadInfo['archivename'] );
1007                 }
1008                 $revision->setSrc( $uploadInfo['src'] );
1009                 if ( isset( $uploadInfo['fileSrc'] ) ) {
1010                         $revision->setFileSrc( $uploadInfo['fileSrc'],
1011                                 !empty( $uploadInfo['isTempSrc'] ) );
1012                 }
1013                 if ( isset( $uploadInfo['sha1base36'] ) ) {
1014                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
1015                 }
1016                 $revision->setSize( intval( $uploadInfo['size'] ) );
1017                 $revision->setComment( $uploadInfo['comment'] );
1018
1019                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
1020                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
1021                 }
1022                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
1023                         $revision->setUsername( $uploadInfo['contributor']['username'] );
1024                 }
1025                 $revision->setNoUpdates( $this->mNoUpdates );
1026
1027                 return call_user_func( $this->mUploadCallback, $revision );
1028         }
1029
1030         /**
1031          * @return array
1032          */
1033         private function handleContributor() {
1034                 $fields = [ 'id', 'ip', 'username' ];
1035                 $info = [];
1036
1037                 if ( $this->reader->isEmptyElement ) {
1038                         return $info;
1039                 }
1040                 while ( $this->reader->read() ) {
1041                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
1042                                         $this->reader->localName == 'contributor' ) {
1043                                 break;
1044                         }
1045
1046                         $tag = $this->reader->localName;
1047
1048                         if ( in_array( $tag, $fields ) ) {
1049                                 $info[$tag] = $this->nodeContents();
1050                         }
1051                 }
1052
1053                 return $info;
1054         }
1055
1056         /**
1057          * @param string $text
1058          * @param string|null $ns
1059          * @return array|bool
1060          */
1061         private function processTitle( $text, $ns = null ) {
1062                 if ( is_null( $this->foreignNamespaces ) ) {
1063                         $foreignTitleFactory = new NaiveForeignTitleFactory();
1064                 } else {
1065                         $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
1066                                 $this->foreignNamespaces );
1067                 }
1068
1069                 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
1070                         intval( $ns ) );
1071
1072                 $title = $this->importTitleFactory->createTitleFromForeignTitle(
1073                         $foreignTitle );
1074
1075                 $commandLineMode = $this->config->get( 'CommandLineMode' );
1076                 if ( is_null( $title ) ) {
1077                         # Invalid page title? Ignore the page
1078                         $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
1079                         return false;
1080                 } elseif ( $title->isExternal() ) {
1081                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
1082                         return false;
1083                 } elseif ( !$title->canExist() ) {
1084                         $this->notice( 'import-error-special', $title->getPrefixedText() );
1085                         return false;
1086                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
1087                         # Do not import if the importing wiki user cannot edit this page
1088                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
1089                         return false;
1090                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
1091                         # Do not import if the importing wiki user cannot create this page
1092                         $this->notice( 'import-error-create', $title->getPrefixedText() );
1093                         return false;
1094                 }
1095
1096                 return [ $title, $foreignTitle ];
1097         }
1098 }