includes/Import.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         /** @var Config */
  42         private $config;
  43         /** @var ImportTitleFactory */
  44         private $importTitleFactory;
  45         /** @var array */
  46         private $countableCache = array();
  47
  48         /**
  49          * Creates an ImportXMLReader drawing from the source provided
  50          * @param ImportSource $source
  51          * @param Config $config
  52          * @throws Exception
  53          */
  54         function __construct( ImportSource $source, Config $config = null ) {
  55                 if ( !class_exists( 'XMLReader' ) ) {
  56                         throw new Exception( 'Import requires PHP to have been compiled with libxml support' );
  57                 }
  58
  59                 $this->reader = new XMLReader();
  60                 if ( !$config ) {
  61                         wfDeprecated( __METHOD__ . ' without a Config instance', '1.25' );
  62                         $config = ConfigFactory::getDefaultInstance()->makeConfig( 'main' );
  63                 }
  64                 $this->config = $config;
  65
  66                 if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) {
  67                         stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  68                 }
  69                 $id = UploadSourceAdapter::registerSource( $source );
  70
  71                 // Enable the entity loader, as it is needed for loading external URLs via
  72                 // XMLReader::open (T86036)
  73                 $oldDisable = libxml_disable_entity_loader( false );
  74                 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
  75                         $status = $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  76                 } else {
  77                         $status = $this->reader->open( "uploadsource://$id" );
  78                 }
  79                 if ( !$status ) {
  80                         $error = libxml_get_last_error();
  81                         libxml_disable_entity_loader( $oldDisable );
  82                         throw new MWException( 'Encountered an internal error while initializing WikiImporter object: ' .
  83                                 $error->message );
  84                 }
  85                 libxml_disable_entity_loader( $oldDisable );
  86
  87                 // Default callbacks
  88                 $this->setPageCallback( array( $this, 'beforeImportPage' ) );
  89                 $this->setRevisionCallback( array( $this, "importRevision" ) );
  90                 $this->setUploadCallback( array( $this, 'importUpload' ) );
  91                 $this->setLogItemCallback( array( $this, 'importLogItem' ) );
  92                 $this->setPageOutCallback( array( $this, 'finishImportPage' ) );
  93
  94                 $this->importTitleFactory = new NaiveImportTitleFactory();
  95         }
  96
  97         /**
  98          * @return null|XMLReader
  99          */
 100         public function getReader() {
 101                 return $this->reader;
 102         }
 103
 104         public function throwXmlError( $err ) {
 105                 $this->debug( "FAILURE: $err" );
 106                 wfDebug( "WikiImporter XML error: $err\n" );
 107         }
 108
 109         public function debug( $data ) {
 110                 if ( $this->mDebug ) {
 111                         wfDebug( "IMPORT: $data\n" );
 112                 }
 113         }
 114
 115         public function warn( $data ) {
 116                 wfDebug( "IMPORT: $data\n" );
 117         }
 118
 119         public function notice( $msg /*, $param, ...*/ ) {
 120                 $params = func_get_args();
 121                 array_shift( $params );
 122
 123                 if ( is_callable( $this->mNoticeCallback ) ) {
 124                         call_user_func( $this->mNoticeCallback, $msg, $params );
 125                 } else { # No ImportReporter -> CLI
 126                         echo wfMessage( $msg, $params )->text() . "\n";
 127                 }
 128         }
 129
 130         /**
 131          * Set debug mode...
 132          * @param bool $debug
 133          */
 134         function setDebug( $debug ) {
 135                 $this->mDebug = $debug;
 136         }
 137
 138         /**
 139          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
 140          * @param bool $noupdates
 141          */
 142         function setNoUpdates( $noupdates ) {
 143                 $this->mNoUpdates = $noupdates;
 144         }
 145
 146         /**
 147          * Set a callback that displays notice messages
 148          *
 149          * @param callable $callback
 150          * @return callable
 151          */
 152         public function setNoticeCallback( $callback ) {
 153                 return wfSetVar( $this->mNoticeCallback, $callback );
 154         }
 155
 156         /**
 157          * Sets the action to perform as each new page in the stream is reached.
 158          * @param callable $callback
 159          * @return callable
 160          */
 161         public function setPageCallback( $callback ) {
 162                 $previous = $this->mPageCallback;
 163                 $this->mPageCallback = $callback;
 164                 return $previous;
 165         }
 166
 167         /**
 168          * Sets the action to perform as each page in the stream is completed.
 169          * Callback accepts the page title (as a Title object), a second object
 170          * with the original title form (in case it's been overridden into a
 171          * local namespace), and a count of revisions.
 172          *
 173          * @param callable $callback
 174          * @return callable
 175          */
 176         public function setPageOutCallback( $callback ) {
 177                 $previous = $this->mPageOutCallback;
 178                 $this->mPageOutCallback = $callback;
 179                 return $previous;
 180         }
 181
 182         /**
 183          * Sets the action to perform as each page revision is reached.
 184          * @param callable $callback
 185          * @return callable
 186          */
 187         public function setRevisionCallback( $callback ) {
 188                 $previous = $this->mRevisionCallback;
 189                 $this->mRevisionCallback = $callback;
 190                 return $previous;
 191         }
 192
 193         /**
 194          * Sets the action to perform as each file upload version is reached.
 195          * @param callable $callback
 196          * @return callable
 197          */
 198         public function setUploadCallback( $callback ) {
 199                 $previous = $this->mUploadCallback;
 200                 $this->mUploadCallback = $callback;
 201                 return $previous;
 202         }
 203
 204         /**
 205          * Sets the action to perform as each log item reached.
 206          * @param callable $callback
 207          * @return callable
 208          */
 209         public function setLogItemCallback( $callback ) {
 210                 $previous = $this->mLogItemCallback;
 211                 $this->mLogItemCallback = $callback;
 212                 return $previous;
 213         }
 214
 215         /**
 216          * Sets the action to perform when site info is encountered
 217          * @param callable $callback
 218          * @return callable
 219          */
 220         public function setSiteInfoCallback( $callback ) {
 221                 $previous = $this->mSiteInfoCallback;
 222                 $this->mSiteInfoCallback = $callback;
 223                 return $previous;
 224         }
 225
 226         /**
 227          * Sets the factory object to use to convert ForeignTitle objects into local
 228          * Title objects
 229          * @param ImportTitleFactory $factory
 230          */
 231         public function setImportTitleFactory( $factory ) {
 232                 $this->importTitleFactory = $factory;
 233         }
 234
 235         /**
 236          * Set a target namespace to override the defaults
 237          * @param null|int $namespace
 238          * @return bool
 239          */
 240         public function setTargetNamespace( $namespace ) {
 241                 if ( is_null( $namespace ) ) {
 242                         // Don't override namespaces
 243                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 244                         return true;
 245                 } elseif (
 246                         $namespace >= 0 &&
 247                         MWNamespace::exists( intval( $namespace ) )
 248                 ) {
 249                         $namespace = intval( $namespace );
 250                         $this->setImportTitleFactory( new NamespaceImportTitleFactory( $namespace ) );
 251                         return true;
 252                 } else {
 253                         return false;
 254                 }
 255         }
 256
 257         /**
 258          * Set a target root page under which all pages are imported
 259          * @param null|string $rootpage
 260          * @return Status
 261          */
 262         public function setTargetRootPage( $rootpage ) {
 263                 $status = Status::newGood();
 264                 if ( is_null( $rootpage ) ) {
 265                         // No rootpage
 266                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 267                 } elseif ( $rootpage !== '' ) {
 268                         $rootpage = rtrim( $rootpage, '/' ); //avoid double slashes
 269                         $title = Title::newFromText( $rootpage );
 270
 271                         if ( !$title || $title->isExternal() ) {
 272                                 $status->fatal( 'import-rootpage-invalid' );
 273                         } else {
 274                                 if ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 275                                         global $wgContLang;
 276
 277                                         $displayNSText = $title->getNamespace() == NS_MAIN
 278                                                 ? wfMessage( 'blanknamespace' )->text()
 279                                                 : $wgContLang->getNsText( $title->getNamespace() );
 280                                         $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 281                                 } else {
 282                                         // set namespace to 'all', so the namespace check in processTitle() can pass
 283                                         $this->setTargetNamespace( null );
 284                                         $this->setImportTitleFactory( new SubpageImportTitleFactory( $title ) );
 285                                 }
 286                         }
 287                 }
 288                 return $status;
 289         }
 290
 291         /**
 292          * @param string $dir
 293          */
 294         public function setImageBasePath( $dir ) {
 295                 $this->mImageBasePath = $dir;
 296         }
 297
 298         /**
 299          * @param bool $import
 300          */
 301         public function setImportUploads( $import ) {
 302                 $this->mImportUploads = $import;
 303         }
 304
 305         /**
 306          * Default per-page callback. Sets up some things related to site statistics
 307          * @param array $titleAndForeignTitle Two-element array, with Title object at
 308          * index 0 and ForeignTitle object at index 1
 309          * @return bool
 310          */
 311         public function beforeImportPage( $titleAndForeignTitle ) {
 312                 $title = $titleAndForeignTitle[0];
 313                 $page = WikiPage::factory( $title );
 314                 $this->countableCache['title_' . $title->getPrefixedText()] = $page->isCountable();
 315                 return true;
 316         }
 317
 318         /**
 319          * Default per-revision callback, performs the import.
 320          * @param WikiRevision $revision
 321          * @return bool
 322          */
 323         public function importRevision( $revision ) {
 324                 if ( !$revision->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
 325                         $this->notice( 'import-error-bad-location',
 326                                 $revision->getTitle()->getPrefixedText(),
 327                                 $revision->getID(),
 328                                 $revision->getModel(),
 329                                 $revision->getFormat() );
 330
 331                         return false;
 332                 }
 333
 334                 try {
 335                         $dbw = wfGetDB( DB_MASTER );
 336                         return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
 337                 } catch ( MWContentSerializationException $ex ) {
 338                         $this->notice( 'import-error-unserialize',
 339                                 $revision->getTitle()->getPrefixedText(),
 340                                 $revision->getID(),
 341                                 $revision->getModel(),
 342                                 $revision->getFormat() );
 343                 }
 344
 345                 return false;
 346         }
 347
 348         /**
 349          * Default per-revision callback, performs the import.
 350          * @param WikiRevision $revision
 351          * @return bool
 352          */
 353         public function importLogItem( $revision ) {
 354                 $dbw = wfGetDB( DB_MASTER );
 355                 return $dbw->deadlockLoop( array( $revision, 'importLogItem' ) );
 356         }
 357
 358         /**
 359          * Dummy for now...
 360          * @param WikiRevision $revision
 361          * @return bool
 362          */
 363         public function importUpload( $revision ) {
 364                 $dbw = wfGetDB( DB_MASTER );
 365                 return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
 366         }
 367
 368         /**
 369          * Mostly for hook use
 370          * @param Title $title
 371          * @param ForeignTitle $foreignTitle
 372          * @param int $revCount
 373          * @param int $sRevCount
 374          * @param array $pageInfo
 375          * @return bool
 376          */
 377         public function finishImportPage( $title, $foreignTitle, $revCount,
 378                         $sRevCount, $pageInfo ) {
 379
 380                 // Update article count statistics (T42009)
 381                 // The normal counting logic in WikiPage->doEditUpdates() is designed for
 382                 // one-revision-at-a-time editing, not bulk imports. In this situation it
 383                 // suffers from issues of slave lag. We let WikiPage handle the total page
 384                 // and revision count, and we implement our own custom logic for the
 385                 // article (content page) count.
 386                 $page = WikiPage::factory( $title );
 387                 $page->loadPageData( 'fromdbmaster' );
 388                 $content = $page->getContent();
 389                 if ( $content === null ) {
 390                         wfDebug( __METHOD__ . ': Skipping article count adjustment for ' . $title .
 391                                 ' because WikiPage::getContent() returned null' );
 392                 } else {
 393                         $editInfo = $page->prepareContentForEdit( $content );
 394                         $countKey = 'title_' . $title->getPrefixedText();
 395                         $countable = $page->isCountable( $editInfo );
 396                         if ( array_key_exists( $countKey, $this->countableCache ) &&
 397                                 $countable != $this->countableCache[$countKey] ) {
 398                                 DeferredUpdates::addUpdate( SiteStatsUpdate::factory( array(
 399                                         'articles' => ( (int)$countable - (int)$this->countableCache[$countKey] )
 400                                 ) ) );
 401                         }
 402                 }
 403
 404                 $args = func_get_args();
 405                 return Hooks::run( 'AfterImportPage', $args );
 406         }
 407
 408         /**
 409          * Alternate per-revision callback, for debugging.
 410          * @param WikiRevision $revision
 411          */
 412         public function debugRevisionHandler( &$revision ) {
 413                 $this->debug( "Got revision:" );
 414                 if ( is_object( $revision->title ) ) {
 415                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 416                 } else {
 417                         $this->debug( "-- Title: <invalid>" );
 418                 }
 419                 $this->debug( "-- User: " . $revision->user_text );
 420                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 421                 $this->debug( "-- Comment: " . $revision->comment );
 422                 $this->debug( "-- Text: " . $revision->text );
 423         }
 424
 425         /**
 426          * Notify the callback function of site info
 427          * @param array $siteInfo
 428          * @return bool|mixed
 429          */
 430         private function siteInfoCallback( $siteInfo ) {
 431                 if ( isset( $this->mSiteInfoCallback ) ) {
 432                         return call_user_func_array( $this->mSiteInfoCallback,
 433                                         array( $siteInfo, $this ) );
 434                 } else {
 435                         return false;
 436                 }
 437         }
 438
 439         /**
 440          * Notify the callback function when a new "<page>" is reached.
 441          * @param Title $title
 442          */
 443         function pageCallback( $title ) {
 444                 if ( isset( $this->mPageCallback ) ) {
 445                         call_user_func( $this->mPageCallback, $title );
 446                 }
 447         }
 448
 449         /**
 450          * Notify the callback function when a "</page>" is closed.
 451          * @param Title $title
 452          * @param ForeignTitle $foreignTitle
 453          * @param int $revCount
 454          * @param int $sucCount Number of revisions for which callback returned true
 455          * @param array $pageInfo Associative array of page information
 456          */
 457         private function pageOutCallback( $title, $foreignTitle, $revCount,
 458                         $sucCount, $pageInfo ) {
 459                 if ( isset( $this->mPageOutCallback ) ) {
 460                         $args = func_get_args();
 461                         call_user_func_array( $this->mPageOutCallback, $args );
 462                 }
 463         }
 464
 465         /**
 466          * Notify the callback function of a revision
 467          * @param WikiRevision $revision
 468          * @return bool|mixed
 469          */
 470         private function revisionCallback( $revision ) {
 471                 if ( isset( $this->mRevisionCallback ) ) {
 472                         return call_user_func_array( $this->mRevisionCallback,
 473                                         array( $revision, $this ) );
 474                 } else {
 475                         return false;
 476                 }
 477         }
 478
 479         /**
 480          * Notify the callback function of a new log item
 481          * @param WikiRevision $revision
 482          * @return bool|mixed
 483          */
 484         private function logItemCallback( $revision ) {
 485                 if ( isset( $this->mLogItemCallback ) ) {
 486                         return call_user_func_array( $this->mLogItemCallback,
 487                                         array( $revision, $this ) );
 488                 } else {
 489                         return false;
 490                 }
 491         }
 492
 493         /**
 494          * Retrieves the contents of the named attribute of the current element.
 495          * @param string $attr The name of the attribute
 496          * @return string The value of the attribute or an empty string if it is not set in the current
 497          * element.
 498          */
 499         public function nodeAttribute( $attr ) {
 500                 return $this->reader->getAttribute( $attr );
 501         }
 502
 503         /**
 504          * Shouldn't something like this be built-in to XMLReader?
 505          * Fetches text contents of the current element, assuming
 506          * no sub-elements or such scary things.
 507          * @return string
 508          * @access private
 509          */
 510         public function nodeContents() {
 511                 if ( $this->reader->isEmptyElement ) {
 512                         return "";
 513                 }
 514                 $buffer = "";
 515                 while ( $this->reader->read() ) {
 516                         switch ( $this->reader->nodeType ) {
 517                         case XMLReader::TEXT:
 518                         case XMLReader::SIGNIFICANT_WHITESPACE:
 519                                 $buffer .= $this->reader->value;
 520                                 break;
 521                         case XMLReader::END_ELEMENT:
 522                                 return $buffer;
 523                         }
 524                 }
 525
 526                 $this->reader->close();
 527                 return '';
 528         }
 529
 530         /**
 531          * Primary entry point
 532          * @throws MWException
 533          * @return bool
 534          */
 535         public function doImport() {
 536                 // Calls to reader->read need to be wrapped in calls to
 537                 // libxml_disable_entity_loader() to avoid local file
 538                 // inclusion attacks (bug 46932).
 539                 $oldDisable = libxml_disable_entity_loader( true );
 540                 $this->reader->read();
 541
 542                 if ( $this->reader->localName != 'mediawiki' ) {
 543                         libxml_disable_entity_loader( $oldDisable );
 544                         throw new MWException( "Expected <mediawiki> tag, got " .
 545                                 $this->reader->localName );
 546                 }
 547                 $this->debug( "<mediawiki> tag is correct." );
 548
 549                 $this->debug( "Starting primary dump processing loop." );
 550
 551                 $keepReading = $this->reader->read();
 552                 $skip = false;
 553                 $rethrow = null;
 554                 try {
 555                         while ( $keepReading ) {
 556                                 $tag = $this->reader->localName;
 557                                 $type = $this->reader->nodeType;
 558
 559                                 if ( !Hooks::run( 'ImportHandleToplevelXMLTag', array( $this ) ) ) {
 560                                         // Do nothing
 561                                 } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) {
 562                                         break;
 563                                 } elseif ( $tag == 'siteinfo' ) {
 564                                         $this->handleSiteInfo();
 565                                 } elseif ( $tag == 'page' ) {
 566                                         $this->handlePage();
 567                                 } elseif ( $tag == 'logitem' ) {
 568                                         $this->handleLogItem();
 569                                 } elseif ( $tag != '#text' ) {
 570                                         $this->warn( "Unhandled top-level XML tag $tag" );
 571
 572                                         $skip = true;
 573                                 }
 574
 575                                 if ( $skip ) {
 576                                         $keepReading = $this->reader->next();
 577                                         $skip = false;
 578                                         $this->debug( "Skip" );
 579                                 } else {
 580                                         $keepReading = $this->reader->read();
 581                                 }
 582                         }
 583                 } catch ( Exception $ex ) {
 584                         $rethrow = $ex;
 585                 }
 586
 587                 // finally
 588                 libxml_disable_entity_loader( $oldDisable );
 589                 $this->reader->close();
 590
 591                 if ( $rethrow ) {
 592                         throw $rethrow;
 593                 }
 594
 595                 return true;
 596         }
 597
 598         private function handleSiteInfo() {
 599                 $this->debug( "Enter site info handler." );
 600                 $siteInfo = array();
 601
 602                 // Fields that can just be stuffed in the siteInfo object
 603                 $normalFields = array( 'sitename', 'base', 'generator', 'case' );
 604
 605                 while ( $this->reader->read() ) {
 606                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 607                                         $this->reader->localName == 'siteinfo' ) {
 608                                 break;
 609                         }
 610
 611                         $tag = $this->reader->localName;
 612
 613                         if ( $tag == 'namespace' ) {
 614                                 $this->foreignNamespaces[$this->nodeAttribute( 'key' )] =
 615                                         $this->nodeContents();
 616                         } elseif ( in_array( $tag, $normalFields ) ) {
 617                                 $siteInfo[$tag] = $this->nodeContents();
 618                         }
 619                 }
 620
 621                 $siteInfo['_namespaces'] = $this->foreignNamespaces;
 622                 $this->siteInfoCallback( $siteInfo );
 623         }
 624
 625         private function handleLogItem() {
 626                 $this->debug( "Enter log item handler." );
 627                 $logInfo = array();
 628
 629                 // Fields that can just be stuffed in the pageInfo object
 630                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 631                                         'logtitle', 'params' );
 632
 633                 while ( $this->reader->read() ) {
 634                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 635                                         $this->reader->localName == 'logitem' ) {
 636                                 break;
 637                         }
 638
 639                         $tag = $this->reader->localName;
 640
 641                         if ( !Hooks::run( 'ImportHandleLogItemXMLTag', array(
 642                                 $this, $logInfo
 643                         ) ) ) {
 644                                 // Do nothing
 645                         } elseif ( in_array( $tag, $normalFields ) ) {
 646                                 $logInfo[$tag] = $this->nodeContents();
 647                         } elseif ( $tag == 'contributor' ) {
 648                                 $logInfo['contributor'] = $this->handleContributor();
 649                         } elseif ( $tag != '#text' ) {
 650                                 $this->warn( "Unhandled log-item XML tag $tag" );
 651                         }
 652                 }
 653
 654                 $this->processLogItem( $logInfo );
 655         }
 656
 657         /**
 658          * @param array $logInfo
 659          * @return bool|mixed
 660          */
 661         private function processLogItem( $logInfo ) {
 662                 $revision = new WikiRevision( $this->config );
 663
 664                 $revision->setID( $logInfo['id'] );
 665                 $revision->setType( $logInfo['type'] );
 666                 $revision->setAction( $logInfo['action'] );
 667                 $revision->setTimestamp( $logInfo['timestamp'] );
 668                 $revision->setParams( $logInfo['params'] );
 669                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 670                 $revision->setNoUpdates( $this->mNoUpdates );
 671
 672                 if ( isset( $logInfo['comment'] ) ) {
 673                         $revision->setComment( $logInfo['comment'] );
 674                 }
 675
 676                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 677                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 678                 }
 679                 if ( isset( $logInfo['contributor']['username'] ) ) {
 680                         $revision->setUserName( $logInfo['contributor']['username'] );
 681                 }
 682
 683                 return $this->logItemCallback( $revision );
 684         }
 685
 686         private function handlePage() {
 687                 // Handle page data.
 688                 $this->debug( "Enter page handler." );
 689                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 690
 691                 // Fields that can just be stuffed in the pageInfo object
 692                 $normalFields = array( 'title', 'ns', 'id', 'redirect', 'restrictions' );
 693
 694                 $skip = false;
 695                 $badTitle = false;
 696
 697                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 698                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 699                                         $this->reader->localName == 'page' ) {
 700                                 break;
 701                         }
 702
 703                         $skip = false;
 704
 705                         $tag = $this->reader->localName;
 706
 707                         if ( $badTitle ) {
 708                                 // The title is invalid, bail out of this page
 709                                 $skip = true;
 710                         } elseif ( !Hooks::run( 'ImportHandlePageXMLTag', array( $this,
 711                                                 &$pageInfo ) ) ) {
 712                                 // Do nothing
 713                         } elseif ( in_array( $tag, $normalFields ) ) {
 714                                 // An XML snippet:
 715                                 // <page>
 716                                 //     <id>123</id>
 717                                 //     <title>Page</title>
 718                                 //     <redirect title="NewTitle"/>
 719                                 //     ...
 720                                 // Because the redirect tag is built differently, we need special handling for that case.
 721                                 if ( $tag == 'redirect' ) {
 722                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 723                                 } else {
 724                                         $pageInfo[$tag] = $this->nodeContents();
 725                                 }
 726                         } elseif ( $tag == 'revision' || $tag == 'upload' ) {
 727                                 if ( !isset( $title ) ) {
 728                                         $title = $this->processTitle( $pageInfo['title'],
 729                                                 isset( $pageInfo['ns'] ) ? $pageInfo['ns'] : null );
 730
 731                                         if ( !$title ) {
 732                                                 $badTitle = true;
 733                                                 $skip = true;
 734                                         }
 735
 736                                         $this->pageCallback( $title );
 737                                         list( $pageInfo['_title'], $foreignTitle ) = $title;
 738                                 }
 739
 740                                 if ( $title ) {
 741                                         if ( $tag == 'revision' ) {
 742                                                 $this->handleRevision( $pageInfo );
 743                                         } else {
 744                                                 $this->handleUpload( $pageInfo );
 745                                         }
 746                                 }
 747                         } elseif ( $tag != '#text' ) {
 748                                 $this->warn( "Unhandled page XML tag $tag" );
 749                                 $skip = true;
 750                         }
 751                 }
 752
 753                 $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
 754                                         $pageInfo['revisionCount'],
 755                                         $pageInfo['successfulRevisionCount'],
 756                                         $pageInfo );
 757         }
 758
 759         /**
 760          * @param array $pageInfo
 761          */
 762         private function handleRevision( &$pageInfo ) {
 763                 $this->debug( "Enter revision handler" );
 764                 $revisionInfo = array();
 765
 766                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' );
 767
 768                 $skip = false;
 769
 770                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 771                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 772                                         $this->reader->localName == 'revision' ) {
 773                                 break;
 774                         }
 775
 776                         $tag = $this->reader->localName;
 777
 778                         if ( !Hooks::run( 'ImportHandleRevisionXMLTag', array(
 779                                 $this, $pageInfo, $revisionInfo
 780                         ) ) ) {
 781                                 // Do nothing
 782                         } elseif ( in_array( $tag, $normalFields ) ) {
 783                                 $revisionInfo[$tag] = $this->nodeContents();
 784                         } elseif ( $tag == 'contributor' ) {
 785                                 $revisionInfo['contributor'] = $this->handleContributor();
 786                         } elseif ( $tag != '#text' ) {
 787                                 $this->warn( "Unhandled revision XML tag $tag" );
 788                                 $skip = true;
 789                         }
 790                 }
 791
 792                 $pageInfo['revisionCount']++;
 793                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 794                         $pageInfo['successfulRevisionCount']++;
 795                 }
 796         }
 797
 798         /**
 799          * @param array $pageInfo
 800          * @param array $revisionInfo
 801          * @return bool|mixed
 802          */
 803         private function processRevision( $pageInfo, $revisionInfo ) {
 804                 $revision = new WikiRevision( $this->config );
 805
 806                 if ( isset( $revisionInfo['id'] ) ) {
 807                         $revision->setID( $revisionInfo['id'] );
 808                 }
 809                 if ( isset( $revisionInfo['model'] ) ) {
 810                         $revision->setModel( $revisionInfo['model'] );
 811                 }
 812                 if ( isset( $revisionInfo['format'] ) ) {
 813                         $revision->setFormat( $revisionInfo['format'] );
 814                 }
 815                 $revision->setTitle( $pageInfo['_title'] );
 816
 817                 if ( isset( $revisionInfo['text'] ) ) {
 818                         $handler = $revision->getContentHandler();
 819                         $text = $handler->importTransform(
 820                                 $revisionInfo['text'],
 821                                 $revision->getFormat() );
 822
 823                         $revision->setText( $text );
 824                 }
 825                 if ( isset( $revisionInfo['timestamp'] ) ) {
 826                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 827                 } else {
 828                         $revision->setTimestamp( wfTimestampNow() );
 829                 }
 830
 831                 if ( isset( $revisionInfo['comment'] ) ) {
 832                         $revision->setComment( $revisionInfo['comment'] );
 833                 }
 834
 835                 if ( isset( $revisionInfo['minor'] ) ) {
 836                         $revision->setMinor( true );
 837                 }
 838                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 839                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 840                 }
 841                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 842                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 843                 }
 844                 $revision->setNoUpdates( $this->mNoUpdates );
 845
 846                 return $this->revisionCallback( $revision );
 847         }
 848
 849         /**
 850          * @param array $pageInfo
 851          * @return mixed
 852          */
 853         private function handleUpload( &$pageInfo ) {
 854                 $this->debug( "Enter upload handler" );
 855                 $uploadInfo = array();
 856
 857                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 858                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 859
 860                 $skip = false;
 861
 862                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 863                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 864                                         $this->reader->localName == 'upload' ) {
 865                                 break;
 866                         }
 867
 868                         $tag = $this->reader->localName;
 869
 870                         if ( !Hooks::run( 'ImportHandleUploadXMLTag', array(
 871                                 $this, $pageInfo
 872                         ) ) ) {
 873                                 // Do nothing
 874                         } elseif ( in_array( $tag, $normalFields ) ) {
 875                                 $uploadInfo[$tag] = $this->nodeContents();
 876                         } elseif ( $tag == 'contributor' ) {
 877                                 $uploadInfo['contributor'] = $this->handleContributor();
 878                         } elseif ( $tag == 'contents' ) {
 879                                 $contents = $this->nodeContents();
 880                                 $encoding = $this->reader->getAttribute( 'encoding' );
 881                                 if ( $encoding === 'base64' ) {
 882                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 883                                         $uploadInfo['isTempSrc'] = true;
 884                                 }
 885                         } elseif ( $tag != '#text' ) {
 886                                 $this->warn( "Unhandled upload XML tag $tag" );
 887                                 $skip = true;
 888                         }
 889                 }
 890
 891                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 892                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 893                         if ( file_exists( $path ) ) {
 894                                 $uploadInfo['fileSrc'] = $path;
 895                                 $uploadInfo['isTempSrc'] = false;
 896                         }
 897                 }
 898
 899                 if ( $this->mImportUploads ) {
 900                         return $this->processUpload( $pageInfo, $uploadInfo );
 901                 }
 902         }
 903
 904         /**
 905          * @param string $contents
 906          * @return string
 907          */
 908         private function dumpTemp( $contents ) {
 909                 $filename = tempnam( wfTempDir(), 'importupload' );
 910                 file_put_contents( $filename, $contents );
 911                 return $filename;
 912         }
 913
 914         /**
 915          * @param array $pageInfo
 916          * @param array $uploadInfo
 917          * @return mixed
 918          */
 919         private function processUpload( $pageInfo, $uploadInfo ) {
 920                 $revision = new WikiRevision( $this->config );
 921                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 922
 923                 $revision->setTitle( $pageInfo['_title'] );
 924                 $revision->setID( $pageInfo['id'] );
 925                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 926                 $revision->setText( $text );
 927                 $revision->setFilename( $uploadInfo['filename'] );
 928                 if ( isset( $uploadInfo['archivename'] ) ) {
 929                         $revision->setArchiveName( $uploadInfo['archivename'] );
 930                 }
 931                 $revision->setSrc( $uploadInfo['src'] );
 932                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 933                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 934                                 !empty( $uploadInfo['isTempSrc'] ) );
 935                 }
 936                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 937                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 938                 }
 939                 $revision->setSize( intval( $uploadInfo['size'] ) );
 940                 $revision->setComment( $uploadInfo['comment'] );
 941
 942                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 943                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 944                 }
 945                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 946                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 947                 }
 948                 $revision->setNoUpdates( $this->mNoUpdates );
 949
 950                 return call_user_func( $this->mUploadCallback, $revision );
 951         }
 952
 953         /**
 954          * @return array
 955          */
 956         private function handleContributor() {
 957                 $fields = array( 'id', 'ip', 'username' );
 958                 $info = array();
 959
 960                 while ( $this->reader->read() ) {
 961                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 962                                         $this->reader->localName == 'contributor' ) {
 963                                 break;
 964                         }
 965
 966                         $tag = $this->reader->localName;
 967
 968                         if ( in_array( $tag, $fields ) ) {
 969                                 $info[$tag] = $this->nodeContents();
 970                         }
 971                 }
 972
 973                 return $info;
 974         }
 975
 976         /**
 977          * @param string $text
 978          * @param string|null $ns
 979          * @return array|bool
 980          */
 981         private function processTitle( $text, $ns = null ) {
 982                 if ( is_null( $this->foreignNamespaces ) ) {
 983                         $foreignTitleFactory = new NaiveForeignTitleFactory();
 984                 } else {
 985                         $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
 986                                 $this->foreignNamespaces );
 987                 }
 988
 989                 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
 990                         intval( $ns ) );
 991
 992                 $title = $this->importTitleFactory->createTitleFromForeignTitle(
 993                         $foreignTitle );
 994
 995                 $commandLineMode = $this->config->get( 'CommandLineMode' );
 996                 if ( is_null( $title ) ) {
 997                         # Invalid page title? Ignore the page
 998                         $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
 999                         return false;
1000                 } elseif ( $title->isExternal() ) {
1001                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
1002                         return false;
1003                 } elseif ( !$title->canExist() ) {
1004                         $this->notice( 'import-error-special', $title->getPrefixedText() );
1005                         return false;
1006                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
1007                         # Do not import if the importing wiki user cannot edit this page
1008                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
1009                         return false;
1010                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
1011                         # Do not import if the importing wiki user cannot create this page
1012                         $this->notice( 'import-error-create', $title->getPrefixedText() );
1013                         return false;
1014                 }
1015
1016                 return array( $title, $foreignTitle );
1017         }
1018 }
1019
1020 /** This is a horrible hack used to keep source compatibility */
1021 class UploadSourceAdapter {
1022         /** @var array */
1023         public static $sourceRegistrations = array();
1024
1025         /** @var string */
1026         private $mSource;
1027
1028         /** @var string */
1029         private $mBuffer;
1030
1031         /** @var int */
1032         private $mPosition;
1033
1034         /**
1035          * @param ImportSource $source
1036          * @return string
1037          */
1038         static function registerSource( ImportSource $source ) {
1039                 $id = wfRandomString();
1040
1041                 self::$sourceRegistrations[$id] = $source;
1042
1043                 return $id;
1044         }
1045
1046         /**
1047          * @param string $path
1048          * @param string $mode
1049          * @param array $options
1050          * @param string $opened_path
1051          * @return bool
1052          */
1053         function stream_open( $path, $mode, $options, &$opened_path ) {
1054                 $url = parse_url( $path );
1055                 $id = $url['host'];
1056
1057                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
1058                         return false;
1059                 }
1060
1061                 $this->mSource = self::$sourceRegistrations[$id];
1062
1063                 return true;
1064         }
1065
1066         /**
1067          * @param int $count
1068          * @return string
1069          */
1070         function stream_read( $count ) {
1071                 $return = '';
1072                 $leave = false;
1073
1074                 while ( !$leave && !$this->mSource->atEnd() &&
1075                                 strlen( $this->mBuffer ) < $count ) {
1076                         $read = $this->mSource->readChunk();
1077
1078                         if ( !strlen( $read ) ) {
1079                                 $leave = true;
1080                         }
1081
1082                         $this->mBuffer .= $read;
1083                 }
1084
1085                 if ( strlen( $this->mBuffer ) ) {
1086                         $return = substr( $this->mBuffer, 0, $count );
1087                         $this->mBuffer = substr( $this->mBuffer, $count );
1088                 }
1089
1090                 $this->mPosition += strlen( $return );
1091
1092                 return $return;
1093         }
1094
1095         /**
1096          * @param string $data
1097          * @return bool
1098          */
1099         function stream_write( $data ) {
1100                 return false;
1101         }
1102
1103         /**
1104          * @return mixed
1105          */
1106         function stream_tell() {
1107                 return $this->mPosition;
1108         }
1109
1110         /**
1111          * @return bool
1112          */
1113         function stream_eof() {
1114                 return $this->mSource->atEnd();
1115         }
1116
1117         /**
1118          * @return array
1119          */
1120         function url_stat() {
1121                 $result = array();
1122
1123                 $result['dev'] = $result[0] = 0;
1124                 $result['ino'] = $result[1] = 0;
1125                 $result['mode'] = $result[2] = 0;
1126                 $result['nlink'] = $result[3] = 0;
1127                 $result['uid'] = $result[4] = 0;
1128                 $result['gid'] = $result[5] = 0;
1129                 $result['rdev'] = $result[6] = 0;
1130                 $result['size'] = $result[7] = 0;
1131                 $result['atime'] = $result[8] = 0;
1132                 $result['mtime'] = $result[9] = 0;
1133                 $result['ctime'] = $result[10] = 0;
1134                 $result['blksize'] = $result[11] = 0;
1135                 $result['blocks'] = $result[12] = 0;
1136
1137                 return $result;
1138         }
1139 }
1140
1141 /**
1142  * @todo document (e.g. one-sentence class description).
1143  * @ingroup SpecialPage
1144  */
1145 class WikiRevision {
1146         /** @todo Unused? */
1147         public $importer = null;
1148
1149         /** @var Title */
1150         public $title = null;
1151
1152         /** @var int */
1153         public $id = 0;
1154
1155         /** @var string */
1156         public $timestamp = "20010115000000";
1157
1158         /**
1159          * @var int
1160          * @todo Can't find any uses. Public, because that's suspicious. Get clarity. */
1161         public $user = 0;
1162
1163         /** @var string */
1164         public $user_text = "";
1165
1166         /** @var string */
1167         public $model = null;
1168
1169         /** @var string */
1170         public $format = null;
1171
1172         /** @var string */
1173         public $text = "";
1174
1175         /** @var int */
1176         protected $size;
1177
1178         /** @var Content */
1179         public $content = null;
1180
1181         /** @var ContentHandler */
1182         protected $contentHandler = null;
1183
1184         /** @var string */
1185         public $comment = "";
1186
1187         /** @var bool */
1188         public $minor = false;
1189
1190         /** @var string */
1191         public $type = "";
1192
1193         /** @var string */
1194         public $action = "";
1195
1196         /** @var string */
1197         public $params = "";
1198
1199         /** @var string */
1200         public $fileSrc = '';
1201
1202         /** @var bool|string */
1203         public $sha1base36 = false;
1204
1205         /**
1206          * @var bool
1207          * @todo Unused?
1208          */
1209         public $isTemp = false;
1210
1211         /** @var string */
1212         public $archiveName = '';
1213
1214         protected $filename;
1215
1216         /** @var mixed */
1217         protected $src;
1218
1219         /** @todo Unused? */
1220         public $fileIsTemp;
1221
1222         /** @var bool */
1223         private $mNoUpdates = false;
1224
1225         /** @var Config $config */
1226         private $config;
1227
1228         public function __construct( Config $config ) {
1229                 $this->config = $config;
1230         }
1231
1232         /**
1233          * @param Title $title
1234          * @throws MWException
1235          */
1236         function setTitle( $title ) {
1237                 if ( is_object( $title ) ) {
1238                         $this->title = $title;
1239                 } elseif ( is_null( $title ) ) {
1240                         throw new MWException( "WikiRevision given a null title in import. "
1241                                 . "You may need to adjust \$wgLegalTitleChars." );
1242                 } else {
1243                         throw new MWException( "WikiRevision given non-object title in import." );
1244                 }
1245         }
1246
1247         /**
1248          * @param int $id
1249          */
1250         function setID( $id ) {
1251                 $this->id = $id;
1252         }
1253
1254         /**
1255          * @param string $ts
1256          */
1257         function setTimestamp( $ts ) {
1258                 # 2003-08-05T18:30:02Z
1259                 $this->timestamp = wfTimestamp( TS_MW, $ts );
1260         }
1261
1262         /**
1263          * @param string $user
1264          */
1265         function setUsername( $user ) {
1266                 $this->user_text = $user;
1267         }
1268
1269         /**
1270          * @param string $ip
1271          */
1272         function setUserIP( $ip ) {
1273                 $this->user_text = $ip;
1274         }
1275
1276         /**
1277          * @param string $model
1278          */
1279         function setModel( $model ) {
1280                 $this->model = $model;
1281         }
1282
1283         /**
1284          * @param string $format
1285          */
1286         function setFormat( $format ) {
1287                 $this->format = $format;
1288         }
1289
1290         /**
1291          * @param string $text
1292          */
1293         function setText( $text ) {
1294                 $this->text = $text;
1295         }
1296
1297         /**
1298          * @param string $text
1299          */
1300         function setComment( $text ) {
1301                 $this->comment = $text;
1302         }
1303
1304         /**
1305          * @param bool $minor
1306          */
1307         function setMinor( $minor ) {
1308                 $this->minor = (bool)$minor;
1309         }
1310
1311         /**
1312          * @param mixed $src
1313          */
1314         function setSrc( $src ) {
1315                 $this->src = $src;
1316         }
1317
1318         /**
1319          * @param string $src
1320          * @param bool $isTemp
1321          */
1322         function setFileSrc( $src, $isTemp ) {
1323                 $this->fileSrc = $src;
1324                 $this->fileIsTemp = $isTemp;
1325         }
1326
1327         /**
1328          * @param string $sha1base36
1329          */
1330         function setSha1Base36( $sha1base36 ) {
1331                 $this->sha1base36 = $sha1base36;
1332         }
1333
1334         /**
1335          * @param string $filename
1336          */
1337         function setFilename( $filename ) {
1338                 $this->filename = $filename;
1339         }
1340
1341         /**
1342          * @param string $archiveName
1343          */
1344         function setArchiveName( $archiveName ) {
1345                 $this->archiveName = $archiveName;
1346         }
1347
1348         /**
1349          * @param int $size
1350          */
1351         function setSize( $size ) {
1352                 $this->size = intval( $size );
1353         }
1354
1355         /**
1356          * @param string $type
1357          */
1358         function setType( $type ) {
1359                 $this->type = $type;
1360         }
1361
1362         /**
1363          * @param string $action
1364          */
1365         function setAction( $action ) {
1366                 $this->action = $action;
1367         }
1368
1369         /**
1370          * @param array $params
1371          */
1372         function setParams( $params ) {
1373                 $this->params = $params;
1374         }
1375
1376         /**
1377          * @param bool $noupdates
1378          */
1379         public function setNoUpdates( $noupdates ) {
1380                 $this->mNoUpdates = $noupdates;
1381         }
1382
1383         /**
1384          * @return Title
1385          */
1386         function getTitle() {
1387                 return $this->title;
1388         }
1389
1390         /**
1391          * @return int
1392          */
1393         function getID() {
1394                 return $this->id;
1395         }
1396
1397         /**
1398          * @return string
1399          */
1400         function getTimestamp() {
1401                 return $this->timestamp;
1402         }
1403
1404         /**
1405          * @return string
1406          */
1407         function getUser() {
1408                 return $this->user_text;
1409         }
1410
1411         /**
1412          * @return string
1413          *
1414          * @deprecated Since 1.21, use getContent() instead.
1415          */
1416         function getText() {
1417                 ContentHandler::deprecated( __METHOD__, '1.21' );
1418
1419                 return $this->text;
1420         }
1421
1422         /**
1423          * @return ContentHandler
1424          */
1425         function getContentHandler() {
1426                 if ( is_null( $this->contentHandler ) ) {
1427                         $this->contentHandler = ContentHandler::getForModelID( $this->getModel() );
1428                 }
1429
1430                 return $this->contentHandler;
1431         }
1432
1433         /**
1434          * @return Content
1435          */
1436         function getContent() {
1437                 if ( is_null( $this->content ) ) {
1438                         $handler = $this->getContentHandler();
1439                         $this->content = $handler->unserializeContent( $this->text, $this->getFormat() );
1440                 }
1441
1442                 return $this->content;
1443         }
1444
1445         /**
1446          * @return string
1447          */
1448         function getModel() {
1449                 if ( is_null( $this->model ) ) {
1450                         $this->model = $this->getTitle()->getContentModel();
1451                 }
1452
1453                 return $this->model;
1454         }
1455
1456         /**
1457          * @return string
1458          */
1459         function getFormat() {
1460                 if ( is_null( $this->format ) ) {
1461                         $this->format = $this->getContentHandler()->getDefaultFormat();
1462                 }
1463
1464                 return $this->format;
1465         }
1466
1467         /**
1468          * @return string
1469          */
1470         function getComment() {
1471                 return $this->comment;
1472         }
1473
1474         /**
1475          * @return bool
1476          */
1477         function getMinor() {
1478                 return $this->minor;
1479         }
1480
1481         /**
1482          * @return mixed
1483          */
1484         function getSrc() {
1485                 return $this->src;
1486         }
1487
1488         /**
1489          * @return bool|string
1490          */
1491         function getSha1() {
1492                 if ( $this->sha1base36 ) {
1493                         return wfBaseConvert( $this->sha1base36, 36, 16 );
1494                 }
1495                 return false;
1496         }
1497
1498         /**
1499          * @return string
1500          */
1501         function getFileSrc() {
1502                 return $this->fileSrc;
1503         }
1504
1505         /**
1506          * @return bool
1507          */
1508         function isTempSrc() {
1509                 return $this->isTemp;
1510         }
1511
1512         /**
1513          * @return mixed
1514          */
1515         function getFilename() {
1516                 return $this->filename;
1517         }
1518
1519         /**
1520          * @return string
1521          */
1522         function getArchiveName() {
1523                 return $this->archiveName;
1524         }
1525
1526         /**
1527          * @return mixed
1528          */
1529         function getSize() {
1530                 return $this->size;
1531         }
1532
1533         /**
1534          * @return string
1535          */
1536         function getType() {
1537                 return $this->type;
1538         }
1539
1540         /**
1541          * @return string
1542          */
1543         function getAction() {
1544                 return $this->action;
1545         }
1546
1547         /**
1548          * @return string
1549          */
1550         function getParams() {
1551                 return $this->params;
1552         }
1553
1554         /**
1555          * @return bool
1556          */
1557         function importOldRevision() {
1558                 $dbw = wfGetDB( DB_MASTER );
1559
1560                 # Sneak a single revision into place
1561                 $user = User::newFromName( $this->getUser() );
1562                 if ( $user ) {
1563                         $userId = intval( $user->getId() );
1564                         $userText = $user->getName();
1565                         $userObj = $user;
1566                 } else {
1567                         $userId = 0;
1568                         $userText = $this->getUser();
1569                         $userObj = new User;
1570                 }
1571
1572                 // avoid memory leak...?
1573                 Title::clearCaches();
1574
1575                 $page = WikiPage::factory( $this->title );
1576                 $page->loadPageData( 'fromdbmaster' );
1577                 if ( !$page->exists() ) {
1578                         # must create the page...
1579                         $pageId = $page->insertOn( $dbw );
1580                         $created = true;
1581                         $oldcountable = null;
1582                 } else {
1583                         $pageId = $page->getId();
1584                         $created = false;
1585
1586                         $prior = $dbw->selectField( 'revision', '1',
1587                                 array( 'rev_page' => $pageId,
1588                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1589                                         'rev_user_text' => $userText,
1590                                         'rev_comment' => $this->getComment() ),
1591                                 __METHOD__
1592                         );
1593                         if ( $prior ) {
1594                                 // @todo FIXME: This could fail slightly for multiple matches :P
1595                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1596                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1597                                 return false;
1598                         }
1599                 }
1600
1601                 # @todo FIXME: Use original rev_id optionally (better for backups)
1602                 # Insert the row
1603                 $revision = new Revision( array(
1604                         'title' => $this->title,
1605                         'page' => $pageId,
1606                         'content_model' => $this->getModel(),
1607                         'content_format' => $this->getFormat(),
1608                         //XXX: just set 'content' => $this->getContent()?
1609                         'text' => $this->getContent()->serialize( $this->getFormat() ),
1610                         'comment' => $this->getComment(),
1611                         'user' => $userId,
1612                         'user_text' => $userText,
1613                         'timestamp' => $this->timestamp,
1614                         'minor_edit' => $this->minor,
1615                         ) );
1616                 $revision->insertOn( $dbw );
1617                 $changed = $page->updateIfNewerOn( $dbw, $revision );
1618
1619                 if ( $changed !== false && !$this->mNoUpdates ) {
1620                         wfDebug( __METHOD__ . ": running updates\n" );
1621                         // countable/oldcountable stuff is handled in WikiImporter::finishImportPage
1622                         $page->doEditUpdates(
1623                                 $revision,
1624                                 $userObj,
1625                                 array( 'created' => $created, 'oldcountable' => 'no-change' )
1626                         );
1627                 }
1628
1629                 return true;
1630         }
1631
1632         function importLogItem() {
1633                 $dbw = wfGetDB( DB_MASTER );
1634                 # @todo FIXME: This will not record autoblocks
1635                 if ( !$this->getTitle() ) {
1636                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1637                                 $this->timestamp . "\n" );
1638                         return;
1639                 }
1640                 # Check if it exists already
1641                 // @todo FIXME: Use original log ID (better for backups)
1642                 $prior = $dbw->selectField( 'logging', '1',
1643                         array( 'log_type' => $this->getType(),
1644                                 'log_action' => $this->getAction(),
1645                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1646                                 'log_namespace' => $this->getTitle()->getNamespace(),
1647                                 'log_title' => $this->getTitle()->getDBkey(),
1648                                 'log_comment' => $this->getComment(),
1649                                 #'log_user_text' => $this->user_text,
1650                                 'log_params' => $this->params ),
1651                         __METHOD__
1652                 );
1653                 // @todo FIXME: This could fail slightly for multiple matches :P
1654                 if ( $prior ) {
1655                         wfDebug( __METHOD__
1656                                 . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp "
1657                                 . $this->timestamp . "\n" );
1658                         return;
1659                 }
1660                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1661                 $data = array(
1662                         'log_id' => $log_id,
1663                         'log_type' => $this->type,
1664                         'log_action' => $this->action,
1665                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1666                         'log_user' => User::idFromName( $this->user_text ),
1667                         #'log_user_text' => $this->user_text,
1668                         'log_namespace' => $this->getTitle()->getNamespace(),
1669                         'log_title' => $this->getTitle()->getDBkey(),
1670                         'log_comment' => $this->getComment(),
1671                         'log_params' => $this->params
1672                 );
1673                 $dbw->insert( 'logging', $data, __METHOD__ );
1674         }
1675
1676         /**
1677          * @return bool
1678          */
1679         function importUpload() {
1680                 # Construct a file
1681                 $archiveName = $this->getArchiveName();
1682                 if ( $archiveName ) {
1683                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1684                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1685                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1686                 } else {
1687                         $file = wfLocalFile( $this->getTitle() );
1688                         $file->load( File::READ_LATEST );
1689                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1690                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1691                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1692                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1693                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1694                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1695                         }
1696                 }
1697                 if ( !$file ) {
1698                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1699                         return false;
1700                 }
1701
1702                 # Get the file source or download if necessary
1703                 $source = $this->getFileSrc();
1704                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1705                 if ( !$source ) {
1706                         $source = $this->downloadSource();
1707                         $flags |= File::DELETE_SOURCE;
1708                 }
1709                 if ( !$source ) {
1710                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1711                         return false;
1712                 }
1713                 $sha1 = $this->getSha1();
1714                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1715                         if ( $flags & File::DELETE_SOURCE ) {
1716                                 # Broken file; delete it if it is a temporary file
1717                                 unlink( $source );
1718                         }
1719                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1720                         return false;
1721                 }
1722
1723                 $user = User::newFromName( $this->user_text );
1724
1725                 # Do the actual upload
1726                 if ( $archiveName ) {
1727                         $status = $file->uploadOld( $source, $archiveName,
1728                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1729                 } else {
1730                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1731                                 $flags, false, $this->getTimestamp(), $user );
1732                 }
1733
1734                 if ( $status->isGood() ) {
1735                         wfDebug( __METHOD__ . ": Successful\n" );
1736                         return true;
1737                 } else {
1738                         wfDebug( __METHOD__ . ': failed: ' . $status->getHTML() . "\n" );
1739                         return false;
1740                 }
1741         }
1742
1743         /**
1744          * @return bool|string
1745          */
1746         function downloadSource() {
1747                 if ( !$this->config->get( 'EnableUploads' ) ) {
1748                         return false;
1749                 }
1750
1751                 $tempo = tempnam( wfTempDir(), 'download' );
1752                 $f = fopen( $tempo, 'wb' );
1753                 if ( !$f ) {
1754                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1755                         return false;
1756                 }
1757
1758                 // @todo FIXME!
1759                 $src = $this->getSrc();
1760                 $data = Http::get( $src, array(), __METHOD__ );
1761                 if ( !$data ) {
1762                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1763                         fclose( $f );
1764                         unlink( $tempo );
1765                         return false;
1766                 }
1767
1768                 fwrite( $f, $data );
1769                 fclose( $f );
1770
1771                 return $tempo;
1772         }
1773
1774 }
1775
1776 /**
1777  * Source interface for XML import.
1778  */
1779 interface ImportSource {
1780
1781         /**
1782          * Indicates whether the end of the input has been reached.
1783          * Will return true after a finite number of calls to readChunk.
1784          *
1785          * @return bool true if there is no more input, false otherwise.
1786          */
1787         function atEnd();
1788
1789         /**
1790          * Return a chunk of the input, as a (possibly empty) string.
1791          * When the end of input is reached, readChunk() returns false.
1792          * If atEnd() returns false, readChunk() will return a string.
1793          * If atEnd() returns true, readChunk() will return false.
1794          *
1795          * @return bool|string
1796          */
1797         function readChunk();
1798 }
1799
1800 /**
1801  * Used for importing XML dumps where the content of the dump is in a string.
1802  * This class is ineffecient, and should only be used for small dumps.
1803  * For larger dumps, ImportStreamSource should be used instead.
1804  *
1805  * @ingroup SpecialPage
1806  */
1807 class ImportStringSource implements ImportSource {
1808         function __construct( $string ) {
1809                 $this->mString = $string;
1810                 $this->mRead = false;
1811         }
1812
1813         /**
1814          * @return bool
1815          */
1816         function atEnd() {
1817                 return $this->mRead;
1818         }
1819
1820         /**
1821          * @return bool|string
1822          */
1823         function readChunk() {
1824                 if ( $this->atEnd() ) {
1825                         return false;
1826                 }
1827                 $this->mRead = true;
1828                 return $this->mString;
1829         }
1830 }
1831
1832 /**
1833  * Imports a XML dump from a file (either from file upload, files on disk, or HTTP)
1834  * @ingroup SpecialPage
1835  */
1836 class ImportStreamSource implements ImportSource {
1837         function __construct( $handle ) {
1838                 $this->mHandle = $handle;
1839         }
1840
1841         /**
1842          * @return bool
1843          */
1844         function atEnd() {
1845                 return feof( $this->mHandle );
1846         }
1847
1848         /**
1849          * @return string
1850          */
1851         function readChunk() {
1852                 return fread( $this->mHandle, 32768 );
1853         }
1854
1855         /**
1856          * @param string $filename
1857          * @return Status
1858          */
1859         static function newFromFile( $filename ) {
1860                 MediaWiki\suppressWarnings();
1861                 $file = fopen( $filename, 'rt' );
1862                 MediaWiki\restoreWarnings();
1863                 if ( !$file ) {
1864                         return Status::newFatal( "importcantopen" );
1865                 }
1866                 return Status::newGood( new ImportStreamSource( $file ) );
1867         }
1868
1869         /**
1870          * @param string $fieldname
1871          * @return Status
1872          */
1873         static function newFromUpload( $fieldname = "xmlimport" ) {
1874                 $upload =& $_FILES[$fieldname];
1875
1876                 if ( $upload === null || !$upload['name'] ) {
1877                         return Status::newFatal( 'importnofile' );
1878                 }
1879                 if ( !empty( $upload['error'] ) ) {
1880                         switch ( $upload['error'] ) {
1881                                 case 1:
1882                                         # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1883                                         return Status::newFatal( 'importuploaderrorsize' );
1884                                 case 2:
1885                                         # The uploaded file exceeds the MAX_FILE_SIZE directive that
1886                                         # was specified in the HTML form.
1887                                         return Status::newFatal( 'importuploaderrorsize' );
1888                                 case 3:
1889                                         # The uploaded file was only partially uploaded
1890                                         return Status::newFatal( 'importuploaderrorpartial' );
1891                                 case 6:
1892                                         # Missing a temporary folder.
1893                                         return Status::newFatal( 'importuploaderrortemp' );
1894                                 # case else: # Currently impossible
1895                         }
1896
1897                 }
1898                 $fname = $upload['tmp_name'];
1899                 if ( is_uploaded_file( $fname ) ) {
1900                         return ImportStreamSource::newFromFile( $fname );
1901                 } else {
1902                         return Status::newFatal( 'importnofile' );
1903                 }
1904         }
1905
1906         /**
1907          * @param string $url
1908          * @param string $method
1909          * @return Status
1910          */
1911         static function newFromURL( $url, $method = 'GET' ) {
1912                 wfDebug( __METHOD__ . ": opening $url\n" );
1913                 # Use the standard HTTP fetch function; it times out
1914                 # quicker and sorts out user-agent problems which might
1915                 # otherwise prevent importing from large sites, such
1916                 # as the Wikimedia cluster, etc.
1917                 $data = Http::request( $method, $url, array( 'followRedirects' => true ), __METHOD__ );
1918                 if ( $data !== false ) {
1919                         $file = tmpfile();
1920                         fwrite( $file, $data );
1921                         fflush( $file );
1922                         fseek( $file, 0 );
1923                         return Status::newGood( new ImportStreamSource( $file ) );
1924                 } else {
1925                         return Status::newFatal( 'importcantopen' );
1926                 }
1927         }
1928
1929         /**
1930          * @param string $interwiki
1931          * @param string $page
1932          * @param bool $history
1933          * @param bool $templates
1934          * @param int $pageLinkDepth
1935          * @return Status
1936          */
1937         public static function newFromInterwiki( $interwiki, $page, $history = false,
1938                 $templates = false, $pageLinkDepth = 0
1939         ) {
1940                 if ( $page == '' ) {
1941                         return Status::newFatal( 'import-noarticle' );
1942                 }
1943                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1944                 if ( is_null( $link ) || !$link->isExternal() ) {
1945                         return Status::newFatal( 'importbadinterwiki' );
1946                 } else {
1947                         $params = array();
1948                         if ( $history ) {
1949                                 $params['history'] = 1;
1950                         }
1951                         if ( $templates ) {
1952                                 $params['templates'] = 1;
1953                         }
1954                         if ( $pageLinkDepth ) {
1955                                 $params['pagelink-depth'] = $pageLinkDepth;
1956                         }
1957                         $url = $link->getFullURL( $params );
1958                         # For interwikis, use POST to avoid redirects.
1959                         return ImportStreamSource::newFromURL( $url, "POST" );
1960                 }
1961         }
1962 }