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, $mTargetNamespace, $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->mTargetNamespace = null;
 244                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 245                         return true;
 246                 } elseif (
 247                         $namespace >= 0 &&
 248                         MWNamespace::exists( intval( $namespace ) )
 249                 ) {
 250                         $namespace = intval( $namespace );
 251                         $this->mTargetNamespace = $namespace;
 252                         $this->setImportTitleFactory( new NamespaceImportTitleFactory( $namespace ) );
 253                         return true;
 254                 } else {
 255                         return false;
 256                 }
 257         }
 258
 259         /**
 260          * Set a target root page under which all pages are imported
 261          * @param null|string $rootpage
 262          * @return Status
 263          */
 264         public function setTargetRootPage( $rootpage ) {
 265                 $status = Status::newGood();
 266                 if ( is_null( $rootpage ) ) {
 267                         // No rootpage
 268                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 269                 } elseif ( $rootpage !== '' ) {
 270                         $rootpage = rtrim( $rootpage, '/' ); //avoid double slashes
 271                         $title = Title::newFromText( $rootpage, !is_null( $this->mTargetNamespace )
 272                                 ? $this->mTargetNamespace
 273                                 : NS_MAIN
 274                         );
 275
 276                         if ( !$title || $title->isExternal() ) {
 277                                 $status->fatal( 'import-rootpage-invalid' );
 278                         } else {
 279                                 if ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 280                                         global $wgContLang;
 281
 282                                         $displayNSText = $title->getNamespace() == NS_MAIN
 283                                                 ? wfMessage( 'blanknamespace' )->text()
 284                                                 : $wgContLang->getNsText( $title->getNamespace() );
 285                                         $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 286                                 } else {
 287                                         // set namespace to 'all', so the namespace check in processTitle() can pass
 288                                         $this->setTargetNamespace( null );
 289                                         $this->setImportTitleFactory( new SubpageImportTitleFactory( $title ) );
 290                                 }
 291                         }
 292                 }
 293                 return $status;
 294         }
 295
 296         /**
 297          * @param string $dir
 298          */
 299         public function setImageBasePath( $dir ) {
 300                 $this->mImageBasePath = $dir;
 301         }
 302
 303         /**
 304          * @param bool $import
 305          */
 306         public function setImportUploads( $import ) {
 307                 $this->mImportUploads = $import;
 308         }
 309
 310         /**
 311          * Default per-page callback. Sets up some things related to site statistics
 312          * @param array $titleAndForeignTitle Two-element array, with Title object at
 313          * index 0 and ForeignTitle object at index 1
 314          * @return bool
 315          */
 316         public function beforeImportPage( $titleAndForeignTitle ) {
 317                 $title = $titleAndForeignTitle[0];
 318                 $page = WikiPage::factory( $title );
 319                 $this->countableCache['title_' . $title->getPrefixedText()] = $page->isCountable();
 320                 return true;
 321         }
 322
 323         /**
 324          * Default per-revision callback, performs the import.
 325          * @param WikiRevision $revision
 326          * @return bool
 327          */
 328         public function importRevision( $revision ) {
 329                 if ( !$revision->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
 330                         $this->notice( 'import-error-bad-location',
 331                                 $revision->getTitle()->getPrefixedText(),
 332                                 $revision->getID(),
 333                                 $revision->getModel(),
 334                                 $revision->getFormat() );
 335
 336                         return false;
 337                 }
 338
 339                 try {
 340                         $dbw = wfGetDB( DB_MASTER );
 341                         return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
 342                 } catch ( MWContentSerializationException $ex ) {
 343                         $this->notice( 'import-error-unserialize',
 344                                 $revision->getTitle()->getPrefixedText(),
 345                                 $revision->getID(),
 346                                 $revision->getModel(),
 347                                 $revision->getFormat() );
 348                 }
 349
 350                 return false;
 351         }
 352
 353         /**
 354          * Default per-revision callback, performs the import.
 355          * @param WikiRevision $revision
 356          * @return bool
 357          */
 358         public function importLogItem( $revision ) {
 359                 $dbw = wfGetDB( DB_MASTER );
 360                 return $dbw->deadlockLoop( array( $revision, 'importLogItem' ) );
 361         }
 362
 363         /**
 364          * Dummy for now...
 365          * @param WikiRevision $revision
 366          * @return bool
 367          */
 368         public function importUpload( $revision ) {
 369                 $dbw = wfGetDB( DB_MASTER );
 370                 return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
 371         }
 372
 373         /**
 374          * Mostly for hook use
 375          * @param Title $title
 376          * @param ForeignTitle $foreignTitle
 377          * @param int $revCount
 378          * @param int $sRevCount
 379          * @param array $pageInfo
 380          * @return bool
 381          */
 382         public function finishImportPage( $title, $foreignTitle, $revCount,
 383                         $sRevCount, $pageInfo ) {
 384
 385                 // Update article count statistics (T42009)
 386                 // The normal counting logic in WikiPage->doEditUpdates() is designed for
 387                 // one-revision-at-a-time editing, not bulk imports. In this situation it
 388                 // suffers from issues of slave lag. We let WikiPage handle the total page
 389                 // and revision count, and we implement our own custom logic for the
 390                 // article (content page) count.
 391                 $page = WikiPage::factory( $title );
 392                 $page->loadPageData( 'fromdbmaster' );
 393                 $content = $page->getContent();
 394                 if ( $content === null ) {
 395                         wfDebug( __METHOD__ . ': Skipping article count adjustment for ' . $title .
 396                                 ' because WikiPage::getContent() returned null' );
 397                 } else {
 398                         $editInfo = $page->prepareContentForEdit( $content );
 399                         $countKey = 'title_' . $title->getPrefixedText();
 400                         $countable = $page->isCountable( $editInfo );
 401                         if ( array_key_exists( $countKey, $this->countableCache ) &&
 402                                 $countable != $this->countableCache[ $countKey ] ) {
 403                                 DeferredUpdates::addUpdate( SiteStatsUpdate::factory( array(
 404                                         'articles' => ( (int)$countable - (int)$this->countableCache[ $countKey ] )
 405                                 ) ) );
 406                         }
 407                 }
 408
 409                 $args = func_get_args();
 410                 return Hooks::run( 'AfterImportPage', $args );
 411         }
 412
 413         /**
 414          * Alternate per-revision callback, for debugging.
 415          * @param WikiRevision $revision
 416          */
 417         public function debugRevisionHandler( &$revision ) {
 418                 $this->debug( "Got revision:" );
 419                 if ( is_object( $revision->title ) ) {
 420                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 421                 } else {
 422                         $this->debug( "-- Title: <invalid>" );
 423                 }
 424                 $this->debug( "-- User: " . $revision->user_text );
 425                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 426                 $this->debug( "-- Comment: " . $revision->comment );
 427                 $this->debug( "-- Text: " . $revision->text );
 428         }
 429
 430         /**
 431          * Notify the callback function of site info
 432          * @param array $siteInfo
 433          * @return bool|mixed
 434          */
 435         private function siteInfoCallback( $siteInfo ) {
 436                 if ( isset( $this->mSiteInfoCallback ) ) {
 437                         return call_user_func_array( $this->mSiteInfoCallback,
 438                                         array( $siteInfo, $this ) );
 439                 } else {
 440                         return false;
 441                 }
 442         }
 443
 444         /**
 445          * Notify the callback function when a new "<page>" is reached.
 446          * @param Title $title
 447          */
 448         function pageCallback( $title ) {
 449                 if ( isset( $this->mPageCallback ) ) {
 450                         call_user_func( $this->mPageCallback, $title );
 451                 }
 452         }
 453
 454         /**
 455          * Notify the callback function when a "</page>" is closed.
 456          * @param Title $title
 457          * @param ForeignTitle $foreignTitle
 458          * @param int $revCount
 459          * @param int $sucCount Number of revisions for which callback returned true
 460          * @param array $pageInfo Associative array of page information
 461          */
 462         private function pageOutCallback( $title, $foreignTitle, $revCount,
 463                         $sucCount, $pageInfo ) {
 464                 if ( isset( $this->mPageOutCallback ) ) {
 465                         $args = func_get_args();
 466                         call_user_func_array( $this->mPageOutCallback, $args );
 467                 }
 468         }
 469
 470         /**
 471          * Notify the callback function of a revision
 472          * @param WikiRevision $revision
 473          * @return bool|mixed
 474          */
 475         private function revisionCallback( $revision ) {
 476                 if ( isset( $this->mRevisionCallback ) ) {
 477                         return call_user_func_array( $this->mRevisionCallback,
 478                                         array( $revision, $this ) );
 479                 } else {
 480                         return false;
 481                 }
 482         }
 483
 484         /**
 485          * Notify the callback function of a new log item
 486          * @param WikiRevision $revision
 487          * @return bool|mixed
 488          */
 489         private function logItemCallback( $revision ) {
 490                 if ( isset( $this->mLogItemCallback ) ) {
 491                         return call_user_func_array( $this->mLogItemCallback,
 492                                         array( $revision, $this ) );
 493                 } else {
 494                         return false;
 495                 }
 496         }
 497
 498         /**
 499          * Retrieves the contents of the named attribute of the current element.
 500          * @param string $attr The name of the attribute
 501          * @return string The value of the attribute or an empty string if it is not set in the current
 502          * element.
 503          */
 504         public function nodeAttribute( $attr ) {
 505                 return $this->reader->getAttribute( $attr );
 506         }
 507
 508         /**
 509          * Shouldn't something like this be built-in to XMLReader?
 510          * Fetches text contents of the current element, assuming
 511          * no sub-elements or such scary things.
 512          * @return string
 513          * @access private
 514          */
 515         public function nodeContents() {
 516                 if ( $this->reader->isEmptyElement ) {
 517                         return "";
 518                 }
 519                 $buffer = "";
 520                 while ( $this->reader->read() ) {
 521                         switch ( $this->reader->nodeType ) {
 522                         case XMLReader::TEXT:
 523                         case XMLReader::SIGNIFICANT_WHITESPACE:
 524                                 $buffer .= $this->reader->value;
 525                                 break;
 526                         case XMLReader::END_ELEMENT:
 527                                 return $buffer;
 528                         }
 529                 }
 530
 531                 $this->reader->close();
 532                 return '';
 533         }
 534
 535         /**
 536          * Primary entry point
 537          * @throws MWException
 538          * @return bool
 539          */
 540         public function doImport() {
 541                 // Calls to reader->read need to be wrapped in calls to
 542                 // libxml_disable_entity_loader() to avoid local file
 543                 // inclusion attacks (bug 46932).
 544                 $oldDisable = libxml_disable_entity_loader( true );
 545                 $this->reader->read();
 546
 547                 if ( $this->reader->localName != 'mediawiki' ) {
 548                         libxml_disable_entity_loader( $oldDisable );
 549                         throw new MWException( "Expected <mediawiki> tag, got " .
 550                                 $this->reader->localName );
 551                 }
 552                 $this->debug( "<mediawiki> tag is correct." );
 553
 554                 $this->debug( "Starting primary dump processing loop." );
 555
 556                 $keepReading = $this->reader->read();
 557                 $skip = false;
 558                 $rethrow = null;
 559                 try {
 560                         while ( $keepReading ) {
 561                                 $tag = $this->reader->localName;
 562                                 $type = $this->reader->nodeType;
 563
 564                                 if ( !Hooks::run( 'ImportHandleToplevelXMLTag', array( $this ) ) ) {
 565                                         // Do nothing
 566                                 } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) {
 567                                         break;
 568                                 } elseif ( $tag == 'siteinfo' ) {
 569                                         $this->handleSiteInfo();
 570                                 } elseif ( $tag == 'page' ) {
 571                                         $this->handlePage();
 572                                 } elseif ( $tag == 'logitem' ) {
 573                                         $this->handleLogItem();
 574                                 } elseif ( $tag != '#text' ) {
 575                                         $this->warn( "Unhandled top-level XML tag $tag" );
 576
 577                                         $skip = true;
 578                                 }
 579
 580                                 if ( $skip ) {
 581                                         $keepReading = $this->reader->next();
 582                                         $skip = false;
 583                                         $this->debug( "Skip" );
 584                                 } else {
 585                                         $keepReading = $this->reader->read();
 586                                 }
 587                         }
 588                 } catch ( Exception $ex ) {
 589                         $rethrow = $ex;
 590                 }
 591
 592                 // finally
 593                 libxml_disable_entity_loader( $oldDisable );
 594                 $this->reader->close();
 595
 596                 if ( $rethrow ) {
 597                         throw $rethrow;
 598                 }
 599
 600                 return true;
 601         }
 602
 603         private function handleSiteInfo() {
 604                 $this->debug( "Enter site info handler." );
 605                 $siteInfo = array();
 606
 607                 // Fields that can just be stuffed in the siteInfo object
 608                 $normalFields = array( 'sitename', 'base', 'generator', 'case' );
 609
 610                 while ( $this->reader->read() ) {
 611                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 612                                         $this->reader->localName == 'siteinfo' ) {
 613                                 break;
 614                         }
 615
 616                         $tag = $this->reader->localName;
 617
 618                         if ( $tag == 'namespace' ) {
 619                                 $this->foreignNamespaces[ $this->nodeAttribute( 'key' ) ] =
 620                                         $this->nodeContents();
 621                         } elseif ( in_array( $tag, $normalFields ) ) {
 622                                 $siteInfo[$tag] = $this->nodeContents();
 623                         }
 624                 }
 625
 626                 $siteInfo['_namespaces'] = $this->foreignNamespaces;
 627                 $this->siteInfoCallback( $siteInfo );
 628         }
 629
 630         private function handleLogItem() {
 631                 $this->debug( "Enter log item handler." );
 632                 $logInfo = array();
 633
 634                 // Fields that can just be stuffed in the pageInfo object
 635                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 636                                         'logtitle', 'params' );
 637
 638                 while ( $this->reader->read() ) {
 639                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 640                                         $this->reader->localName == 'logitem' ) {
 641                                 break;
 642                         }
 643
 644                         $tag = $this->reader->localName;
 645
 646                         if ( !Hooks::run( 'ImportHandleLogItemXMLTag', array(
 647                                 $this, $logInfo
 648                         ) ) ) {
 649                                 // Do nothing
 650                         } elseif ( in_array( $tag, $normalFields ) ) {
 651                                 $logInfo[$tag] = $this->nodeContents();
 652                         } elseif ( $tag == 'contributor' ) {
 653                                 $logInfo['contributor'] = $this->handleContributor();
 654                         } elseif ( $tag != '#text' ) {
 655                                 $this->warn( "Unhandled log-item XML tag $tag" );
 656                         }
 657                 }
 658
 659                 $this->processLogItem( $logInfo );
 660         }
 661
 662         /**
 663          * @param array $logInfo
 664          * @return bool|mixed
 665          */
 666         private function processLogItem( $logInfo ) {
 667                 $revision = new WikiRevision( $this->config );
 668
 669                 $revision->setID( $logInfo['id'] );
 670                 $revision->setType( $logInfo['type'] );
 671                 $revision->setAction( $logInfo['action'] );
 672                 $revision->setTimestamp( $logInfo['timestamp'] );
 673                 $revision->setParams( $logInfo['params'] );
 674                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 675                 $revision->setNoUpdates( $this->mNoUpdates );
 676
 677                 if ( isset( $logInfo['comment'] ) ) {
 678                         $revision->setComment( $logInfo['comment'] );
 679                 }
 680
 681                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 682                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 683                 }
 684                 if ( isset( $logInfo['contributor']['username'] ) ) {
 685                         $revision->setUserName( $logInfo['contributor']['username'] );
 686                 }
 687
 688                 return $this->logItemCallback( $revision );
 689         }
 690
 691         private function handlePage() {
 692                 // Handle page data.
 693                 $this->debug( "Enter page handler." );
 694                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 695
 696                 // Fields that can just be stuffed in the pageInfo object
 697                 $normalFields = array( 'title', 'ns', 'id', 'redirect', 'restrictions' );
 698
 699                 $skip = false;
 700                 $badTitle = false;
 701
 702                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 703                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 704                                         $this->reader->localName == 'page' ) {
 705                                 break;
 706                         }
 707
 708                         $skip = false;
 709
 710                         $tag = $this->reader->localName;
 711
 712                         if ( $badTitle ) {
 713                                 // The title is invalid, bail out of this page
 714                                 $skip = true;
 715                         } elseif ( !Hooks::run( 'ImportHandlePageXMLTag', array( $this,
 716                                                 &$pageInfo ) ) ) {
 717                                 // Do nothing
 718                         } elseif ( in_array( $tag, $normalFields ) ) {
 719                                 // An XML snippet:
 720                                 // <page>
 721                                 //     <id>123</id>
 722                                 //     <title>Page</title>
 723                                 //     <redirect title="NewTitle"/>
 724                                 //     ...
 725                                 // Because the redirect tag is built differently, we need special handling for that case.
 726                                 if ( $tag == 'redirect' ) {
 727                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 728                                 } else {
 729                                         $pageInfo[$tag] = $this->nodeContents();
 730                                 }
 731                         } elseif ( $tag == 'revision' || $tag == 'upload' ) {
 732                                 if ( !isset( $title ) ) {
 733                                         $title = $this->processTitle( $pageInfo['title'],
 734                                                 isset( $pageInfo['ns'] ) ? $pageInfo['ns'] : null );
 735
 736                                         if ( !$title ) {
 737                                                 $badTitle = true;
 738                                                 $skip = true;
 739                                         }
 740
 741                                         $this->pageCallback( $title );
 742                                         list( $pageInfo['_title'], $foreignTitle ) = $title;
 743                                 }
 744
 745                                 if ( $title ) {
 746                                         if ( $tag == 'revision' ) {
 747                                                 $this->handleRevision( $pageInfo );
 748                                         } else {
 749                                                 $this->handleUpload( $pageInfo );
 750                                         }
 751                                 }
 752                         } elseif ( $tag != '#text' ) {
 753                                 $this->warn( "Unhandled page XML tag $tag" );
 754                                 $skip = true;
 755                         }
 756                 }
 757
 758                 $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
 759                                         $pageInfo['revisionCount'],
 760                                         $pageInfo['successfulRevisionCount'],
 761                                         $pageInfo );
 762         }
 763
 764         /**
 765          * @param array $pageInfo
 766          */
 767         private function handleRevision( &$pageInfo ) {
 768                 $this->debug( "Enter revision handler" );
 769                 $revisionInfo = array();
 770
 771                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' );
 772
 773                 $skip = false;
 774
 775                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 776                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 777                                         $this->reader->localName == 'revision' ) {
 778                                 break;
 779                         }
 780
 781                         $tag = $this->reader->localName;
 782
 783                         if ( !Hooks::run( 'ImportHandleRevisionXMLTag', array(
 784                                 $this, $pageInfo, $revisionInfo
 785                         ) ) ) {
 786                                 // Do nothing
 787                         } elseif ( in_array( $tag, $normalFields ) ) {
 788                                 $revisionInfo[$tag] = $this->nodeContents();
 789                         } elseif ( $tag == 'contributor' ) {
 790                                 $revisionInfo['contributor'] = $this->handleContributor();
 791                         } elseif ( $tag != '#text' ) {
 792                                 $this->warn( "Unhandled revision XML tag $tag" );
 793                                 $skip = true;
 794                         }
 795                 }
 796
 797                 $pageInfo['revisionCount']++;
 798                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 799                         $pageInfo['successfulRevisionCount']++;
 800                 }
 801         }
 802
 803         /**
 804          * @param array $pageInfo
 805          * @param array $revisionInfo
 806          * @return bool|mixed
 807          */
 808         private function processRevision( $pageInfo, $revisionInfo ) {
 809                 $revision = new WikiRevision( $this->config );
 810
 811                 if ( isset( $revisionInfo['id'] ) ) {
 812                         $revision->setID( $revisionInfo['id'] );
 813                 }
 814                 if ( isset( $revisionInfo['model'] ) ) {
 815                         $revision->setModel( $revisionInfo['model'] );
 816                 }
 817                 if ( isset( $revisionInfo['format'] ) ) {
 818                         $revision->setFormat( $revisionInfo['format'] );
 819                 }
 820                 $revision->setTitle( $pageInfo['_title'] );
 821
 822                 if ( isset( $revisionInfo['text'] ) ) {
 823                         $handler = $revision->getContentHandler();
 824                         $text = $handler->importTransform(
 825                                 $revisionInfo['text'],
 826                                 $revision->getFormat() );
 827
 828                         $revision->setText( $text );
 829                 }
 830                 if ( isset( $revisionInfo['timestamp'] ) ) {
 831                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 832                 } else {
 833                         $revision->setTimestamp( wfTimestampNow() );
 834                 }
 835
 836                 if ( isset( $revisionInfo['comment'] ) ) {
 837                         $revision->setComment( $revisionInfo['comment'] );
 838                 }
 839
 840                 if ( isset( $revisionInfo['minor'] ) ) {
 841                         $revision->setMinor( true );
 842                 }
 843                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 844                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 845                 }
 846                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 847                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 848                 }
 849                 $revision->setNoUpdates( $this->mNoUpdates );
 850
 851                 return $this->revisionCallback( $revision );
 852         }
 853
 854         /**
 855          * @param array $pageInfo
 856          * @return mixed
 857          */
 858         private function handleUpload( &$pageInfo ) {
 859                 $this->debug( "Enter upload handler" );
 860                 $uploadInfo = array();
 861
 862                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 863                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 864
 865                 $skip = false;
 866
 867                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 868                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 869                                         $this->reader->localName == 'upload' ) {
 870                                 break;
 871                         }
 872
 873                         $tag = $this->reader->localName;
 874
 875                         if ( !Hooks::run( 'ImportHandleUploadXMLTag', array(
 876                                 $this, $pageInfo
 877                         ) ) ) {
 878                                 // Do nothing
 879                         } elseif ( in_array( $tag, $normalFields ) ) {
 880                                 $uploadInfo[$tag] = $this->nodeContents();
 881                         } elseif ( $tag == 'contributor' ) {
 882                                 $uploadInfo['contributor'] = $this->handleContributor();
 883                         } elseif ( $tag == 'contents' ) {
 884                                 $contents = $this->nodeContents();
 885                                 $encoding = $this->reader->getAttribute( 'encoding' );
 886                                 if ( $encoding === 'base64' ) {
 887                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 888                                         $uploadInfo['isTempSrc'] = true;
 889                                 }
 890                         } elseif ( $tag != '#text' ) {
 891                                 $this->warn( "Unhandled upload XML tag $tag" );
 892                                 $skip = true;
 893                         }
 894                 }
 895
 896                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 897                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 898                         if ( file_exists( $path ) ) {
 899                                 $uploadInfo['fileSrc'] = $path;
 900                                 $uploadInfo['isTempSrc'] = false;
 901                         }
 902                 }
 903
 904                 if ( $this->mImportUploads ) {
 905                         return $this->processUpload( $pageInfo, $uploadInfo );
 906                 }
 907         }
 908
 909         /**
 910          * @param string $contents
 911          * @return string
 912          */
 913         private function dumpTemp( $contents ) {
 914                 $filename = tempnam( wfTempDir(), 'importupload' );
 915                 file_put_contents( $filename, $contents );
 916                 return $filename;
 917         }
 918
 919         /**
 920          * @param array $pageInfo
 921          * @param array $uploadInfo
 922          * @return mixed
 923          */
 924         private function processUpload( $pageInfo, $uploadInfo ) {
 925                 $revision = new WikiRevision( $this->config );
 926                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 927
 928                 $revision->setTitle( $pageInfo['_title'] );
 929                 $revision->setID( $pageInfo['id'] );
 930                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 931                 $revision->setText( $text );
 932                 $revision->setFilename( $uploadInfo['filename'] );
 933                 if ( isset( $uploadInfo['archivename'] ) ) {
 934                         $revision->setArchiveName( $uploadInfo['archivename'] );
 935                 }
 936                 $revision->setSrc( $uploadInfo['src'] );
 937                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 938                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 939                                 !empty( $uploadInfo['isTempSrc'] ) );
 940                 }
 941                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 942                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 943                 }
 944                 $revision->setSize( intval( $uploadInfo['size'] ) );
 945                 $revision->setComment( $uploadInfo['comment'] );
 946
 947                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 948                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 949                 }
 950                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 951                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 952                 }
 953                 $revision->setNoUpdates( $this->mNoUpdates );
 954
 955                 return call_user_func( $this->mUploadCallback, $revision );
 956         }
 957
 958         /**
 959          * @return array
 960          */
 961         private function handleContributor() {
 962                 $fields = array( 'id', 'ip', 'username' );
 963                 $info = array();
 964
 965                 while ( $this->reader->read() ) {
 966                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 967                                         $this->reader->localName == 'contributor' ) {
 968                                 break;
 969                         }
 970
 971                         $tag = $this->reader->localName;
 972
 973                         if ( in_array( $tag, $fields ) ) {
 974                                 $info[$tag] = $this->nodeContents();
 975                         }
 976                 }
 977
 978                 return $info;
 979         }
 980
 981         /**
 982          * @param string $text
 983          * @param string|null $ns
 984          * @return array|bool
 985          */
 986         private function processTitle( $text, $ns = null ) {
 987                 if ( is_null( $this->foreignNamespaces ) ) {
 988                         $foreignTitleFactory = new NaiveForeignTitleFactory();
 989                 } else {
 990                         $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
 991                                 $this->foreignNamespaces );
 992                 }
 993
 994                 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
 995                         intval( $ns ) );
 996
 997                 $title = $this->importTitleFactory->createTitleFromForeignTitle(
 998                         $foreignTitle );
 999
1000                 $commandLineMode = $this->config->get( 'CommandLineMode' );
1001                 if ( is_null( $title ) ) {
1002                         # Invalid page title? Ignore the page
1003                         $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
1004                         return false;
1005                 } elseif ( $title->isExternal() ) {
1006                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
1007                         return false;
1008                 } elseif ( !$title->canExist() ) {
1009                         $this->notice( 'import-error-special', $title->getPrefixedText() );
1010                         return false;
1011                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
1012                         # Do not import if the importing wiki user cannot edit this page
1013                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
1014                         return false;
1015                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
1016                         # Do not import if the importing wiki user cannot create this page
1017                         $this->notice( 'import-error-create', $title->getPrefixedText() );
1018                         return false;
1019                 }
1020
1021                 return array( $title, $foreignTitle );
1022         }
1023 }
1024
1025 /** This is a horrible hack used to keep source compatibility */
1026 class UploadSourceAdapter {
1027         /** @var array */
1028         public static $sourceRegistrations = array();
1029
1030         /** @var string */
1031         private $mSource;
1032
1033         /** @var string */
1034         private $mBuffer;
1035
1036         /** @var int */
1037         private $mPosition;
1038
1039         /**
1040          * @param ImportSource $source
1041          * @return string
1042          */
1043         static function registerSource( ImportSource $source ) {
1044                 $id = wfRandomString();
1045
1046                 self::$sourceRegistrations[$id] = $source;
1047
1048                 return $id;
1049         }
1050
1051         /**
1052          * @param string $path
1053          * @param string $mode
1054          * @param array $options
1055          * @param string $opened_path
1056          * @return bool
1057          */
1058         function stream_open( $path, $mode, $options, &$opened_path ) {
1059                 $url = parse_url( $path );
1060                 $id = $url['host'];
1061
1062                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
1063                         return false;
1064                 }
1065
1066                 $this->mSource = self::$sourceRegistrations[$id];
1067
1068                 return true;
1069         }
1070
1071         /**
1072          * @param int $count
1073          * @return string
1074          */
1075         function stream_read( $count ) {
1076                 $return = '';
1077                 $leave = false;
1078
1079                 while ( !$leave && !$this->mSource->atEnd() &&
1080                                 strlen( $this->mBuffer ) < $count ) {
1081                         $read = $this->mSource->readChunk();
1082
1083                         if ( !strlen( $read ) ) {
1084                                 $leave = true;
1085                         }
1086
1087                         $this->mBuffer .= $read;
1088                 }
1089
1090                 if ( strlen( $this->mBuffer ) ) {
1091                         $return = substr( $this->mBuffer, 0, $count );
1092                         $this->mBuffer = substr( $this->mBuffer, $count );
1093                 }
1094
1095                 $this->mPosition += strlen( $return );
1096
1097                 return $return;
1098         }
1099
1100         /**
1101          * @param string $data
1102          * @return bool
1103          */
1104         function stream_write( $data ) {
1105                 return false;
1106         }
1107
1108         /**
1109          * @return mixed
1110          */
1111         function stream_tell() {
1112                 return $this->mPosition;
1113         }
1114
1115         /**
1116          * @return bool
1117          */
1118         function stream_eof() {
1119                 return $this->mSource->atEnd();
1120         }
1121
1122         /**
1123          * @return array
1124          */
1125         function url_stat() {
1126                 $result = array();
1127
1128                 $result['dev'] = $result[0] = 0;
1129                 $result['ino'] = $result[1] = 0;
1130                 $result['mode'] = $result[2] = 0;
1131                 $result['nlink'] = $result[3] = 0;
1132                 $result['uid'] = $result[4] = 0;
1133                 $result['gid'] = $result[5] = 0;
1134                 $result['rdev'] = $result[6] = 0;
1135                 $result['size'] = $result[7] = 0;
1136                 $result['atime'] = $result[8] = 0;
1137                 $result['mtime'] = $result[9] = 0;
1138                 $result['ctime'] = $result[10] = 0;
1139                 $result['blksize'] = $result[11] = 0;
1140                 $result['blocks'] = $result[12] = 0;
1141
1142                 return $result;
1143         }
1144 }
1145
1146 /**
1147  * @todo document (e.g. one-sentence class description).
1148  * @ingroup SpecialPage
1149  */
1150 class WikiRevision {
1151         /** @todo Unused? */
1152         public $importer = null;
1153
1154         /** @var Title */
1155         public $title = null;
1156
1157         /** @var int */
1158         public $id = 0;
1159
1160         /** @var string */
1161         public $timestamp = "20010115000000";
1162
1163         /**
1164          * @var int
1165          * @todo Can't find any uses. Public, because that's suspicious. Get clarity. */
1166         public $user = 0;
1167
1168         /** @var string */
1169         public $user_text = "";
1170
1171         /** @var string */
1172         public $model = null;
1173
1174         /** @var string */
1175         public $format = null;
1176
1177         /** @var string */
1178         public $text = "";
1179
1180         /** @var int */
1181         protected $size;
1182
1183         /** @var Content */
1184         public $content = null;
1185
1186         /** @var ContentHandler */
1187         protected $contentHandler = null;
1188
1189         /** @var string */
1190         public $comment = "";
1191
1192         /** @var bool */
1193         public $minor = false;
1194
1195         /** @var string */
1196         public $type = "";
1197
1198         /** @var string */
1199         public $action = "";
1200
1201         /** @var string */
1202         public $params = "";
1203
1204         /** @var string */
1205         public $fileSrc = '';
1206
1207         /** @var bool|string */
1208         public $sha1base36 = false;
1209
1210         /**
1211          * @var bool
1212          * @todo Unused?
1213          */
1214         public $isTemp = false;
1215
1216         /** @var string */
1217         public $archiveName = '';
1218
1219         protected $filename;
1220
1221         /** @var mixed */
1222         protected $src;
1223
1224         /** @todo Unused? */
1225         public $fileIsTemp;
1226
1227         /** @var bool */
1228         private $mNoUpdates = false;
1229
1230         /** @var Config $config */
1231         private $config;
1232
1233         public function __construct( Config $config ) {
1234                 $this->config = $config;
1235         }
1236
1237         /**
1238          * @param Title $title
1239          * @throws MWException
1240          */
1241         function setTitle( $title ) {
1242                 if ( is_object( $title ) ) {
1243                         $this->title = $title;
1244                 } elseif ( is_null( $title ) ) {
1245                         throw new MWException( "WikiRevision given a null title in import. "
1246                                 . "You may need to adjust \$wgLegalTitleChars." );
1247                 } else {
1248                         throw new MWException( "WikiRevision given non-object title in import." );
1249                 }
1250         }
1251
1252         /**
1253          * @param int $id
1254          */
1255         function setID( $id ) {
1256                 $this->id = $id;
1257         }
1258
1259         /**
1260          * @param string $ts
1261          */
1262         function setTimestamp( $ts ) {
1263                 # 2003-08-05T18:30:02Z
1264                 $this->timestamp = wfTimestamp( TS_MW, $ts );
1265         }
1266
1267         /**
1268          * @param string $user
1269          */
1270         function setUsername( $user ) {
1271                 $this->user_text = $user;
1272         }
1273
1274         /**
1275          * @param string $ip
1276          */
1277         function setUserIP( $ip ) {
1278                 $this->user_text = $ip;
1279         }
1280
1281         /**
1282          * @param string $model
1283          */
1284         function setModel( $model ) {
1285                 $this->model = $model;
1286         }
1287
1288         /**
1289          * @param string $format
1290          */
1291         function setFormat( $format ) {
1292                 $this->format = $format;
1293         }
1294
1295         /**
1296          * @param string $text
1297          */
1298         function setText( $text ) {
1299                 $this->text = $text;
1300         }
1301
1302         /**
1303          * @param string $text
1304          */
1305         function setComment( $text ) {
1306                 $this->comment = $text;
1307         }
1308
1309         /**
1310          * @param bool $minor
1311          */
1312         function setMinor( $minor ) {
1313                 $this->minor = (bool)$minor;
1314         }
1315
1316         /**
1317          * @param mixed $src
1318          */
1319         function setSrc( $src ) {
1320                 $this->src = $src;
1321         }
1322
1323         /**
1324          * @param string $src
1325          * @param bool $isTemp
1326          */
1327         function setFileSrc( $src, $isTemp ) {
1328                 $this->fileSrc = $src;
1329                 $this->fileIsTemp = $isTemp;
1330         }
1331
1332         /**
1333          * @param string $sha1base36
1334          */
1335         function setSha1Base36( $sha1base36 ) {
1336                 $this->sha1base36 = $sha1base36;
1337         }
1338
1339         /**
1340          * @param string $filename
1341          */
1342         function setFilename( $filename ) {
1343                 $this->filename = $filename;
1344         }
1345
1346         /**
1347          * @param string $archiveName
1348          */
1349         function setArchiveName( $archiveName ) {
1350                 $this->archiveName = $archiveName;
1351         }
1352
1353         /**
1354          * @param int $size
1355          */
1356         function setSize( $size ) {
1357                 $this->size = intval( $size );
1358         }
1359
1360         /**
1361          * @param string $type
1362          */
1363         function setType( $type ) {
1364                 $this->type = $type;
1365         }
1366
1367         /**
1368          * @param string $action
1369          */
1370         function setAction( $action ) {
1371                 $this->action = $action;
1372         }
1373
1374         /**
1375          * @param array $params
1376          */
1377         function setParams( $params ) {
1378                 $this->params = $params;
1379         }
1380
1381         /**
1382          * @param bool $noupdates
1383          */
1384         public function setNoUpdates( $noupdates ) {
1385                 $this->mNoUpdates = $noupdates;
1386         }
1387
1388         /**
1389          * @return Title
1390          */
1391         function getTitle() {
1392                 return $this->title;
1393         }
1394
1395         /**
1396          * @return int
1397          */
1398         function getID() {
1399                 return $this->id;
1400         }
1401
1402         /**
1403          * @return string
1404          */
1405         function getTimestamp() {
1406                 return $this->timestamp;
1407         }
1408
1409         /**
1410          * @return string
1411          */
1412         function getUser() {
1413                 return $this->user_text;
1414         }
1415
1416         /**
1417          * @return string
1418          *
1419          * @deprecated Since 1.21, use getContent() instead.
1420          */
1421         function getText() {
1422                 ContentHandler::deprecated( __METHOD__, '1.21' );
1423
1424                 return $this->text;
1425         }
1426
1427         /**
1428          * @return ContentHandler
1429          */
1430         function getContentHandler() {
1431                 if ( is_null( $this->contentHandler ) ) {
1432                         $this->contentHandler = ContentHandler::getForModelID( $this->getModel() );
1433                 }
1434
1435                 return $this->contentHandler;
1436         }
1437
1438         /**
1439          * @return Content
1440          */
1441         function getContent() {
1442                 if ( is_null( $this->content ) ) {
1443                         $handler = $this->getContentHandler();
1444                         $this->content = $handler->unserializeContent( $this->text, $this->getFormat() );
1445                 }
1446
1447                 return $this->content;
1448         }
1449
1450         /**
1451          * @return string
1452          */
1453         function getModel() {
1454                 if ( is_null( $this->model ) ) {
1455                         $this->model = $this->getTitle()->getContentModel();
1456                 }
1457
1458                 return $this->model;
1459         }
1460
1461         /**
1462          * @return string
1463          */
1464         function getFormat() {
1465                 if ( is_null( $this->format ) ) {
1466                         $this->format = $this->getContentHandler()->getDefaultFormat();
1467                 }
1468
1469                 return $this->format;
1470         }
1471
1472         /**
1473          * @return string
1474          */
1475         function getComment() {
1476                 return $this->comment;
1477         }
1478
1479         /**
1480          * @return bool
1481          */
1482         function getMinor() {
1483                 return $this->minor;
1484         }
1485
1486         /**
1487          * @return mixed
1488          */
1489         function getSrc() {
1490                 return $this->src;
1491         }
1492
1493         /**
1494          * @return bool|string
1495          */
1496         function getSha1() {
1497                 if ( $this->sha1base36 ) {
1498                         return wfBaseConvert( $this->sha1base36, 36, 16 );
1499                 }
1500                 return false;
1501         }
1502
1503         /**
1504          * @return string
1505          */
1506         function getFileSrc() {
1507                 return $this->fileSrc;
1508         }
1509
1510         /**
1511          * @return bool
1512          */
1513         function isTempSrc() {
1514                 return $this->isTemp;
1515         }
1516
1517         /**
1518          * @return mixed
1519          */
1520         function getFilename() {
1521                 return $this->filename;
1522         }
1523
1524         /**
1525          * @return string
1526          */
1527         function getArchiveName() {
1528                 return $this->archiveName;
1529         }
1530
1531         /**
1532          * @return mixed
1533          */
1534         function getSize() {
1535                 return $this->size;
1536         }
1537
1538         /**
1539          * @return string
1540          */
1541         function getType() {
1542                 return $this->type;
1543         }
1544
1545         /**
1546          * @return string
1547          */
1548         function getAction() {
1549                 return $this->action;
1550         }
1551
1552         /**
1553          * @return string
1554          */
1555         function getParams() {
1556                 return $this->params;
1557         }
1558
1559         /**
1560          * @return bool
1561          */
1562         function importOldRevision() {
1563                 $dbw = wfGetDB( DB_MASTER );
1564
1565                 # Sneak a single revision into place
1566                 $user = User::newFromName( $this->getUser() );
1567                 if ( $user ) {
1568                         $userId = intval( $user->getId() );
1569                         $userText = $user->getName();
1570                         $userObj = $user;
1571                 } else {
1572                         $userId = 0;
1573                         $userText = $this->getUser();
1574                         $userObj = new User;
1575                 }
1576
1577                 // avoid memory leak...?
1578                 $linkCache = LinkCache::singleton();
1579                 $linkCache->clear();
1580
1581                 $page = WikiPage::factory( $this->title );
1582                 $page->loadPageData( 'fromdbmaster' );
1583                 if ( !$page->exists() ) {
1584                         # must create the page...
1585                         $pageId = $page->insertOn( $dbw );
1586                         $created = true;
1587                         $oldcountable = null;
1588                 } else {
1589                         $pageId = $page->getId();
1590                         $created = false;
1591
1592                         $prior = $dbw->selectField( 'revision', '1',
1593                                 array( 'rev_page' => $pageId,
1594                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1595                                         'rev_user_text' => $userText,
1596                                         'rev_comment' => $this->getComment() ),
1597                                 __METHOD__
1598                         );
1599                         if ( $prior ) {
1600                                 // @todo FIXME: This could fail slightly for multiple matches :P
1601                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1602                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1603                                 return false;
1604                         }
1605                 }
1606
1607                 # @todo FIXME: Use original rev_id optionally (better for backups)
1608                 # Insert the row
1609                 $revision = new Revision( array(
1610                         'title' => $this->title,
1611                         'page' => $pageId,
1612                         'content_model' => $this->getModel(),
1613                         'content_format' => $this->getFormat(),
1614                         //XXX: just set 'content' => $this->getContent()?
1615                         'text' => $this->getContent()->serialize( $this->getFormat() ),
1616                         'comment' => $this->getComment(),
1617                         'user' => $userId,
1618                         'user_text' => $userText,
1619                         'timestamp' => $this->timestamp,
1620                         'minor_edit' => $this->minor,
1621                         ) );
1622                 $revision->insertOn( $dbw );
1623                 $changed = $page->updateIfNewerOn( $dbw, $revision );
1624
1625                 if ( $changed !== false && !$this->mNoUpdates ) {
1626                         wfDebug( __METHOD__ . ": running updates\n" );
1627                         // countable/oldcountable stuff is handled in WikiImporter::finishImportPage
1628                         $page->doEditUpdates(
1629                                 $revision,
1630                                 $userObj,
1631                                 array( 'created' => $created, 'oldcountable' => 'no-change' )
1632                         );
1633                 }
1634
1635                 return true;
1636         }
1637
1638         function importLogItem() {
1639                 $dbw = wfGetDB( DB_MASTER );
1640                 # @todo FIXME: This will not record autoblocks
1641                 if ( !$this->getTitle() ) {
1642                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1643                                 $this->timestamp . "\n" );
1644                         return;
1645                 }
1646                 # Check if it exists already
1647                 // @todo FIXME: Use original log ID (better for backups)
1648                 $prior = $dbw->selectField( 'logging', '1',
1649                         array( 'log_type' => $this->getType(),
1650                                 'log_action' => $this->getAction(),
1651                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1652                                 'log_namespace' => $this->getTitle()->getNamespace(),
1653                                 'log_title' => $this->getTitle()->getDBkey(),
1654                                 'log_comment' => $this->getComment(),
1655                                 #'log_user_text' => $this->user_text,
1656                                 'log_params' => $this->params ),
1657                         __METHOD__
1658                 );
1659                 // @todo FIXME: This could fail slightly for multiple matches :P
1660                 if ( $prior ) {
1661                         wfDebug( __METHOD__
1662                                 . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp "
1663                                 . $this->timestamp . "\n" );
1664                         return;
1665                 }
1666                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1667                 $data = array(
1668                         'log_id' => $log_id,
1669                         'log_type' => $this->type,
1670                         'log_action' => $this->action,
1671                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1672                         'log_user' => User::idFromName( $this->user_text ),
1673                         #'log_user_text' => $this->user_text,
1674                         'log_namespace' => $this->getTitle()->getNamespace(),
1675                         'log_title' => $this->getTitle()->getDBkey(),
1676                         'log_comment' => $this->getComment(),
1677                         'log_params' => $this->params
1678                 );
1679                 $dbw->insert( 'logging', $data, __METHOD__ );
1680         }
1681
1682         /**
1683          * @return bool
1684          */
1685         function importUpload() {
1686                 # Construct a file
1687                 $archiveName = $this->getArchiveName();
1688                 if ( $archiveName ) {
1689                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1690                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1691                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1692                 } else {
1693                         $file = wfLocalFile( $this->getTitle() );
1694                         $file->load( File::READ_LATEST );
1695                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1696                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1697                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1698                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1699                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1700                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1701                         }
1702                 }
1703                 if ( !$file ) {
1704                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1705                         return false;
1706                 }
1707
1708                 # Get the file source or download if necessary
1709                 $source = $this->getFileSrc();
1710                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1711                 if ( !$source ) {
1712                         $source = $this->downloadSource();
1713                         $flags |= File::DELETE_SOURCE;
1714                 }
1715                 if ( !$source ) {
1716                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1717                         return false;
1718                 }
1719                 $sha1 = $this->getSha1();
1720                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1721                         if ( $flags & File::DELETE_SOURCE ) {
1722                                 # Broken file; delete it if it is a temporary file
1723                                 unlink( $source );
1724                         }
1725                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1726                         return false;
1727                 }
1728
1729                 $user = User::newFromName( $this->user_text );
1730
1731                 # Do the actual upload
1732                 if ( $archiveName ) {
1733                         $status = $file->uploadOld( $source, $archiveName,
1734                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1735                 } else {
1736                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1737                                 $flags, false, $this->getTimestamp(), $user );
1738                 }
1739
1740                 if ( $status->isGood() ) {
1741                         wfDebug( __METHOD__ . ": Successful\n" );
1742                         return true;
1743                 } else {
1744                         wfDebug( __METHOD__ . ': failed: ' . $status->getHTML() . "\n" );
1745                         return false;
1746                 }
1747         }
1748
1749         /**
1750          * @return bool|string
1751          */
1752         function downloadSource() {
1753                 if ( !$this->config->get( 'EnableUploads' ) ) {
1754                         return false;
1755                 }
1756
1757                 $tempo = tempnam( wfTempDir(), 'download' );
1758                 $f = fopen( $tempo, 'wb' );
1759                 if ( !$f ) {
1760                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1761                         return false;
1762                 }
1763
1764                 // @todo FIXME!
1765                 $src = $this->getSrc();
1766                 $data = Http::get( $src, array(), __METHOD__ );
1767                 if ( !$data ) {
1768                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1769                         fclose( $f );
1770                         unlink( $tempo );
1771                         return false;
1772                 }
1773
1774                 fwrite( $f, $data );
1775                 fclose( $f );
1776
1777                 return $tempo;
1778         }
1779
1780 }
1781
1782 /**
1783  * Source interface for XML import.
1784  */
1785 interface ImportSource {
1786
1787         /**
1788          * Indicates whether the end of the input has been reached.
1789          * Will return true after a finite number of calls to readChunk.
1790          *
1791          * @return bool true if there is no more input, false otherwise.
1792          */
1793         function atEnd();
1794
1795         /**
1796          * Return a chunk of the input, as a (possibly empty) string.
1797          * When the end of input is reached, readChunk() returns false.
1798          * If atEnd() returns false, readChunk() will return a string.
1799          * If atEnd() returns true, readChunk() will return false.
1800          *
1801          * @return bool|string
1802          */
1803         function readChunk();
1804 }
1805
1806 /**
1807  * Used for importing XML dumps where the content of the dump is in a string.
1808  * This class is ineffecient, and should only be used for small dumps.
1809  * For larger dumps, ImportStreamSource should be used instead.
1810  *
1811  * @ingroup SpecialPage
1812  */
1813 class ImportStringSource implements ImportSource {
1814         function __construct( $string ) {
1815                 $this->mString = $string;
1816                 $this->mRead = false;
1817         }
1818
1819         /**
1820          * @return bool
1821          */
1822         function atEnd() {
1823                 return $this->mRead;
1824         }
1825
1826         /**
1827          * @return bool|string
1828          */
1829         function readChunk() {
1830                 if ( $this->atEnd() ) {
1831                         return false;
1832                 }
1833                 $this->mRead = true;
1834                 return $this->mString;
1835         }
1836 }
1837
1838 /**
1839  * Imports a XML dump from a file (either from file upload, files on disk, or HTTP)
1840  * @ingroup SpecialPage
1841  */
1842 class ImportStreamSource implements ImportSource {
1843         function __construct( $handle ) {
1844                 $this->mHandle = $handle;
1845         }
1846
1847         /**
1848          * @return bool
1849          */
1850         function atEnd() {
1851                 return feof( $this->mHandle );
1852         }
1853
1854         /**
1855          * @return string
1856          */
1857         function readChunk() {
1858                 return fread( $this->mHandle, 32768 );
1859         }
1860
1861         /**
1862          * @param string $filename
1863          * @return Status
1864          */
1865         static function newFromFile( $filename ) {
1866                 wfSuppressWarnings();
1867                 $file = fopen( $filename, 'rt' );
1868                 wfRestoreWarnings();
1869                 if ( !$file ) {
1870                         return Status::newFatal( "importcantopen" );
1871                 }
1872                 return Status::newGood( new ImportStreamSource( $file ) );
1873         }
1874
1875         /**
1876          * @param string $fieldname
1877          * @return Status
1878          */
1879         static function newFromUpload( $fieldname = "xmlimport" ) {
1880                 $upload =& $_FILES[$fieldname];
1881
1882                 if ( $upload === null || !$upload['name'] ) {
1883                         return Status::newFatal( 'importnofile' );
1884                 }
1885                 if ( !empty( $upload['error'] ) ) {
1886                         switch ( $upload['error'] ) {
1887                                 case 1:
1888                                         # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1889                                         return Status::newFatal( 'importuploaderrorsize' );
1890                                 case 2:
1891                                         # The uploaded file exceeds the MAX_FILE_SIZE directive that
1892                                         # was specified in the HTML form.
1893                                         return Status::newFatal( 'importuploaderrorsize' );
1894                                 case 3:
1895                                         # The uploaded file was only partially uploaded
1896                                         return Status::newFatal( 'importuploaderrorpartial' );
1897                                 case 6:
1898                                         # Missing a temporary folder.
1899                                         return Status::newFatal( 'importuploaderrortemp' );
1900                                 # case else: # Currently impossible
1901                         }
1902
1903                 }
1904                 $fname = $upload['tmp_name'];
1905                 if ( is_uploaded_file( $fname ) ) {
1906                         return ImportStreamSource::newFromFile( $fname );
1907                 } else {
1908                         return Status::newFatal( 'importnofile' );
1909                 }
1910         }
1911
1912         /**
1913          * @param string $url
1914          * @param string $method
1915          * @return Status
1916          */
1917         static function newFromURL( $url, $method = 'GET' ) {
1918                 wfDebug( __METHOD__ . ": opening $url\n" );
1919                 # Use the standard HTTP fetch function; it times out
1920                 # quicker and sorts out user-agent problems which might
1921                 # otherwise prevent importing from large sites, such
1922                 # as the Wikimedia cluster, etc.
1923                 $data = Http::request( $method, $url, array( 'followRedirects' => true ), __METHOD__ );
1924                 if ( $data !== false ) {
1925                         $file = tmpfile();
1926                         fwrite( $file, $data );
1927                         fflush( $file );
1928                         fseek( $file, 0 );
1929                         return Status::newGood( new ImportStreamSource( $file ) );
1930                 } else {
1931                         return Status::newFatal( 'importcantopen' );
1932                 }
1933         }
1934
1935         /**
1936          * @param string $interwiki
1937          * @param string $page
1938          * @param bool $history
1939          * @param bool $templates
1940          * @param int $pageLinkDepth
1941          * @return Status
1942          */
1943         public static function newFromInterwiki( $interwiki, $page, $history = false,
1944                 $templates = false, $pageLinkDepth = 0
1945         ) {
1946                 if ( $page == '' ) {
1947                         return Status::newFatal( 'import-noarticle' );
1948                 }
1949                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1950                 if ( is_null( $link ) || !$link->isExternal() ) {
1951                         return Status::newFatal( 'importbadinterwiki' );
1952                 } else {
1953                         $params = array();
1954                         if ( $history ) {
1955                                 $params['history'] = 1;
1956                         }
1957                         if ( $templates ) {
1958                                 $params['templates'] = 1;
1959                         }
1960                         if ( $pageLinkDepth ) {
1961                                 $params['pagelink-depth'] = $pageLinkDepth;
1962                         }
1963                         $url = $link->getFullURL( $params );
1964                         # For interwikis, use POST to avoid redirects.
1965                         return ImportStreamSource::newFromURL( $url, "POST" );
1966                 }
1967         }
1968 }