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::CDATA:
 519                         case XMLReader::SIGNIFICANT_WHITESPACE:
 520                                 $buffer .= $this->reader->value;
 521                                 break;
 522                         case XMLReader::END_ELEMENT:
 523                                 return $buffer;
 524                         }
 525                 }
 526
 527                 $this->reader->close();
 528                 return '';
 529         }
 530
 531         /**
 532          * Primary entry point
 533          * @throws MWException
 534          * @return bool
 535          */
 536         public function doImport() {
 537                 // Calls to reader->read need to be wrapped in calls to
 538                 // libxml_disable_entity_loader() to avoid local file
 539                 // inclusion attacks (bug 46932).
 540                 $oldDisable = libxml_disable_entity_loader( true );
 541                 $this->reader->read();
 542
 543                 if ( $this->reader->localName != 'mediawiki' ) {
 544                         libxml_disable_entity_loader( $oldDisable );
 545                         throw new MWException( "Expected <mediawiki> tag, got " .
 546                                 $this->reader->localName );
 547                 }
 548                 $this->debug( "<mediawiki> tag is correct." );
 549
 550                 $this->debug( "Starting primary dump processing loop." );
 551
 552                 $keepReading = $this->reader->read();
 553                 $skip = false;
 554                 $rethrow = null;
 555                 try {
 556                         while ( $keepReading ) {
 557                                 $tag = $this->reader->localName;
 558                                 $type = $this->reader->nodeType;
 559
 560                                 if ( !Hooks::run( 'ImportHandleToplevelXMLTag', array( $this ) ) ) {
 561                                         // Do nothing
 562                                 } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) {
 563                                         break;
 564                                 } elseif ( $tag == 'siteinfo' ) {
 565                                         $this->handleSiteInfo();
 566                                 } elseif ( $tag == 'page' ) {
 567                                         $this->handlePage();
 568                                 } elseif ( $tag == 'logitem' ) {
 569                                         $this->handleLogItem();
 570                                 } elseif ( $tag != '#text' ) {
 571                                         $this->warn( "Unhandled top-level XML tag $tag" );
 572
 573                                         $skip = true;
 574                                 }
 575
 576                                 if ( $skip ) {
 577                                         $keepReading = $this->reader->next();
 578                                         $skip = false;
 579                                         $this->debug( "Skip" );
 580                                 } else {
 581                                         $keepReading = $this->reader->read();
 582                                 }
 583                         }
 584                 } catch ( Exception $ex ) {
 585                         $rethrow = $ex;
 586                 }
 587
 588                 // finally
 589                 libxml_disable_entity_loader( $oldDisable );
 590                 $this->reader->close();
 591
 592                 if ( $rethrow ) {
 593                         throw $rethrow;
 594                 }
 595
 596                 return true;
 597         }
 598
 599         private function handleSiteInfo() {
 600                 $this->debug( "Enter site info handler." );
 601                 $siteInfo = array();
 602
 603                 // Fields that can just be stuffed in the siteInfo object
 604                 $normalFields = array( 'sitename', 'base', 'generator', 'case' );
 605
 606                 while ( $this->reader->read() ) {
 607                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 608                                         $this->reader->localName == 'siteinfo' ) {
 609                                 break;
 610                         }
 611
 612                         $tag = $this->reader->localName;
 613
 614                         if ( $tag == 'namespace' ) {
 615                                 $this->foreignNamespaces[$this->nodeAttribute( 'key' )] =
 616                                         $this->nodeContents();
 617                         } elseif ( in_array( $tag, $normalFields ) ) {
 618                                 $siteInfo[$tag] = $this->nodeContents();
 619                         }
 620                 }
 621
 622                 $siteInfo['_namespaces'] = $this->foreignNamespaces;
 623                 $this->siteInfoCallback( $siteInfo );
 624         }
 625
 626         private function handleLogItem() {
 627                 $this->debug( "Enter log item handler." );
 628                 $logInfo = array();
 629
 630                 // Fields that can just be stuffed in the pageInfo object
 631                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 632                                         'logtitle', 'params' );
 633
 634                 while ( $this->reader->read() ) {
 635                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 636                                         $this->reader->localName == 'logitem' ) {
 637                                 break;
 638                         }
 639
 640                         $tag = $this->reader->localName;
 641
 642                         if ( !Hooks::run( 'ImportHandleLogItemXMLTag', array(
 643                                 $this, $logInfo
 644                         ) ) ) {
 645                                 // Do nothing
 646                         } elseif ( in_array( $tag, $normalFields ) ) {
 647                                 $logInfo[$tag] = $this->nodeContents();
 648                         } elseif ( $tag == 'contributor' ) {
 649                                 $logInfo['contributor'] = $this->handleContributor();
 650                         } elseif ( $tag != '#text' ) {
 651                                 $this->warn( "Unhandled log-item XML tag $tag" );
 652                         }
 653                 }
 654
 655                 $this->processLogItem( $logInfo );
 656         }
 657
 658         /**
 659          * @param array $logInfo
 660          * @return bool|mixed
 661          */
 662         private function processLogItem( $logInfo ) {
 663
 664                 $revision = new WikiRevision( $this->config );
 665
 666                 if ( isset( $logInfo['id'] ) ) {
 667                         $revision->setID( $logInfo['id'] );
 668                 }
 669                 $revision->setType( $logInfo['type'] );
 670                 $revision->setAction( $logInfo['action'] );
 671                 if ( isset( $logInfo['timestamp'] ) ) {
 672                         $revision->setTimestamp( $logInfo['timestamp'] );
 673                 }
 674                 if ( isset( $logInfo['params'] ) ) {
 675                         $revision->setParams( $logInfo['params'] );
 676                 }
 677                 if ( isset( $logInfo['logtitle'] ) ) {
 678                         // @todo Using Title for non-local titles is a recipe for disaster.
 679                         // We should use ForeignTitle here instead.
 680                         $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 681                 }
 682
 683                 $revision->setNoUpdates( $this->mNoUpdates );
 684
 685                 if ( isset( $logInfo['comment'] ) ) {
 686                         $revision->setComment( $logInfo['comment'] );
 687                 }
 688
 689                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 690                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 691                 }
 692
 693                 if ( !isset( $logInfo['contributor']['username'] ) ) {
 694                         $revision->setUsername( 'Unknown user' );
 695                 } else {
 696                         $revision->setUserName( $logInfo['contributor']['username'] );
 697                 }
 698
 699                 return $this->logItemCallback( $revision );
 700         }
 701
 702         private function handlePage() {
 703                 // Handle page data.
 704                 $this->debug( "Enter page handler." );
 705                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 706
 707                 // Fields that can just be stuffed in the pageInfo object
 708                 $normalFields = array( 'title', 'ns', 'id', 'redirect', 'restrictions' );
 709
 710                 $skip = false;
 711                 $badTitle = false;
 712
 713                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 714                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 715                                         $this->reader->localName == 'page' ) {
 716                                 break;
 717                         }
 718
 719                         $skip = false;
 720
 721                         $tag = $this->reader->localName;
 722
 723                         if ( $badTitle ) {
 724                                 // The title is invalid, bail out of this page
 725                                 $skip = true;
 726                         } elseif ( !Hooks::run( 'ImportHandlePageXMLTag', array( $this,
 727                                                 &$pageInfo ) ) ) {
 728                                 // Do nothing
 729                         } elseif ( in_array( $tag, $normalFields ) ) {
 730                                 // An XML snippet:
 731                                 // <page>
 732                                 //     <id>123</id>
 733                                 //     <title>Page</title>
 734                                 //     <redirect title="NewTitle"/>
 735                                 //     ...
 736                                 // Because the redirect tag is built differently, we need special handling for that case.
 737                                 if ( $tag == 'redirect' ) {
 738                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 739                                 } else {
 740                                         $pageInfo[$tag] = $this->nodeContents();
 741                                 }
 742                         } elseif ( $tag == 'revision' || $tag == 'upload' ) {
 743                                 if ( !isset( $title ) ) {
 744                                         $title = $this->processTitle( $pageInfo['title'],
 745                                                 isset( $pageInfo['ns'] ) ? $pageInfo['ns'] : null );
 746
 747                                         // $title is either an array of two titles or false.
 748                                         if ( is_array( $title ) ) {
 749                                                 $this->pageCallback( $title );
 750                                                 list( $pageInfo['_title'], $foreignTitle ) = $title;
 751                                         } else {
 752                                                 $badTitle = true;
 753                                                 $skip = true;
 754                                         }
 755                                 }
 756
 757                                 if ( $title ) {
 758                                         if ( $tag == 'revision' ) {
 759                                                 $this->handleRevision( $pageInfo );
 760                                         } else {
 761                                                 $this->handleUpload( $pageInfo );
 762                                         }
 763                                 }
 764                         } elseif ( $tag != '#text' ) {
 765                                 $this->warn( "Unhandled page XML tag $tag" );
 766                                 $skip = true;
 767                         }
 768                 }
 769
 770                 // @note $pageInfo is only set if a valid $title is processed above with
 771                 //       no error. If we have a valid $title, then pageCallback is called
 772                 //       above, $pageInfo['title'] is set and we do pageOutCallback here.
 773                 //       If $pageInfo['_title'] is not set, then $foreignTitle is also not
 774                 //       set since they both come from $title above.
 775                 if ( array_key_exists( '_title', $pageInfo ) ) {
 776                         $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
 777                                         $pageInfo['revisionCount'],
 778                                         $pageInfo['successfulRevisionCount'],
 779                                         $pageInfo );
 780                 }
 781         }
 782
 783         /**
 784          * @param array $pageInfo
 785          */
 786         private function handleRevision( &$pageInfo ) {
 787                 $this->debug( "Enter revision handler" );
 788                 $revisionInfo = array();
 789
 790                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' );
 791
 792                 $skip = false;
 793
 794                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 795                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 796                                         $this->reader->localName == 'revision' ) {
 797                                 break;
 798                         }
 799
 800                         $tag = $this->reader->localName;
 801
 802                         if ( !Hooks::run( 'ImportHandleRevisionXMLTag', array(
 803                                 $this, $pageInfo, $revisionInfo
 804                         ) ) ) {
 805                                 // Do nothing
 806                         } elseif ( in_array( $tag, $normalFields ) ) {
 807                                 $revisionInfo[$tag] = $this->nodeContents();
 808                         } elseif ( $tag == 'contributor' ) {
 809                                 $revisionInfo['contributor'] = $this->handleContributor();
 810                         } elseif ( $tag != '#text' ) {
 811                                 $this->warn( "Unhandled revision XML tag $tag" );
 812                                 $skip = true;
 813                         }
 814                 }
 815
 816                 $pageInfo['revisionCount']++;
 817                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 818                         $pageInfo['successfulRevisionCount']++;
 819                 }
 820         }
 821
 822         /**
 823          * @param array $pageInfo
 824          * @param array $revisionInfo
 825          * @return bool|mixed
 826          */
 827         private function processRevision( $pageInfo, $revisionInfo ) {
 828                 $revision = new WikiRevision( $this->config );
 829
 830                 if ( isset( $revisionInfo['id'] ) ) {
 831                         $revision->setID( $revisionInfo['id'] );
 832                 }
 833                 if ( isset( $revisionInfo['model'] ) ) {
 834                         $revision->setModel( $revisionInfo['model'] );
 835                 }
 836                 if ( isset( $revisionInfo['format'] ) ) {
 837                         $revision->setFormat( $revisionInfo['format'] );
 838                 }
 839                 $revision->setTitle( $pageInfo['_title'] );
 840
 841                 if ( isset( $revisionInfo['text'] ) ) {
 842                         $handler = $revision->getContentHandler();
 843                         $text = $handler->importTransform(
 844                                 $revisionInfo['text'],
 845                                 $revision->getFormat() );
 846
 847                         $revision->setText( $text );
 848                 }
 849                 if ( isset( $revisionInfo['timestamp'] ) ) {
 850                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 851                 } else {
 852                         $revision->setTimestamp( wfTimestampNow() );
 853                 }
 854
 855                 if ( isset( $revisionInfo['comment'] ) ) {
 856                         $revision->setComment( $revisionInfo['comment'] );
 857                 }
 858
 859                 if ( isset( $revisionInfo['minor'] ) ) {
 860                         $revision->setMinor( true );
 861                 }
 862                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 863                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 864                 } elseif ( isset( $revisionInfo['contributor']['username'] ) ) {
 865                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 866                 } else {
 867                         $revision->setUserName( 'Unknown user' );
 868                 }
 869                 $revision->setNoUpdates( $this->mNoUpdates );
 870
 871                 return $this->revisionCallback( $revision );
 872         }
 873
 874         /**
 875          * @param array $pageInfo
 876          * @return mixed
 877          */
 878         private function handleUpload( &$pageInfo ) {
 879                 $this->debug( "Enter upload handler" );
 880                 $uploadInfo = array();
 881
 882                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 883                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 884
 885                 $skip = false;
 886
 887                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 888                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 889                                         $this->reader->localName == 'upload' ) {
 890                                 break;
 891                         }
 892
 893                         $tag = $this->reader->localName;
 894
 895                         if ( !Hooks::run( 'ImportHandleUploadXMLTag', array(
 896                                 $this, $pageInfo
 897                         ) ) ) {
 898                                 // Do nothing
 899                         } elseif ( in_array( $tag, $normalFields ) ) {
 900                                 $uploadInfo[$tag] = $this->nodeContents();
 901                         } elseif ( $tag == 'contributor' ) {
 902                                 $uploadInfo['contributor'] = $this->handleContributor();
 903                         } elseif ( $tag == 'contents' ) {
 904                                 $contents = $this->nodeContents();
 905                                 $encoding = $this->reader->getAttribute( 'encoding' );
 906                                 if ( $encoding === 'base64' ) {
 907                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 908                                         $uploadInfo['isTempSrc'] = true;
 909                                 }
 910                         } elseif ( $tag != '#text' ) {
 911                                 $this->warn( "Unhandled upload XML tag $tag" );
 912                                 $skip = true;
 913                         }
 914                 }
 915
 916                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 917                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 918                         if ( file_exists( $path ) ) {
 919                                 $uploadInfo['fileSrc'] = $path;
 920                                 $uploadInfo['isTempSrc'] = false;
 921                         }
 922                 }
 923
 924                 if ( $this->mImportUploads ) {
 925                         return $this->processUpload( $pageInfo, $uploadInfo );
 926                 }
 927         }
 928
 929         /**
 930          * @param string $contents
 931          * @return string
 932          */
 933         private function dumpTemp( $contents ) {
 934                 $filename = tempnam( wfTempDir(), 'importupload' );
 935                 file_put_contents( $filename, $contents );
 936                 return $filename;
 937         }
 938
 939         /**
 940          * @param array $pageInfo
 941          * @param array $uploadInfo
 942          * @return mixed
 943          */
 944         private function processUpload( $pageInfo, $uploadInfo ) {
 945                 $revision = new WikiRevision( $this->config );
 946                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 947
 948                 $revision->setTitle( $pageInfo['_title'] );
 949                 $revision->setID( $pageInfo['id'] );
 950                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 951                 $revision->setText( $text );
 952                 $revision->setFilename( $uploadInfo['filename'] );
 953                 if ( isset( $uploadInfo['archivename'] ) ) {
 954                         $revision->setArchiveName( $uploadInfo['archivename'] );
 955                 }
 956                 $revision->setSrc( $uploadInfo['src'] );
 957                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 958                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 959                                 !empty( $uploadInfo['isTempSrc'] ) );
 960                 }
 961                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 962                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 963                 }
 964                 $revision->setSize( intval( $uploadInfo['size'] ) );
 965                 $revision->setComment( $uploadInfo['comment'] );
 966
 967                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 968                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 969                 }
 970                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 971                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 972                 }
 973                 $revision->setNoUpdates( $this->mNoUpdates );
 974
 975                 return call_user_func( $this->mUploadCallback, $revision );
 976         }
 977
 978         /**
 979          * @return array
 980          */
 981         private function handleContributor() {
 982                 $fields = array( 'id', 'ip', 'username' );
 983                 $info = array();
 984
 985                 if ( $this->reader->isEmptyElement ) {
 986                         return $info;
 987                 }
 988                 while ( $this->reader->read() ) {
 989                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 990                                         $this->reader->localName == 'contributor' ) {
 991                                 break;
 992                         }
 993
 994                         $tag = $this->reader->localName;
 995
 996                         if ( in_array( $tag, $fields ) ) {
 997                                 $info[$tag] = $this->nodeContents();
 998                         }
 999                 }
1000
1001                 return $info;
1002         }
1003
1004         /**
1005          * @param string $text
1006          * @param string|null $ns
1007          * @return array|bool
1008          */
1009         private function processTitle( $text, $ns = null ) {
1010                 if ( is_null( $this->foreignNamespaces ) ) {
1011                         $foreignTitleFactory = new NaiveForeignTitleFactory();
1012                 } else {
1013                         $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
1014                                 $this->foreignNamespaces );
1015                 }
1016
1017                 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
1018                         intval( $ns ) );
1019
1020                 $title = $this->importTitleFactory->createTitleFromForeignTitle(
1021                         $foreignTitle );
1022
1023                 $commandLineMode = $this->config->get( 'CommandLineMode' );
1024                 if ( is_null( $title ) ) {
1025                         # Invalid page title? Ignore the page
1026                         $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
1027                         return false;
1028                 } elseif ( $title->isExternal() ) {
1029                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
1030                         return false;
1031                 } elseif ( !$title->canExist() ) {
1032                         $this->notice( 'import-error-special', $title->getPrefixedText() );
1033                         return false;
1034                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
1035                         # Do not import if the importing wiki user cannot edit this page
1036                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
1037                         return false;
1038                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
1039                         # Do not import if the importing wiki user cannot create this page
1040                         $this->notice( 'import-error-create', $title->getPrefixedText() );
1041                         return false;
1042                 }
1043
1044                 return array( $title, $foreignTitle );
1045         }
1046 }
1047
1048 /** This is a horrible hack used to keep source compatibility */
1049 class UploadSourceAdapter {
1050         /** @var array */
1051         public static $sourceRegistrations = array();
1052
1053         /** @var string */
1054         private $mSource;
1055
1056         /** @var string */
1057         private $mBuffer;
1058
1059         /** @var int */
1060         private $mPosition;
1061
1062         /**
1063          * @param ImportSource $source
1064          * @return string
1065          */
1066         static function registerSource( ImportSource $source ) {
1067                 $id = wfRandomString();
1068
1069                 self::$sourceRegistrations[$id] = $source;
1070
1071                 return $id;
1072         }
1073
1074         /**
1075          * @param string $path
1076          * @param string $mode
1077          * @param array $options
1078          * @param string $opened_path
1079          * @return bool
1080          */
1081         function stream_open( $path, $mode, $options, &$opened_path ) {
1082                 $url = parse_url( $path );
1083                 $id = $url['host'];
1084
1085                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
1086                         return false;
1087                 }
1088
1089                 $this->mSource = self::$sourceRegistrations[$id];
1090
1091                 return true;
1092         }
1093
1094         /**
1095          * @param int $count
1096          * @return string
1097          */
1098         function stream_read( $count ) {
1099                 $return = '';
1100                 $leave = false;
1101
1102                 while ( !$leave && !$this->mSource->atEnd() &&
1103                                 strlen( $this->mBuffer ) < $count ) {
1104                         $read = $this->mSource->readChunk();
1105
1106                         if ( !strlen( $read ) ) {
1107                                 $leave = true;
1108                         }
1109
1110                         $this->mBuffer .= $read;
1111                 }
1112
1113                 if ( strlen( $this->mBuffer ) ) {
1114                         $return = substr( $this->mBuffer, 0, $count );
1115                         $this->mBuffer = substr( $this->mBuffer, $count );
1116                 }
1117
1118                 $this->mPosition += strlen( $return );
1119
1120                 return $return;
1121         }
1122
1123         /**
1124          * @param string $data
1125          * @return bool
1126          */
1127         function stream_write( $data ) {
1128                 return false;
1129         }
1130
1131         /**
1132          * @return mixed
1133          */
1134         function stream_tell() {
1135                 return $this->mPosition;
1136         }
1137
1138         /**
1139          * @return bool
1140          */
1141         function stream_eof() {
1142                 return $this->mSource->atEnd();
1143         }
1144
1145         /**
1146          * @return array
1147          */
1148         function url_stat() {
1149                 $result = array();
1150
1151                 $result['dev'] = $result[0] = 0;
1152                 $result['ino'] = $result[1] = 0;
1153                 $result['mode'] = $result[2] = 0;
1154                 $result['nlink'] = $result[3] = 0;
1155                 $result['uid'] = $result[4] = 0;
1156                 $result['gid'] = $result[5] = 0;
1157                 $result['rdev'] = $result[6] = 0;
1158                 $result['size'] = $result[7] = 0;
1159                 $result['atime'] = $result[8] = 0;
1160                 $result['mtime'] = $result[9] = 0;
1161                 $result['ctime'] = $result[10] = 0;
1162                 $result['blksize'] = $result[11] = 0;
1163                 $result['blocks'] = $result[12] = 0;
1164
1165                 return $result;
1166         }
1167 }
1168
1169 /**
1170  * @todo document (e.g. one-sentence class description).
1171  * @ingroup SpecialPage
1172  */
1173 class WikiRevision {
1174         /** @todo Unused? */
1175         public $importer = null;
1176
1177         /** @var Title */
1178         public $title = null;
1179
1180         /** @var int */
1181         public $id = 0;
1182
1183         /** @var string */
1184         public $timestamp = "20010115000000";
1185
1186         /**
1187          * @var int
1188          * @todo Can't find any uses. Public, because that's suspicious. Get clarity. */
1189         public $user = 0;
1190
1191         /** @var string */
1192         public $user_text = "";
1193
1194         /** @var string */
1195         public $model = null;
1196
1197         /** @var string */
1198         public $format = null;
1199
1200         /** @var string */
1201         public $text = "";
1202
1203         /** @var int */
1204         protected $size;
1205
1206         /** @var Content */
1207         public $content = null;
1208
1209         /** @var ContentHandler */
1210         protected $contentHandler = null;
1211
1212         /** @var string */
1213         public $comment = "";
1214
1215         /** @var bool */
1216         public $minor = false;
1217
1218         /** @var string */
1219         public $type = "";
1220
1221         /** @var string */
1222         public $action = "";
1223
1224         /** @var string */
1225         public $params = "";
1226
1227         /** @var string */
1228         public $fileSrc = '';
1229
1230         /** @var bool|string */
1231         public $sha1base36 = false;
1232
1233         /**
1234          * @var bool
1235          * @todo Unused?
1236          */
1237         public $isTemp = false;
1238
1239         /** @var string */
1240         public $archiveName = '';
1241
1242         protected $filename;
1243
1244         /** @var mixed */
1245         protected $src;
1246
1247         /** @todo Unused? */
1248         public $fileIsTemp;
1249
1250         /** @var bool */
1251         private $mNoUpdates = false;
1252
1253         /** @var Config $config */
1254         private $config;
1255
1256         public function __construct( Config $config ) {
1257                 $this->config = $config;
1258         }
1259
1260         /**
1261          * @param Title $title
1262          * @throws MWException
1263          */
1264         function setTitle( $title ) {
1265                 if ( is_object( $title ) ) {
1266                         $this->title = $title;
1267                 } elseif ( is_null( $title ) ) {
1268                         throw new MWException( "WikiRevision given a null title in import. "
1269                                 . "You may need to adjust \$wgLegalTitleChars." );
1270                 } else {
1271                         throw new MWException( "WikiRevision given non-object title in import." );
1272                 }
1273         }
1274
1275         /**
1276          * @param int $id
1277          */
1278         function setID( $id ) {
1279                 $this->id = $id;
1280         }
1281
1282         /**
1283          * @param string $ts
1284          */
1285         function setTimestamp( $ts ) {
1286                 # 2003-08-05T18:30:02Z
1287                 $this->timestamp = wfTimestamp( TS_MW, $ts );
1288         }
1289
1290         /**
1291          * @param string $user
1292          */
1293         function setUsername( $user ) {
1294                 $this->user_text = $user;
1295         }
1296
1297         /**
1298          * @param string $ip
1299          */
1300         function setUserIP( $ip ) {
1301                 $this->user_text = $ip;
1302         }
1303
1304         /**
1305          * @param string $model
1306          */
1307         function setModel( $model ) {
1308                 $this->model = $model;
1309         }
1310
1311         /**
1312          * @param string $format
1313          */
1314         function setFormat( $format ) {
1315                 $this->format = $format;
1316         }
1317
1318         /**
1319          * @param string $text
1320          */
1321         function setText( $text ) {
1322                 $this->text = $text;
1323         }
1324
1325         /**
1326          * @param string $text
1327          */
1328         function setComment( $text ) {
1329                 $this->comment = $text;
1330         }
1331
1332         /**
1333          * @param bool $minor
1334          */
1335         function setMinor( $minor ) {
1336                 $this->minor = (bool)$minor;
1337         }
1338
1339         /**
1340          * @param mixed $src
1341          */
1342         function setSrc( $src ) {
1343                 $this->src = $src;
1344         }
1345
1346         /**
1347          * @param string $src
1348          * @param bool $isTemp
1349          */
1350         function setFileSrc( $src, $isTemp ) {
1351                 $this->fileSrc = $src;
1352                 $this->fileIsTemp = $isTemp;
1353         }
1354
1355         /**
1356          * @param string $sha1base36
1357          */
1358         function setSha1Base36( $sha1base36 ) {
1359                 $this->sha1base36 = $sha1base36;
1360         }
1361
1362         /**
1363          * @param string $filename
1364          */
1365         function setFilename( $filename ) {
1366                 $this->filename = $filename;
1367         }
1368
1369         /**
1370          * @param string $archiveName
1371          */
1372         function setArchiveName( $archiveName ) {
1373                 $this->archiveName = $archiveName;
1374         }
1375
1376         /**
1377          * @param int $size
1378          */
1379         function setSize( $size ) {
1380                 $this->size = intval( $size );
1381         }
1382
1383         /**
1384          * @param string $type
1385          */
1386         function setType( $type ) {
1387                 $this->type = $type;
1388         }
1389
1390         /**
1391          * @param string $action
1392          */
1393         function setAction( $action ) {
1394                 $this->action = $action;
1395         }
1396
1397         /**
1398          * @param array $params
1399          */
1400         function setParams( $params ) {
1401                 $this->params = $params;
1402         }
1403
1404         /**
1405          * @param bool $noupdates
1406          */
1407         public function setNoUpdates( $noupdates ) {
1408                 $this->mNoUpdates = $noupdates;
1409         }
1410
1411         /**
1412          * @return Title
1413          */
1414         function getTitle() {
1415                 return $this->title;
1416         }
1417
1418         /**
1419          * @return int
1420          */
1421         function getID() {
1422                 return $this->id;
1423         }
1424
1425         /**
1426          * @return string
1427          */
1428         function getTimestamp() {
1429                 return $this->timestamp;
1430         }
1431
1432         /**
1433          * @return string
1434          */
1435         function getUser() {
1436                 return $this->user_text;
1437         }
1438
1439         /**
1440          * @return string
1441          *
1442          * @deprecated Since 1.21, use getContent() instead.
1443          */
1444         function getText() {
1445                 ContentHandler::deprecated( __METHOD__, '1.21' );
1446
1447                 return $this->text;
1448         }
1449
1450         /**
1451          * @return ContentHandler
1452          */
1453         function getContentHandler() {
1454                 if ( is_null( $this->contentHandler ) ) {
1455                         $this->contentHandler = ContentHandler::getForModelID( $this->getModel() );
1456                 }
1457
1458                 return $this->contentHandler;
1459         }
1460
1461         /**
1462          * @return Content
1463          */
1464         function getContent() {
1465                 if ( is_null( $this->content ) ) {
1466                         $handler = $this->getContentHandler();
1467                         $this->content = $handler->unserializeContent( $this->text, $this->getFormat() );
1468                 }
1469
1470                 return $this->content;
1471         }
1472
1473         /**
1474          * @return string
1475          */
1476         function getModel() {
1477                 if ( is_null( $this->model ) ) {
1478                         $this->model = $this->getTitle()->getContentModel();
1479                 }
1480
1481                 return $this->model;
1482         }
1483
1484         /**
1485          * @return string
1486          */
1487         function getFormat() {
1488                 if ( is_null( $this->format ) ) {
1489                         $this->format = $this->getContentHandler()->getDefaultFormat();
1490                 }
1491
1492                 return $this->format;
1493         }
1494
1495         /**
1496          * @return string
1497          */
1498         function getComment() {
1499                 return $this->comment;
1500         }
1501
1502         /**
1503          * @return bool
1504          */
1505         function getMinor() {
1506                 return $this->minor;
1507         }
1508
1509         /**
1510          * @return mixed
1511          */
1512         function getSrc() {
1513                 return $this->src;
1514         }
1515
1516         /**
1517          * @return bool|string
1518          */
1519         function getSha1() {
1520                 if ( $this->sha1base36 ) {
1521                         return Wikimedia\base_convert( $this->sha1base36, 36, 16 );
1522                 }
1523                 return false;
1524         }
1525
1526         /**
1527          * @return string
1528          */
1529         function getFileSrc() {
1530                 return $this->fileSrc;
1531         }
1532
1533         /**
1534          * @return bool
1535          */
1536         function isTempSrc() {
1537                 return $this->isTemp;
1538         }
1539
1540         /**
1541          * @return mixed
1542          */
1543         function getFilename() {
1544                 return $this->filename;
1545         }
1546
1547         /**
1548          * @return string
1549          */
1550         function getArchiveName() {
1551                 return $this->archiveName;
1552         }
1553
1554         /**
1555          * @return mixed
1556          */
1557         function getSize() {
1558                 return $this->size;
1559         }
1560
1561         /**
1562          * @return string
1563          */
1564         function getType() {
1565                 return $this->type;
1566         }
1567
1568         /**
1569          * @return string
1570          */
1571         function getAction() {
1572                 return $this->action;
1573         }
1574
1575         /**
1576          * @return string
1577          */
1578         function getParams() {
1579                 return $this->params;
1580         }
1581
1582         /**
1583          * @return bool
1584          */
1585         function importOldRevision() {
1586                 $dbw = wfGetDB( DB_MASTER );
1587
1588                 # Sneak a single revision into place
1589                 $user = User::newFromName( $this->getUser() );
1590                 if ( $user ) {
1591                         $userId = intval( $user->getId() );
1592                         $userText = $user->getName();
1593                         $userObj = $user;
1594                 } else {
1595                         $userId = 0;
1596                         $userText = $this->getUser();
1597                         $userObj = new User;
1598                 }
1599
1600                 // avoid memory leak...?
1601                 Title::clearCaches();
1602
1603                 $page = WikiPage::factory( $this->title );
1604                 $page->loadPageData( 'fromdbmaster' );
1605                 if ( !$page->exists() ) {
1606                         # must create the page...
1607                         $pageId = $page->insertOn( $dbw );
1608                         $created = true;
1609                         $oldcountable = null;
1610                 } else {
1611                         $pageId = $page->getId();
1612                         $created = false;
1613
1614                         $prior = $dbw->selectField( 'revision', '1',
1615                                 array( 'rev_page' => $pageId,
1616                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1617                                         'rev_user_text' => $userText,
1618                                         'rev_comment' => $this->getComment() ),
1619                                 __METHOD__
1620                         );
1621                         if ( $prior ) {
1622                                 // @todo FIXME: This could fail slightly for multiple matches :P
1623                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1624                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1625                                 return false;
1626                         }
1627                 }
1628
1629                 // Select previous version to make size diffs correct
1630                 $prevId = $dbw->selectField( 'revision', 'rev_id',
1631                         array(
1632                                 'rev_page' => $pageId,
1633                                 'rev_timestamp <= ' . $dbw->timestamp( $this->timestamp ),
1634                         ),
1635                         __METHOD__,
1636                         array( 'ORDER BY' => array(
1637                                         'rev_timestamp DESC',
1638                                         'rev_id DESC', // timestamp is not unique per page
1639                                 )
1640                         )
1641                 );
1642
1643                 # @todo FIXME: Use original rev_id optionally (better for backups)
1644                 # Insert the row
1645                 $revision = new Revision( array(
1646                         'title' => $this->title,
1647                         'page' => $pageId,
1648                         'content_model' => $this->getModel(),
1649                         'content_format' => $this->getFormat(),
1650                         // XXX: just set 'content' => $this->getContent()?
1651                         'text' => $this->getContent()->serialize( $this->getFormat() ),
1652                         'comment' => $this->getComment(),
1653                         'user' => $userId,
1654                         'user_text' => $userText,
1655                         'timestamp' => $this->timestamp,
1656                         'minor_edit' => $this->minor,
1657                         'parent_id' => $prevId,
1658                         ) );
1659                 $revision->insertOn( $dbw );
1660                 $changed = $page->updateIfNewerOn( $dbw, $revision );
1661
1662                 if ( $changed !== false && !$this->mNoUpdates ) {
1663                         wfDebug( __METHOD__ . ": running updates\n" );
1664                         // countable/oldcountable stuff is handled in WikiImporter::finishImportPage
1665                         $page->doEditUpdates(
1666                                 $revision,
1667                                 $userObj,
1668                                 array( 'created' => $created, 'oldcountable' => 'no-change' )
1669                         );
1670                 }
1671
1672                 return true;
1673         }
1674
1675         function importLogItem() {
1676                 $dbw = wfGetDB( DB_MASTER );
1677
1678                 $user = User::newFromName( $this->getUser() );
1679                 if ( $user ) {
1680                         $userId = intval( $user->getId() );
1681                         $userText = $user->getName();
1682                 } else {
1683                         $userId = 0;
1684                         $userText = $this->getUser();
1685                 }
1686
1687                 # @todo FIXME: This will not record autoblocks
1688                 if ( !$this->getTitle() ) {
1689                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1690                                 $this->timestamp . "\n" );
1691                         return;
1692                 }
1693                 # Check if it exists already
1694                 // @todo FIXME: Use original log ID (better for backups)
1695                 $prior = $dbw->selectField( 'logging', '1',
1696                         array( 'log_type' => $this->getType(),
1697                                 'log_action' => $this->getAction(),
1698                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1699                                 'log_namespace' => $this->getTitle()->getNamespace(),
1700                                 'log_title' => $this->getTitle()->getDBkey(),
1701                                 'log_comment' => $this->getComment(),
1702                                 # 'log_user_text' => $this->user_text,
1703                                 'log_params' => $this->params ),
1704                         __METHOD__
1705                 );
1706                 // @todo FIXME: This could fail slightly for multiple matches :P
1707                 if ( $prior ) {
1708                         wfDebug( __METHOD__
1709                                 . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp "
1710                                 . $this->timestamp . "\n" );
1711                         return;
1712                 }
1713                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1714                 $data = array(
1715                         'log_id' => $log_id,
1716                         'log_type' => $this->type,
1717                         'log_action' => $this->action,
1718                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1719                         'log_user' =>  $userId,
1720                         'log_user_text' => $userText,
1721                         'log_namespace' => $this->getTitle()->getNamespace(),
1722                         'log_title' => $this->getTitle()->getDBkey(),
1723                         'log_comment' => $this->getComment(),
1724                         'log_params' => $this->params
1725                 );
1726                 $dbw->insert( 'logging', $data, __METHOD__ );
1727         }
1728
1729         /**
1730          * @return bool
1731          */
1732         function importUpload() {
1733                 # Construct a file
1734                 $archiveName = $this->getArchiveName();
1735                 if ( $archiveName ) {
1736                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1737                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1738                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1739                 } else {
1740                         $file = wfLocalFile( $this->getTitle() );
1741                         $file->load( File::READ_LATEST );
1742                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1743                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1744                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1745                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1746                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1747                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1748                         }
1749                 }
1750                 if ( !$file ) {
1751                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1752                         return false;
1753                 }
1754
1755                 # Get the file source or download if necessary
1756                 $source = $this->getFileSrc();
1757                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1758                 if ( !$source ) {
1759                         $source = $this->downloadSource();
1760                         $flags |= File::DELETE_SOURCE;
1761                 }
1762                 if ( !$source ) {
1763                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1764                         return false;
1765                 }
1766                 $sha1 = $this->getSha1();
1767                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1768                         if ( $flags & File::DELETE_SOURCE ) {
1769                                 # Broken file; delete it if it is a temporary file
1770                                 unlink( $source );
1771                         }
1772                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1773                         return false;
1774                 }
1775
1776                 $user = User::newFromName( $this->user_text );
1777
1778                 # Do the actual upload
1779                 if ( $archiveName ) {
1780                         $status = $file->uploadOld( $source, $archiveName,
1781                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1782                 } else {
1783                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1784                                 $flags, false, $this->getTimestamp(), $user );
1785                 }
1786
1787                 if ( $status->isGood() ) {
1788                         wfDebug( __METHOD__ . ": Successful\n" );
1789                         return true;
1790                 } else {
1791                         wfDebug( __METHOD__ . ': failed: ' . $status->getHTML() . "\n" );
1792                         return false;
1793                 }
1794         }
1795
1796         /**
1797          * @return bool|string
1798          */
1799         function downloadSource() {
1800                 if ( !$this->config->get( 'EnableUploads' ) ) {
1801                         return false;
1802                 }
1803
1804                 $tempo = tempnam( wfTempDir(), 'download' );
1805                 $f = fopen( $tempo, 'wb' );
1806                 if ( !$f ) {
1807                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1808                         return false;
1809                 }
1810
1811                 // @todo FIXME!
1812                 $src = $this->getSrc();
1813                 $data = Http::get( $src, array(), __METHOD__ );
1814                 if ( !$data ) {
1815                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1816                         fclose( $f );
1817                         unlink( $tempo );
1818                         return false;
1819                 }
1820
1821                 fwrite( $f, $data );
1822                 fclose( $f );
1823
1824                 return $tempo;
1825         }
1826
1827 }
1828
1829 /**
1830  * Source interface for XML import.
1831  */
1832 interface ImportSource {
1833
1834         /**
1835          * Indicates whether the end of the input has been reached.
1836          * Will return true after a finite number of calls to readChunk.
1837          *
1838          * @return bool true if there is no more input, false otherwise.
1839          */
1840         function atEnd();
1841
1842         /**
1843          * Return a chunk of the input, as a (possibly empty) string.
1844          * When the end of input is reached, readChunk() returns false.
1845          * If atEnd() returns false, readChunk() will return a string.
1846          * If atEnd() returns true, readChunk() will return false.
1847          *
1848          * @return bool|string
1849          */
1850         function readChunk();
1851 }
1852
1853 /**
1854  * Used for importing XML dumps where the content of the dump is in a string.
1855  * This class is ineffecient, and should only be used for small dumps.
1856  * For larger dumps, ImportStreamSource should be used instead.
1857  *
1858  * @ingroup SpecialPage
1859  */
1860 class ImportStringSource implements ImportSource {
1861         function __construct( $string ) {
1862                 $this->mString = $string;
1863                 $this->mRead = false;
1864         }
1865
1866         /**
1867          * @return bool
1868          */
1869         function atEnd() {
1870                 return $this->mRead;
1871         }
1872
1873         /**
1874          * @return bool|string
1875          */
1876         function readChunk() {
1877                 if ( $this->atEnd() ) {
1878                         return false;
1879                 }
1880                 $this->mRead = true;
1881                 return $this->mString;
1882         }
1883 }
1884
1885 /**
1886  * Imports a XML dump from a file (either from file upload, files on disk, or HTTP)
1887  * @ingroup SpecialPage
1888  */
1889 class ImportStreamSource implements ImportSource {
1890         function __construct( $handle ) {
1891                 $this->mHandle = $handle;
1892         }
1893
1894         /**
1895          * @return bool
1896          */
1897         function atEnd() {
1898                 return feof( $this->mHandle );
1899         }
1900
1901         /**
1902          * @return string
1903          */
1904         function readChunk() {
1905                 return fread( $this->mHandle, 32768 );
1906         }
1907
1908         /**
1909          * @param string $filename
1910          * @return Status
1911          */
1912         static function newFromFile( $filename ) {
1913                 MediaWiki\suppressWarnings();
1914                 $file = fopen( $filename, 'rt' );
1915                 MediaWiki\restoreWarnings();
1916                 if ( !$file ) {
1917                         return Status::newFatal( "importcantopen" );
1918                 }
1919                 return Status::newGood( new ImportStreamSource( $file ) );
1920         }
1921
1922         /**
1923          * @param string $fieldname
1924          * @return Status
1925          */
1926         static function newFromUpload( $fieldname = "xmlimport" ) {
1927                 $upload =& $_FILES[$fieldname];
1928
1929                 if ( $upload === null || !$upload['name'] ) {
1930                         return Status::newFatal( 'importnofile' );
1931                 }
1932                 if ( !empty( $upload['error'] ) ) {
1933                         switch ( $upload['error'] ) {
1934                                 case 1:
1935                                         # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1936                                         return Status::newFatal( 'importuploaderrorsize' );
1937                                 case 2:
1938                                         # The uploaded file exceeds the MAX_FILE_SIZE directive that
1939                                         # was specified in the HTML form.
1940                                         return Status::newFatal( 'importuploaderrorsize' );
1941                                 case 3:
1942                                         # The uploaded file was only partially uploaded
1943                                         return Status::newFatal( 'importuploaderrorpartial' );
1944                                 case 6:
1945                                         # Missing a temporary folder.
1946                                         return Status::newFatal( 'importuploaderrortemp' );
1947                                 # case else: # Currently impossible
1948                         }
1949
1950                 }
1951                 $fname = $upload['tmp_name'];
1952                 if ( is_uploaded_file( $fname ) ) {
1953                         return ImportStreamSource::newFromFile( $fname );
1954                 } else {
1955                         return Status::newFatal( 'importnofile' );
1956                 }
1957         }
1958
1959         /**
1960          * @param string $url
1961          * @param string $method
1962          * @return Status
1963          */
1964         static function newFromURL( $url, $method = 'GET' ) {
1965                 wfDebug( __METHOD__ . ": opening $url\n" );
1966                 # Use the standard HTTP fetch function; it times out
1967                 # quicker and sorts out user-agent problems which might
1968                 # otherwise prevent importing from large sites, such
1969                 # as the Wikimedia cluster, etc.
1970                 $data = Http::request( $method, $url, array( 'followRedirects' => true ), __METHOD__ );
1971                 if ( $data !== false ) {
1972                         $file = tmpfile();
1973                         fwrite( $file, $data );
1974                         fflush( $file );
1975                         fseek( $file, 0 );
1976                         return Status::newGood( new ImportStreamSource( $file ) );
1977                 } else {
1978                         return Status::newFatal( 'importcantopen' );
1979                 }
1980         }
1981
1982         /**
1983          * @param string $interwiki
1984          * @param string $page
1985          * @param bool $history
1986          * @param bool $templates
1987          * @param int $pageLinkDepth
1988          * @return Status
1989          */
1990         public static function newFromInterwiki( $interwiki, $page, $history = false,
1991                 $templates = false, $pageLinkDepth = 0
1992         ) {
1993                 if ( $page == '' ) {
1994                         return Status::newFatal( 'import-noarticle' );
1995                 }
1996
1997                 # Look up the first interwiki prefix, and let the foreign site handle
1998                 # subsequent interwiki prefixes
1999                 $firstIwPrefix = strtok( $interwiki, ':' );
2000                 $firstIw = Interwiki::fetch( $firstIwPrefix );
2001                 if ( !$firstIw ) {
2002                         return Status::newFatal( 'importbadinterwiki' );
2003                 }
2004
2005                 $additionalIwPrefixes = strtok( '' );
2006                 if ( $additionalIwPrefixes ) {
2007                         $additionalIwPrefixes .= ':';
2008                 }
2009                 # Have to do a DB-key replacement ourselves; otherwise spaces get
2010                 # URL-encoded to +, which is wrong in this case. Similar to logic in
2011                 # Title::getLocalURL
2012                 $link = $firstIw->getURL( strtr( "${additionalIwPrefixes}Special:Export/$page",
2013                         ' ', '_' ) );
2014
2015                 $params = array();
2016                 if ( $history ) {
2017                         $params['history'] = 1;
2018                 }
2019                 if ( $templates ) {
2020                         $params['templates'] = 1;
2021                 }
2022                 if ( $pageLinkDepth ) {
2023                         $params['pagelink-depth'] = $pageLinkDepth;
2024                 }
2025
2026                 $url = wfAppendQuery( $link, $params );
2027                 # For interwikis, use POST to avoid redirects.
2028                 return ImportStreamSource::newFromURL( $url, "POST" );
2029         }
2030 }