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                 global $wgMaxArticleSize;
 829
 830                 // Make sure revisions won't violate $wgMaxArticleSize, which could lead to
 831                 // database errors and instability. Testing for revisions with only listed
 832                 // content models, as other content models might use serialization formats
 833                 // which aren't checked against $wgMaxArticleSize.
 834                 if ( ( !isset( $revisionInfo['model'] ) ||
 835                         in_array( $revisionInfo['model'], array(
 836                                 'wikitext',
 837                                 'css',
 838                                 'json',
 839                                 'javascript',
 840                                 'text',
 841                                 ''
 842                         ) ) ) &&
 843                         (int)( strlen( $revisionInfo['text'] ) / 1024 ) > $wgMaxArticleSize
 844                 ) {
 845                         throw new MWException( 'The text of ' .
 846                                 ( isset( $revisionInfo['id'] ) ?
 847                                         "the revision with ID $revisionInfo[id]" :
 848                                         'a revision'
 849                                 ) . " exceeds the maximum allowable size ($wgMaxArticleSize KB)" );
 850                 }
 851
 852                 $revision = new WikiRevision( $this->config );
 853
 854                 if ( isset( $revisionInfo['id'] ) ) {
 855                         $revision->setID( $revisionInfo['id'] );
 856                 }
 857                 if ( isset( $revisionInfo['model'] ) ) {
 858                         $revision->setModel( $revisionInfo['model'] );
 859                 }
 860                 if ( isset( $revisionInfo['format'] ) ) {
 861                         $revision->setFormat( $revisionInfo['format'] );
 862                 }
 863                 $revision->setTitle( $pageInfo['_title'] );
 864
 865                 if ( isset( $revisionInfo['text'] ) ) {
 866                         $handler = $revision->getContentHandler();
 867                         $text = $handler->importTransform(
 868                                 $revisionInfo['text'],
 869                                 $revision->getFormat() );
 870
 871                         $revision->setText( $text );
 872                 }
 873                 if ( isset( $revisionInfo['timestamp'] ) ) {
 874                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 875                 } else {
 876                         $revision->setTimestamp( wfTimestampNow() );
 877                 }
 878
 879                 if ( isset( $revisionInfo['comment'] ) ) {
 880                         $revision->setComment( $revisionInfo['comment'] );
 881                 }
 882
 883                 if ( isset( $revisionInfo['minor'] ) ) {
 884                         $revision->setMinor( true );
 885                 }
 886                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 887                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 888                 } elseif ( isset( $revisionInfo['contributor']['username'] ) ) {
 889                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 890                 } else {
 891                         $revision->setUserName( 'Unknown user' );
 892                 }
 893                 $revision->setNoUpdates( $this->mNoUpdates );
 894
 895                 return $this->revisionCallback( $revision );
 896         }
 897
 898         /**
 899          * @param array $pageInfo
 900          * @return mixed
 901          */
 902         private function handleUpload( &$pageInfo ) {
 903                 $this->debug( "Enter upload handler" );
 904                 $uploadInfo = array();
 905
 906                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 907                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 908
 909                 $skip = false;
 910
 911                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 912                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 913                                         $this->reader->localName == 'upload' ) {
 914                                 break;
 915                         }
 916
 917                         $tag = $this->reader->localName;
 918
 919                         if ( !Hooks::run( 'ImportHandleUploadXMLTag', array(
 920                                 $this, $pageInfo
 921                         ) ) ) {
 922                                 // Do nothing
 923                         } elseif ( in_array( $tag, $normalFields ) ) {
 924                                 $uploadInfo[$tag] = $this->nodeContents();
 925                         } elseif ( $tag == 'contributor' ) {
 926                                 $uploadInfo['contributor'] = $this->handleContributor();
 927                         } elseif ( $tag == 'contents' ) {
 928                                 $contents = $this->nodeContents();
 929                                 $encoding = $this->reader->getAttribute( 'encoding' );
 930                                 if ( $encoding === 'base64' ) {
 931                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 932                                         $uploadInfo['isTempSrc'] = true;
 933                                 }
 934                         } elseif ( $tag != '#text' ) {
 935                                 $this->warn( "Unhandled upload XML tag $tag" );
 936                                 $skip = true;
 937                         }
 938                 }
 939
 940                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 941                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 942                         if ( file_exists( $path ) ) {
 943                                 $uploadInfo['fileSrc'] = $path;
 944                                 $uploadInfo['isTempSrc'] = false;
 945                         }
 946                 }
 947
 948                 if ( $this->mImportUploads ) {
 949                         return $this->processUpload( $pageInfo, $uploadInfo );
 950                 }
 951         }
 952
 953         /**
 954          * @param string $contents
 955          * @return string
 956          */
 957         private function dumpTemp( $contents ) {
 958                 $filename = tempnam( wfTempDir(), 'importupload' );
 959                 file_put_contents( $filename, $contents );
 960                 return $filename;
 961         }
 962
 963         /**
 964          * @param array $pageInfo
 965          * @param array $uploadInfo
 966          * @return mixed
 967          */
 968         private function processUpload( $pageInfo, $uploadInfo ) {
 969                 $revision = new WikiRevision( $this->config );
 970                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 971
 972                 $revision->setTitle( $pageInfo['_title'] );
 973                 $revision->setID( $pageInfo['id'] );
 974                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 975                 $revision->setText( $text );
 976                 $revision->setFilename( $uploadInfo['filename'] );
 977                 if ( isset( $uploadInfo['archivename'] ) ) {
 978                         $revision->setArchiveName( $uploadInfo['archivename'] );
 979                 }
 980                 $revision->setSrc( $uploadInfo['src'] );
 981                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 982                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 983                                 !empty( $uploadInfo['isTempSrc'] ) );
 984                 }
 985                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 986                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 987                 }
 988                 $revision->setSize( intval( $uploadInfo['size'] ) );
 989                 $revision->setComment( $uploadInfo['comment'] );
 990
 991                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 992                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 993                 }
 994                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 995                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 996                 }
 997                 $revision->setNoUpdates( $this->mNoUpdates );
 998
 999                 return call_user_func( $this->mUploadCallback, $revision );
1000         }
1001
1002         /**
1003          * @return array
1004          */
1005         private function handleContributor() {
1006                 $fields = array( 'id', 'ip', 'username' );
1007                 $info = array();
1008
1009                 if ( $this->reader->isEmptyElement ) {
1010                         return $info;
1011                 }
1012                 while ( $this->reader->read() ) {
1013                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
1014                                         $this->reader->localName == 'contributor' ) {
1015                                 break;
1016                         }
1017
1018                         $tag = $this->reader->localName;
1019
1020                         if ( in_array( $tag, $fields ) ) {
1021                                 $info[$tag] = $this->nodeContents();
1022                         }
1023                 }
1024
1025                 return $info;
1026         }
1027
1028         /**
1029          * @param string $text
1030          * @param string|null $ns
1031          * @return array|bool
1032          */
1033         private function processTitle( $text, $ns = null ) {
1034                 if ( is_null( $this->foreignNamespaces ) ) {
1035                         $foreignTitleFactory = new NaiveForeignTitleFactory();
1036                 } else {
1037                         $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
1038                                 $this->foreignNamespaces );
1039                 }
1040
1041                 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
1042                         intval( $ns ) );
1043
1044                 $title = $this->importTitleFactory->createTitleFromForeignTitle(
1045                         $foreignTitle );
1046
1047                 $commandLineMode = $this->config->get( 'CommandLineMode' );
1048                 if ( is_null( $title ) ) {
1049                         # Invalid page title? Ignore the page
1050                         $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
1051                         return false;
1052                 } elseif ( $title->isExternal() ) {
1053                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
1054                         return false;
1055                 } elseif ( !$title->canExist() ) {
1056                         $this->notice( 'import-error-special', $title->getPrefixedText() );
1057                         return false;
1058                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
1059                         # Do not import if the importing wiki user cannot edit this page
1060                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
1061                         return false;
1062                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
1063                         # Do not import if the importing wiki user cannot create this page
1064                         $this->notice( 'import-error-create', $title->getPrefixedText() );
1065                         return false;
1066                 }
1067
1068                 return array( $title, $foreignTitle );
1069         }
1070 }
1071
1072 /** This is a horrible hack used to keep source compatibility */
1073 class UploadSourceAdapter {
1074         /** @var array */
1075         public static $sourceRegistrations = array();
1076
1077         /** @var string */
1078         private $mSource;
1079
1080         /** @var string */
1081         private $mBuffer;
1082
1083         /** @var int */
1084         private $mPosition;
1085
1086         /**
1087          * @param ImportSource $source
1088          * @return string
1089          */
1090         static function registerSource( ImportSource $source ) {
1091                 $id = wfRandomString();
1092
1093                 self::$sourceRegistrations[$id] = $source;
1094
1095                 return $id;
1096         }
1097
1098         /**
1099          * @param string $path
1100          * @param string $mode
1101          * @param array $options
1102          * @param string $opened_path
1103          * @return bool
1104          */
1105         function stream_open( $path, $mode, $options, &$opened_path ) {
1106                 $url = parse_url( $path );
1107                 $id = $url['host'];
1108
1109                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
1110                         return false;
1111                 }
1112
1113                 $this->mSource = self::$sourceRegistrations[$id];
1114
1115                 return true;
1116         }
1117
1118         /**
1119          * @param int $count
1120          * @return string
1121          */
1122         function stream_read( $count ) {
1123                 $return = '';
1124                 $leave = false;
1125
1126                 while ( !$leave && !$this->mSource->atEnd() &&
1127                                 strlen( $this->mBuffer ) < $count ) {
1128                         $read = $this->mSource->readChunk();
1129
1130                         if ( !strlen( $read ) ) {
1131                                 $leave = true;
1132                         }
1133
1134                         $this->mBuffer .= $read;
1135                 }
1136
1137                 if ( strlen( $this->mBuffer ) ) {
1138                         $return = substr( $this->mBuffer, 0, $count );
1139                         $this->mBuffer = substr( $this->mBuffer, $count );
1140                 }
1141
1142                 $this->mPosition += strlen( $return );
1143
1144                 return $return;
1145         }
1146
1147         /**
1148          * @param string $data
1149          * @return bool
1150          */
1151         function stream_write( $data ) {
1152                 return false;
1153         }
1154
1155         /**
1156          * @return mixed
1157          */
1158         function stream_tell() {
1159                 return $this->mPosition;
1160         }
1161
1162         /**
1163          * @return bool
1164          */
1165         function stream_eof() {
1166                 return $this->mSource->atEnd();
1167         }
1168
1169         /**
1170          * @return array
1171          */
1172         function url_stat() {
1173                 $result = array();
1174
1175                 $result['dev'] = $result[0] = 0;
1176                 $result['ino'] = $result[1] = 0;
1177                 $result['mode'] = $result[2] = 0;
1178                 $result['nlink'] = $result[3] = 0;
1179                 $result['uid'] = $result[4] = 0;
1180                 $result['gid'] = $result[5] = 0;
1181                 $result['rdev'] = $result[6] = 0;
1182                 $result['size'] = $result[7] = 0;
1183                 $result['atime'] = $result[8] = 0;
1184                 $result['mtime'] = $result[9] = 0;
1185                 $result['ctime'] = $result[10] = 0;
1186                 $result['blksize'] = $result[11] = 0;
1187                 $result['blocks'] = $result[12] = 0;
1188
1189                 return $result;
1190         }
1191 }
1192
1193 /**
1194  * @todo document (e.g. one-sentence class description).
1195  * @ingroup SpecialPage
1196  */
1197 class WikiRevision {
1198         /** @todo Unused? */
1199         public $importer = null;
1200
1201         /** @var Title */
1202         public $title = null;
1203
1204         /** @var int */
1205         public $id = 0;
1206
1207         /** @var string */
1208         public $timestamp = "20010115000000";
1209
1210         /**
1211          * @var int
1212          * @todo Can't find any uses. Public, because that's suspicious. Get clarity. */
1213         public $user = 0;
1214
1215         /** @var string */
1216         public $user_text = "";
1217
1218         /** @var string */
1219         public $model = null;
1220
1221         /** @var string */
1222         public $format = null;
1223
1224         /** @var string */
1225         public $text = "";
1226
1227         /** @var int */
1228         protected $size;
1229
1230         /** @var Content */
1231         public $content = null;
1232
1233         /** @var ContentHandler */
1234         protected $contentHandler = null;
1235
1236         /** @var string */
1237         public $comment = "";
1238
1239         /** @var bool */
1240         public $minor = false;
1241
1242         /** @var string */
1243         public $type = "";
1244
1245         /** @var string */
1246         public $action = "";
1247
1248         /** @var string */
1249         public $params = "";
1250
1251         /** @var string */
1252         public $fileSrc = '';
1253
1254         /** @var bool|string */
1255         public $sha1base36 = false;
1256
1257         /**
1258          * @var bool
1259          * @todo Unused?
1260          */
1261         public $isTemp = false;
1262
1263         /** @var string */
1264         public $archiveName = '';
1265
1266         protected $filename;
1267
1268         /** @var mixed */
1269         protected $src;
1270
1271         /** @todo Unused? */
1272         public $fileIsTemp;
1273
1274         /** @var bool */
1275         private $mNoUpdates = false;
1276
1277         /** @var Config $config */
1278         private $config;
1279
1280         public function __construct( Config $config ) {
1281                 $this->config = $config;
1282         }
1283
1284         /**
1285          * @param Title $title
1286          * @throws MWException
1287          */
1288         function setTitle( $title ) {
1289                 if ( is_object( $title ) ) {
1290                         $this->title = $title;
1291                 } elseif ( is_null( $title ) ) {
1292                         throw new MWException( "WikiRevision given a null title in import. "
1293                                 . "You may need to adjust \$wgLegalTitleChars." );
1294                 } else {
1295                         throw new MWException( "WikiRevision given non-object title in import." );
1296                 }
1297         }
1298
1299         /**
1300          * @param int $id
1301          */
1302         function setID( $id ) {
1303                 $this->id = $id;
1304         }
1305
1306         /**
1307          * @param string $ts
1308          */
1309         function setTimestamp( $ts ) {
1310                 # 2003-08-05T18:30:02Z
1311                 $this->timestamp = wfTimestamp( TS_MW, $ts );
1312         }
1313
1314         /**
1315          * @param string $user
1316          */
1317         function setUsername( $user ) {
1318                 $this->user_text = $user;
1319         }
1320
1321         /**
1322          * @param string $ip
1323          */
1324         function setUserIP( $ip ) {
1325                 $this->user_text = $ip;
1326         }
1327
1328         /**
1329          * @param string $model
1330          */
1331         function setModel( $model ) {
1332                 $this->model = $model;
1333         }
1334
1335         /**
1336          * @param string $format
1337          */
1338         function setFormat( $format ) {
1339                 $this->format = $format;
1340         }
1341
1342         /**
1343          * @param string $text
1344          */
1345         function setText( $text ) {
1346                 $this->text = $text;
1347         }
1348
1349         /**
1350          * @param string $text
1351          */
1352         function setComment( $text ) {
1353                 $this->comment = $text;
1354         }
1355
1356         /**
1357          * @param bool $minor
1358          */
1359         function setMinor( $minor ) {
1360                 $this->minor = (bool)$minor;
1361         }
1362
1363         /**
1364          * @param mixed $src
1365          */
1366         function setSrc( $src ) {
1367                 $this->src = $src;
1368         }
1369
1370         /**
1371          * @param string $src
1372          * @param bool $isTemp
1373          */
1374         function setFileSrc( $src, $isTemp ) {
1375                 $this->fileSrc = $src;
1376                 $this->fileIsTemp = $isTemp;
1377         }
1378
1379         /**
1380          * @param string $sha1base36
1381          */
1382         function setSha1Base36( $sha1base36 ) {
1383                 $this->sha1base36 = $sha1base36;
1384         }
1385
1386         /**
1387          * @param string $filename
1388          */
1389         function setFilename( $filename ) {
1390                 $this->filename = $filename;
1391         }
1392
1393         /**
1394          * @param string $archiveName
1395          */
1396         function setArchiveName( $archiveName ) {
1397                 $this->archiveName = $archiveName;
1398         }
1399
1400         /**
1401          * @param int $size
1402          */
1403         function setSize( $size ) {
1404                 $this->size = intval( $size );
1405         }
1406
1407         /**
1408          * @param string $type
1409          */
1410         function setType( $type ) {
1411                 $this->type = $type;
1412         }
1413
1414         /**
1415          * @param string $action
1416          */
1417         function setAction( $action ) {
1418                 $this->action = $action;
1419         }
1420
1421         /**
1422          * @param array $params
1423          */
1424         function setParams( $params ) {
1425                 $this->params = $params;
1426         }
1427
1428         /**
1429          * @param bool $noupdates
1430          */
1431         public function setNoUpdates( $noupdates ) {
1432                 $this->mNoUpdates = $noupdates;
1433         }
1434
1435         /**
1436          * @return Title
1437          */
1438         function getTitle() {
1439                 return $this->title;
1440         }
1441
1442         /**
1443          * @return int
1444          */
1445         function getID() {
1446                 return $this->id;
1447         }
1448
1449         /**
1450          * @return string
1451          */
1452         function getTimestamp() {
1453                 return $this->timestamp;
1454         }
1455
1456         /**
1457          * @return string
1458          */
1459         function getUser() {
1460                 return $this->user_text;
1461         }
1462
1463         /**
1464          * @return string
1465          *
1466          * @deprecated Since 1.21, use getContent() instead.
1467          */
1468         function getText() {
1469                 ContentHandler::deprecated( __METHOD__, '1.21' );
1470
1471                 return $this->text;
1472         }
1473
1474         /**
1475          * @return ContentHandler
1476          */
1477         function getContentHandler() {
1478                 if ( is_null( $this->contentHandler ) ) {
1479                         $this->contentHandler = ContentHandler::getForModelID( $this->getModel() );
1480                 }
1481
1482                 return $this->contentHandler;
1483         }
1484
1485         /**
1486          * @return Content
1487          */
1488         function getContent() {
1489                 if ( is_null( $this->content ) ) {
1490                         $handler = $this->getContentHandler();
1491                         $this->content = $handler->unserializeContent( $this->text, $this->getFormat() );
1492                 }
1493
1494                 return $this->content;
1495         }
1496
1497         /**
1498          * @return string
1499          */
1500         function getModel() {
1501                 if ( is_null( $this->model ) ) {
1502                         $this->model = $this->getTitle()->getContentModel();
1503                 }
1504
1505                 return $this->model;
1506         }
1507
1508         /**
1509          * @return string
1510          */
1511         function getFormat() {
1512                 if ( is_null( $this->format ) ) {
1513                         $this->format = $this->getContentHandler()->getDefaultFormat();
1514                 }
1515
1516                 return $this->format;
1517         }
1518
1519         /**
1520          * @return string
1521          */
1522         function getComment() {
1523                 return $this->comment;
1524         }
1525
1526         /**
1527          * @return bool
1528          */
1529         function getMinor() {
1530                 return $this->minor;
1531         }
1532
1533         /**
1534          * @return mixed
1535          */
1536         function getSrc() {
1537                 return $this->src;
1538         }
1539
1540         /**
1541          * @return bool|string
1542          */
1543         function getSha1() {
1544                 if ( $this->sha1base36 ) {
1545                         return Wikimedia\base_convert( $this->sha1base36, 36, 16 );
1546                 }
1547                 return false;
1548         }
1549
1550         /**
1551          * @return string
1552          */
1553         function getFileSrc() {
1554                 return $this->fileSrc;
1555         }
1556
1557         /**
1558          * @return bool
1559          */
1560         function isTempSrc() {
1561                 return $this->isTemp;
1562         }
1563
1564         /**
1565          * @return mixed
1566          */
1567         function getFilename() {
1568                 return $this->filename;
1569         }
1570
1571         /**
1572          * @return string
1573          */
1574         function getArchiveName() {
1575                 return $this->archiveName;
1576         }
1577
1578         /**
1579          * @return mixed
1580          */
1581         function getSize() {
1582                 return $this->size;
1583         }
1584
1585         /**
1586          * @return string
1587          */
1588         function getType() {
1589                 return $this->type;
1590         }
1591
1592         /**
1593          * @return string
1594          */
1595         function getAction() {
1596                 return $this->action;
1597         }
1598
1599         /**
1600          * @return string
1601          */
1602         function getParams() {
1603                 return $this->params;
1604         }
1605
1606         /**
1607          * @return bool
1608          */
1609         function importOldRevision() {
1610                 $dbw = wfGetDB( DB_MASTER );
1611
1612                 # Sneak a single revision into place
1613                 $user = User::newFromName( $this->getUser() );
1614                 if ( $user ) {
1615                         $userId = intval( $user->getId() );
1616                         $userText = $user->getName();
1617                         $userObj = $user;
1618                 } else {
1619                         $userId = 0;
1620                         $userText = $this->getUser();
1621                         $userObj = new User;
1622                 }
1623
1624                 // avoid memory leak...?
1625                 Title::clearCaches();
1626
1627                 $page = WikiPage::factory( $this->title );
1628                 $page->loadPageData( 'fromdbmaster' );
1629                 if ( !$page->exists() ) {
1630                         # must create the page...
1631                         $pageId = $page->insertOn( $dbw );
1632                         $created = true;
1633                         $oldcountable = null;
1634                 } else {
1635                         $pageId = $page->getId();
1636                         $created = false;
1637
1638                         $prior = $dbw->selectField( 'revision', '1',
1639                                 array( 'rev_page' => $pageId,
1640                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1641                                         'rev_user_text' => $userText,
1642                                         'rev_comment' => $this->getComment() ),
1643                                 __METHOD__
1644                         );
1645                         if ( $prior ) {
1646                                 // @todo FIXME: This could fail slightly for multiple matches :P
1647                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1648                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1649                                 return false;
1650                         }
1651                 }
1652
1653                 // Select previous version to make size diffs correct
1654                 $prevId = $dbw->selectField( 'revision', 'rev_id',
1655                         array(
1656                                 'rev_page' => $pageId,
1657                                 'rev_timestamp <= ' . $dbw->timestamp( $this->timestamp ),
1658                         ),
1659                         __METHOD__,
1660                         array( 'ORDER BY' => array(
1661                                         'rev_timestamp DESC',
1662                                         'rev_id DESC', // timestamp is not unique per page
1663                                 )
1664                         )
1665                 );
1666
1667                 # @todo FIXME: Use original rev_id optionally (better for backups)
1668                 # Insert the row
1669                 $revision = new Revision( array(
1670                         'title' => $this->title,
1671                         'page' => $pageId,
1672                         'content_model' => $this->getModel(),
1673                         'content_format' => $this->getFormat(),
1674                         // XXX: just set 'content' => $this->getContent()?
1675                         'text' => $this->getContent()->serialize( $this->getFormat() ),
1676                         'comment' => $this->getComment(),
1677                         'user' => $userId,
1678                         'user_text' => $userText,
1679                         'timestamp' => $this->timestamp,
1680                         'minor_edit' => $this->minor,
1681                         'parent_id' => $prevId,
1682                         ) );
1683                 $revision->insertOn( $dbw );
1684                 $changed = $page->updateIfNewerOn( $dbw, $revision );
1685
1686                 if ( $changed !== false && !$this->mNoUpdates ) {
1687                         wfDebug( __METHOD__ . ": running updates\n" );
1688                         // countable/oldcountable stuff is handled in WikiImporter::finishImportPage
1689                         $page->doEditUpdates(
1690                                 $revision,
1691                                 $userObj,
1692                                 array( 'created' => $created, 'oldcountable' => 'no-change' )
1693                         );
1694                 }
1695
1696                 return true;
1697         }
1698
1699         function importLogItem() {
1700                 $dbw = wfGetDB( DB_MASTER );
1701
1702                 $user = User::newFromName( $this->getUser() );
1703                 if ( $user ) {
1704                         $userId = intval( $user->getId() );
1705                         $userText = $user->getName();
1706                 } else {
1707                         $userId = 0;
1708                         $userText = $this->getUser();
1709                 }
1710
1711                 # @todo FIXME: This will not record autoblocks
1712                 if ( !$this->getTitle() ) {
1713                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1714                                 $this->timestamp . "\n" );
1715                         return;
1716                 }
1717                 # Check if it exists already
1718                 // @todo FIXME: Use original log ID (better for backups)
1719                 $prior = $dbw->selectField( 'logging', '1',
1720                         array( 'log_type' => $this->getType(),
1721                                 'log_action' => $this->getAction(),
1722                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1723                                 'log_namespace' => $this->getTitle()->getNamespace(),
1724                                 'log_title' => $this->getTitle()->getDBkey(),
1725                                 'log_comment' => $this->getComment(),
1726                                 # 'log_user_text' => $this->user_text,
1727                                 'log_params' => $this->params ),
1728                         __METHOD__
1729                 );
1730                 // @todo FIXME: This could fail slightly for multiple matches :P
1731                 if ( $prior ) {
1732                         wfDebug( __METHOD__
1733                                 . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp "
1734                                 . $this->timestamp . "\n" );
1735                         return;
1736                 }
1737                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1738                 $data = array(
1739                         'log_id' => $log_id,
1740                         'log_type' => $this->type,
1741                         'log_action' => $this->action,
1742                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1743                         'log_user' =>  $userId,
1744                         'log_user_text' => $userText,
1745                         'log_namespace' => $this->getTitle()->getNamespace(),
1746                         'log_title' => $this->getTitle()->getDBkey(),
1747                         'log_comment' => $this->getComment(),
1748                         'log_params' => $this->params
1749                 );
1750                 $dbw->insert( 'logging', $data, __METHOD__ );
1751         }
1752
1753         /**
1754          * @return bool
1755          */
1756         function importUpload() {
1757                 # Construct a file
1758                 $archiveName = $this->getArchiveName();
1759                 if ( $archiveName ) {
1760                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1761                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1762                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1763                 } else {
1764                         $file = wfLocalFile( $this->getTitle() );
1765                         $file->load( File::READ_LATEST );
1766                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1767                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1768                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1769                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1770                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1771                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1772                         }
1773                 }
1774                 if ( !$file ) {
1775                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1776                         return false;
1777                 }
1778
1779                 # Get the file source or download if necessary
1780                 $source = $this->getFileSrc();
1781                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1782                 if ( !$source ) {
1783                         $source = $this->downloadSource();
1784                         $flags |= File::DELETE_SOURCE;
1785                 }
1786                 if ( !$source ) {
1787                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1788                         return false;
1789                 }
1790                 $sha1 = $this->getSha1();
1791                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1792                         if ( $flags & File::DELETE_SOURCE ) {
1793                                 # Broken file; delete it if it is a temporary file
1794                                 unlink( $source );
1795                         }
1796                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1797                         return false;
1798                 }
1799
1800                 $user = User::newFromName( $this->user_text );
1801
1802                 # Do the actual upload
1803                 if ( $archiveName ) {
1804                         $status = $file->uploadOld( $source, $archiveName,
1805                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1806                 } else {
1807                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1808                                 $flags, false, $this->getTimestamp(), $user );
1809                 }
1810
1811                 if ( $status->isGood() ) {
1812                         wfDebug( __METHOD__ . ": Successful\n" );
1813                         return true;
1814                 } else {
1815                         wfDebug( __METHOD__ . ': failed: ' . $status->getHTML() . "\n" );
1816                         return false;
1817                 }
1818         }
1819
1820         /**
1821          * @return bool|string
1822          */
1823         function downloadSource() {
1824                 if ( !$this->config->get( 'EnableUploads' ) ) {
1825                         return false;
1826                 }
1827
1828                 $tempo = tempnam( wfTempDir(), 'download' );
1829                 $f = fopen( $tempo, 'wb' );
1830                 if ( !$f ) {
1831                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1832                         return false;
1833                 }
1834
1835                 // @todo FIXME!
1836                 $src = $this->getSrc();
1837                 $data = Http::get( $src, array(), __METHOD__ );
1838                 if ( !$data ) {
1839                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1840                         fclose( $f );
1841                         unlink( $tempo );
1842                         return false;
1843                 }
1844
1845                 fwrite( $f, $data );
1846                 fclose( $f );
1847
1848                 return $tempo;
1849         }
1850
1851 }
1852
1853 /**
1854  * Source interface for XML import.
1855  */
1856 interface ImportSource {
1857
1858         /**
1859          * Indicates whether the end of the input has been reached.
1860          * Will return true after a finite number of calls to readChunk.
1861          *
1862          * @return bool true if there is no more input, false otherwise.
1863          */
1864         function atEnd();
1865
1866         /**
1867          * Return a chunk of the input, as a (possibly empty) string.
1868          * When the end of input is reached, readChunk() returns false.
1869          * If atEnd() returns false, readChunk() will return a string.
1870          * If atEnd() returns true, readChunk() will return false.
1871          *
1872          * @return bool|string
1873          */
1874         function readChunk();
1875 }
1876
1877 /**
1878  * Used for importing XML dumps where the content of the dump is in a string.
1879  * This class is ineffecient, and should only be used for small dumps.
1880  * For larger dumps, ImportStreamSource should be used instead.
1881  *
1882  * @ingroup SpecialPage
1883  */
1884 class ImportStringSource implements ImportSource {
1885         function __construct( $string ) {
1886                 $this->mString = $string;
1887                 $this->mRead = false;
1888         }
1889
1890         /**
1891          * @return bool
1892          */
1893         function atEnd() {
1894                 return $this->mRead;
1895         }
1896
1897         /**
1898          * @return bool|string
1899          */
1900         function readChunk() {
1901                 if ( $this->atEnd() ) {
1902                         return false;
1903                 }
1904                 $this->mRead = true;
1905                 return $this->mString;
1906         }
1907 }
1908
1909 /**
1910  * Imports a XML dump from a file (either from file upload, files on disk, or HTTP)
1911  * @ingroup SpecialPage
1912  */
1913 class ImportStreamSource implements ImportSource {
1914         function __construct( $handle ) {
1915                 $this->mHandle = $handle;
1916         }
1917
1918         /**
1919          * @return bool
1920          */
1921         function atEnd() {
1922                 return feof( $this->mHandle );
1923         }
1924
1925         /**
1926          * @return string
1927          */
1928         function readChunk() {
1929                 return fread( $this->mHandle, 32768 );
1930         }
1931
1932         /**
1933          * @param string $filename
1934          * @return Status
1935          */
1936         static function newFromFile( $filename ) {
1937                 MediaWiki\suppressWarnings();
1938                 $file = fopen( $filename, 'rt' );
1939                 MediaWiki\restoreWarnings();
1940                 if ( !$file ) {
1941                         return Status::newFatal( "importcantopen" );
1942                 }
1943                 return Status::newGood( new ImportStreamSource( $file ) );
1944         }
1945
1946         /**
1947          * @param string $fieldname
1948          * @return Status
1949          */
1950         static function newFromUpload( $fieldname = "xmlimport" ) {
1951                 $upload =& $_FILES[$fieldname];
1952
1953                 if ( $upload === null || !$upload['name'] ) {
1954                         return Status::newFatal( 'importnofile' );
1955                 }
1956                 if ( !empty( $upload['error'] ) ) {
1957                         switch ( $upload['error'] ) {
1958                                 case 1:
1959                                         # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1960                                         return Status::newFatal( 'importuploaderrorsize' );
1961                                 case 2:
1962                                         # The uploaded file exceeds the MAX_FILE_SIZE directive that
1963                                         # was specified in the HTML form.
1964                                         return Status::newFatal( 'importuploaderrorsize' );
1965                                 case 3:
1966                                         # The uploaded file was only partially uploaded
1967                                         return Status::newFatal( 'importuploaderrorpartial' );
1968                                 case 6:
1969                                         # Missing a temporary folder.
1970                                         return Status::newFatal( 'importuploaderrortemp' );
1971                                 # case else: # Currently impossible
1972                         }
1973
1974                 }
1975                 $fname = $upload['tmp_name'];
1976                 if ( is_uploaded_file( $fname ) ) {
1977                         return ImportStreamSource::newFromFile( $fname );
1978                 } else {
1979                         return Status::newFatal( 'importnofile' );
1980                 }
1981         }
1982
1983         /**
1984          * @param string $url
1985          * @param string $method
1986          * @return Status
1987          */
1988         static function newFromURL( $url, $method = 'GET' ) {
1989                 wfDebug( __METHOD__ . ": opening $url\n" );
1990                 # Use the standard HTTP fetch function; it times out
1991                 # quicker and sorts out user-agent problems which might
1992                 # otherwise prevent importing from large sites, such
1993                 # as the Wikimedia cluster, etc.
1994                 $data = Http::request( $method, $url, array( 'followRedirects' => true ), __METHOD__ );
1995                 if ( $data !== false ) {
1996                         $file = tmpfile();
1997                         fwrite( $file, $data );
1998                         fflush( $file );
1999                         fseek( $file, 0 );
2000                         return Status::newGood( new ImportStreamSource( $file ) );
2001                 } else {
2002                         return Status::newFatal( 'importcantopen' );
2003                 }
2004         }
2005
2006         /**
2007          * @param string $interwiki
2008          * @param string $page
2009          * @param bool $history
2010          * @param bool $templates
2011          * @param int $pageLinkDepth
2012          * @return Status
2013          */
2014         public static function newFromInterwiki( $interwiki, $page, $history = false,
2015                 $templates = false, $pageLinkDepth = 0
2016         ) {
2017                 if ( $page == '' ) {
2018                         return Status::newFatal( 'import-noarticle' );
2019                 }
2020
2021                 # Look up the first interwiki prefix, and let the foreign site handle
2022                 # subsequent interwiki prefixes
2023                 $firstIwPrefix = strtok( $interwiki, ':' );
2024                 $firstIw = Interwiki::fetch( $firstIwPrefix );
2025                 if ( !$firstIw ) {
2026                         return Status::newFatal( 'importbadinterwiki' );
2027                 }
2028
2029                 $additionalIwPrefixes = strtok( '' );
2030                 if ( $additionalIwPrefixes ) {
2031                         $additionalIwPrefixes .= ':';
2032                 }
2033                 # Have to do a DB-key replacement ourselves; otherwise spaces get
2034                 # URL-encoded to +, which is wrong in this case. Similar to logic in
2035                 # Title::getLocalURL
2036                 $link = $firstIw->getURL( strtr( "${additionalIwPrefixes}Special:Export/$page",
2037                         ' ', '_' ) );
2038
2039                 $params = array();
2040                 if ( $history ) {
2041                         $params['history'] = 1;
2042                 }
2043                 if ( $templates ) {
2044                         $params['templates'] = 1;
2045                 }
2046                 if ( $pageLinkDepth ) {
2047                         $params['pagelink-depth'] = $pageLinkDepth;
2048                 }
2049
2050                 $url = wfAppendQuery( $link, $params );
2051                 # For interwikis, use POST to avoid redirects.
2052                 return ImportStreamSource::newFromURL( $url, "POST" );
2053         }
2054 }