includes/Import.php

   1 <?php
   2 /**
   3  * MediaWiki page data importer.
   4  *
   5  * Copyright © 2003,2005 Brion Vibber <brion@pobox.com>
   6  * https://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup SpecialPage
  25  */
  26
  27 /**
  28  * XML file reader for the page data importer
  29  *
  30  * implements Special:Import
  31  * @ingroup SpecialPage
  32  */
  33 class WikiImporter {
  34         private $reader = null;
  35         private $foreignNamespaces = null;
  36         private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
  37         private $mSiteInfoCallback, $mTargetNamespace, $mPageOutCallback;
  38         private $mNoticeCallback, $mDebug;
  39         private $mImportUploads, $mImageBasePath;
  40         private $mNoUpdates = false;
  41         /** @var Config */
  42         private $config;
  43         /** @var ImportTitleFactory */
  44         private $importTitleFactory;
  45
  46         /**
  47          * Creates an ImportXMLReader drawing from the source provided
  48          * @param ImportSource $source
  49          * @param Config $config
  50          */
  51         function __construct( ImportSource $source, Config $config = null ) {
  52                 $this->reader = new XMLReader();
  53                 if ( !$config ) {
  54                         wfDeprecated( __METHOD__ . ' without a Config instance', '1.25' );
  55                         $config = ConfigFactory::getDefaultInstance()->makeConfig( 'main' );
  56                 }
  57                 $this->config = $config;
  58
  59                 if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) {
  60                         stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  61                 }
  62                 $id = UploadSourceAdapter::registerSource( $source );
  63                 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
  64                         $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  65                 } else {
  66                         $this->reader->open( "uploadsource://$id" );
  67                 }
  68
  69                 // Default callbacks
  70                 $this->setRevisionCallback( array( $this, "importRevision" ) );
  71                 $this->setUploadCallback( array( $this, 'importUpload' ) );
  72                 $this->setLogItemCallback( array( $this, 'importLogItem' ) );
  73                 $this->setPageOutCallback( array( $this, 'finishImportPage' ) );
  74
  75                 $this->importTitleFactory = new NaiveImportTitleFactory();
  76         }
  77
  78         /**
  79          * @return null|XMLReader
  80          */
  81         public function getReader() {
  82                 return $this->reader;
  83         }
  84
  85         public function throwXmlError( $err ) {
  86                 $this->debug( "FAILURE: $err" );
  87                 wfDebug( "WikiImporter XML error: $err\n" );
  88         }
  89
  90         public function debug( $data ) {
  91                 if ( $this->mDebug ) {
  92                         wfDebug( "IMPORT: $data\n" );
  93                 }
  94         }
  95
  96         public function warn( $data ) {
  97                 wfDebug( "IMPORT: $data\n" );
  98         }
  99
 100         public function notice( $msg /*, $param, ...*/ ) {
 101                 $params = func_get_args();
 102                 array_shift( $params );
 103
 104                 if ( is_callable( $this->mNoticeCallback ) ) {
 105                         call_user_func( $this->mNoticeCallback, $msg, $params );
 106                 } else { # No ImportReporter -> CLI
 107                         echo wfMessage( $msg, $params )->text() . "\n";
 108                 }
 109         }
 110
 111         /**
 112          * Set debug mode...
 113          * @param bool $debug
 114          */
 115         function setDebug( $debug ) {
 116                 $this->mDebug = $debug;
 117         }
 118
 119         /**
 120          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
 121          * @param bool $noupdates
 122          */
 123         function setNoUpdates( $noupdates ) {
 124                 $this->mNoUpdates = $noupdates;
 125         }
 126
 127         /**
 128          * Set a callback that displays notice messages
 129          *
 130          * @param callable $callback
 131          * @return callable
 132          */
 133         public function setNoticeCallback( $callback ) {
 134                 return wfSetVar( $this->mNoticeCallback, $callback );
 135         }
 136
 137         /**
 138          * Sets the action to perform as each new page in the stream is reached.
 139          * @param callable $callback
 140          * @return callable
 141          */
 142         public function setPageCallback( $callback ) {
 143                 $previous = $this->mPageCallback;
 144                 $this->mPageCallback = $callback;
 145                 return $previous;
 146         }
 147
 148         /**
 149          * Sets the action to perform as each page in the stream is completed.
 150          * Callback accepts the page title (as a Title object), a second object
 151          * with the original title form (in case it's been overridden into a
 152          * local namespace), and a count of revisions.
 153          *
 154          * @param callable $callback
 155          * @return callable
 156          */
 157         public function setPageOutCallback( $callback ) {
 158                 $previous = $this->mPageOutCallback;
 159                 $this->mPageOutCallback = $callback;
 160                 return $previous;
 161         }
 162
 163         /**
 164          * Sets the action to perform as each page revision is reached.
 165          * @param callable $callback
 166          * @return callable
 167          */
 168         public function setRevisionCallback( $callback ) {
 169                 $previous = $this->mRevisionCallback;
 170                 $this->mRevisionCallback = $callback;
 171                 return $previous;
 172         }
 173
 174         /**
 175          * Sets the action to perform as each file upload version is reached.
 176          * @param callable $callback
 177          * @return callable
 178          */
 179         public function setUploadCallback( $callback ) {
 180                 $previous = $this->mUploadCallback;
 181                 $this->mUploadCallback = $callback;
 182                 return $previous;
 183         }
 184
 185         /**
 186          * Sets the action to perform as each log item reached.
 187          * @param callable $callback
 188          * @return callable
 189          */
 190         public function setLogItemCallback( $callback ) {
 191                 $previous = $this->mLogItemCallback;
 192                 $this->mLogItemCallback = $callback;
 193                 return $previous;
 194         }
 195
 196         /**
 197          * Sets the action to perform when site info is encountered
 198          * @param callable $callback
 199          * @return callable
 200          */
 201         public function setSiteInfoCallback( $callback ) {
 202                 $previous = $this->mSiteInfoCallback;
 203                 $this->mSiteInfoCallback = $callback;
 204                 return $previous;
 205         }
 206
 207         /**
 208          * Sets the factory object to use to convert ForeignTitle objects into local
 209          * Title objects
 210          * @param ImportTitleFactory $factory
 211          */
 212         public function setImportTitleFactory( $factory ) {
 213                 $this->importTitleFactory = $factory;
 214         }
 215
 216         /**
 217          * Set a target namespace to override the defaults
 218          * @param null|int $namespace
 219          * @return bool
 220          */
 221         public function setTargetNamespace( $namespace ) {
 222                 if ( is_null( $namespace ) ) {
 223                         // Don't override namespaces
 224                         $this->mTargetNamespace = null;
 225                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 226                         return true;
 227                 } elseif (
 228                         $namespace >= 0 &&
 229                         MWNamespace::exists( intval( $namespace ) )
 230                 ) {
 231                         $namespace = intval( $namespace );
 232                         $this->mTargetNamespace = $namespace;
 233                         $this->setImportTitleFactory( new NamespaceImportTitleFactory( $namespace ) );
 234                         return true;
 235                 } else {
 236                         return false;
 237                 }
 238         }
 239
 240         /**
 241          * Set a target root page under which all pages are imported
 242          * @param null|string $rootpage
 243          * @return Status
 244          */
 245         public function setTargetRootPage( $rootpage ) {
 246                 $status = Status::newGood();
 247                 if ( is_null( $rootpage ) ) {
 248                         // No rootpage
 249                         $this->setImportTitleFactory( new NaiveImportTitleFactory() );
 250                 } elseif ( $rootpage !== '' ) {
 251                         $rootpage = rtrim( $rootpage, '/' ); //avoid double slashes
 252                         $title = Title::newFromText( $rootpage, !is_null( $this->mTargetNamespace )
 253                                 ? $this->mTargetNamespace
 254                                 : NS_MAIN
 255                         );
 256
 257                         if ( !$title || $title->isExternal() ) {
 258                                 $status->fatal( 'import-rootpage-invalid' );
 259                         } else {
 260                                 if ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 261                                         global $wgContLang;
 262
 263                                         $displayNSText = $title->getNamespace() == NS_MAIN
 264                                                 ? wfMessage( 'blanknamespace' )->text()
 265                                                 : $wgContLang->getNsText( $title->getNamespace() );
 266                                         $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 267                                 } else {
 268                                         // set namespace to 'all', so the namespace check in processTitle() can pass
 269                                         $this->setTargetNamespace( null );
 270                                         $this->setImportTitleFactory( new SubpageImportTitleFactory( $title ) );
 271                                 }
 272                         }
 273                 }
 274                 return $status;
 275         }
 276
 277         /**
 278          * @param string $dir
 279          */
 280         public function setImageBasePath( $dir ) {
 281                 $this->mImageBasePath = $dir;
 282         }
 283
 284         /**
 285          * @param bool $import
 286          */
 287         public function setImportUploads( $import ) {
 288                 $this->mImportUploads = $import;
 289         }
 290
 291         /**
 292          * Default per-revision callback, performs the import.
 293          * @param WikiRevision $revision
 294          * @return bool
 295          */
 296         public function importRevision( $revision ) {
 297                 if ( !$revision->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
 298                         $this->notice( 'import-error-bad-location',
 299                                 $revision->getTitle()->getPrefixedText(),
 300                                 $revision->getID(),
 301                                 $revision->getModel(),
 302                                 $revision->getFormat() );
 303
 304                         return false;
 305                 }
 306
 307                 try {
 308                         $dbw = wfGetDB( DB_MASTER );
 309                         return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
 310                 } catch ( MWContentSerializationException $ex ) {
 311                         $this->notice( 'import-error-unserialize',
 312                                 $revision->getTitle()->getPrefixedText(),
 313                                 $revision->getID(),
 314                                 $revision->getModel(),
 315                                 $revision->getFormat() );
 316                 }
 317
 318                 return false;
 319         }
 320
 321         /**
 322          * Default per-revision callback, performs the import.
 323          * @param WikiRevision $revision
 324          * @return bool
 325          */
 326         public function importLogItem( $revision ) {
 327                 $dbw = wfGetDB( DB_MASTER );
 328                 return $dbw->deadlockLoop( array( $revision, 'importLogItem' ) );
 329         }
 330
 331         /**
 332          * Dummy for now...
 333          * @param WikiRevision $revision
 334          * @return bool
 335          */
 336         public function importUpload( $revision ) {
 337                 $dbw = wfGetDB( DB_MASTER );
 338                 return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
 339         }
 340
 341         /**
 342          * Mostly for hook use
 343          * @param Title $title
 344          * @param ForeignTitle $foreignTitle
 345          * @param int $revCount
 346          * @param int $sRevCount
 347          * @param array $pageInfo
 348          * @return bool
 349          */
 350         public function finishImportPage( $title, $foreignTitle, $revCount,
 351                         $sRevCount, $pageInfo ) {
 352                 $args = func_get_args();
 353                 return Hooks::run( 'AfterImportPage', $args );
 354         }
 355
 356         /**
 357          * Alternate per-revision callback, for debugging.
 358          * @param WikiRevision $revision
 359          */
 360         public function debugRevisionHandler( &$revision ) {
 361                 $this->debug( "Got revision:" );
 362                 if ( is_object( $revision->title ) ) {
 363                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 364                 } else {
 365                         $this->debug( "-- Title: <invalid>" );
 366                 }
 367                 $this->debug( "-- User: " . $revision->user_text );
 368                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 369                 $this->debug( "-- Comment: " . $revision->comment );
 370                 $this->debug( "-- Text: " . $revision->text );
 371         }
 372
 373         /**
 374          * Notify the callback function of site info
 375          * @param array $siteInfo
 376          * @return bool|mixed
 377          */
 378         private function siteInfoCallback( $siteInfo ) {
 379                 if ( isset( $this->mSiteInfoCallback ) ) {
 380                         return call_user_func_array( $this->mSiteInfoCallback,
 381                                         array( $siteInfo, $this ) );
 382                 } else {
 383                         return false;
 384                 }
 385         }
 386
 387         /**
 388          * Notify the callback function when a new "<page>" is reached.
 389          * @param Title $title
 390          */
 391         function pageCallback( $title ) {
 392                 if ( isset( $this->mPageCallback ) ) {
 393                         call_user_func( $this->mPageCallback, $title );
 394                 }
 395         }
 396
 397         /**
 398          * Notify the callback function when a "</page>" is closed.
 399          * @param Title $title
 400          * @param ForeignTitle $foreignTitle
 401          * @param int $revCount
 402          * @param int $sucCount Number of revisions for which callback returned true
 403          * @param array $pageInfo Associative array of page information
 404          */
 405         private function pageOutCallback( $title, $foreignTitle, $revCount,
 406                         $sucCount, $pageInfo ) {
 407                 if ( isset( $this->mPageOutCallback ) ) {
 408                         $args = func_get_args();
 409                         call_user_func_array( $this->mPageOutCallback, $args );
 410                 }
 411         }
 412
 413         /**
 414          * Notify the callback function of a revision
 415          * @param WikiRevision $revision
 416          * @return bool|mixed
 417          */
 418         private function revisionCallback( $revision ) {
 419                 if ( isset( $this->mRevisionCallback ) ) {
 420                         return call_user_func_array( $this->mRevisionCallback,
 421                                         array( $revision, $this ) );
 422                 } else {
 423                         return false;
 424                 }
 425         }
 426
 427         /**
 428          * Notify the callback function of a new log item
 429          * @param WikiRevision $revision
 430          * @return bool|mixed
 431          */
 432         private function logItemCallback( $revision ) {
 433                 if ( isset( $this->mLogItemCallback ) ) {
 434                         return call_user_func_array( $this->mLogItemCallback,
 435                                         array( $revision, $this ) );
 436                 } else {
 437                         return false;
 438                 }
 439         }
 440
 441         /**
 442          * Retrieves the contents of the named attribute of the current element.
 443          * @param string $attr The name of the attribute
 444          * @return string The value of the attribute or an empty string if it is not set in the current element.
 445          */
 446         public function nodeAttribute( $attr ) {
 447                 return $this->reader->getAttribute( $attr );
 448         }
 449
 450         /**
 451          * Shouldn't something like this be built-in to XMLReader?
 452          * Fetches text contents of the current element, assuming
 453          * no sub-elements or such scary things.
 454          * @return string
 455          * @access private
 456          */
 457         public function nodeContents() {
 458                 if ( $this->reader->isEmptyElement ) {
 459                         return "";
 460                 }
 461                 $buffer = "";
 462                 while ( $this->reader->read() ) {
 463                         switch ( $this->reader->nodeType ) {
 464                         case XMLReader::TEXT:
 465                         case XMLReader::SIGNIFICANT_WHITESPACE:
 466                                 $buffer .= $this->reader->value;
 467                                 break;
 468                         case XMLReader::END_ELEMENT:
 469                                 return $buffer;
 470                         }
 471                 }
 472
 473                 $this->reader->close();
 474                 return '';
 475         }
 476
 477         /**
 478          * Primary entry point
 479          * @throws MWException
 480          * @return bool
 481          */
 482         public function doImport() {
 483                 // Calls to reader->read need to be wrapped in calls to
 484                 // libxml_disable_entity_loader() to avoid local file
 485                 // inclusion attacks (bug 46932).
 486                 $oldDisable = libxml_disable_entity_loader( true );
 487                 $this->reader->read();
 488
 489                 if ( $this->reader->name != 'mediawiki' ) {
 490                         libxml_disable_entity_loader( $oldDisable );
 491                         throw new MWException( "Expected <mediawiki> tag, got " .
 492                                 $this->reader->name );
 493                 }
 494                 $this->debug( "<mediawiki> tag is correct." );
 495
 496                 $this->debug( "Starting primary dump processing loop." );
 497
 498                 $keepReading = $this->reader->read();
 499                 $skip = false;
 500                 while ( $keepReading ) {
 501                         $tag = $this->reader->name;
 502                         $type = $this->reader->nodeType;
 503
 504                         if ( !Hooks::run( 'ImportHandleToplevelXMLTag', array( $this ) ) ) {
 505                                 // Do nothing
 506                         } elseif ( $tag == 'mediawiki' && $type == XMLReader::END_ELEMENT ) {
 507                                 break;
 508                         } elseif ( $tag == 'siteinfo' ) {
 509                                 $this->handleSiteInfo();
 510                         } elseif ( $tag == 'page' ) {
 511                                 $this->handlePage();
 512                         } elseif ( $tag == 'logitem' ) {
 513                                 $this->handleLogItem();
 514                         } elseif ( $tag != '#text' ) {
 515                                 $this->warn( "Unhandled top-level XML tag $tag" );
 516
 517                                 $skip = true;
 518                         }
 519
 520                         if ( $skip ) {
 521                                 $keepReading = $this->reader->next();
 522                                 $skip = false;
 523                                 $this->debug( "Skip" );
 524                         } else {
 525                                 $keepReading = $this->reader->read();
 526                         }
 527                 }
 528
 529                 libxml_disable_entity_loader( $oldDisable );
 530                 return true;
 531         }
 532
 533         private function handleSiteInfo() {
 534                 $this->debug( "Enter site info handler." );
 535                 $siteInfo = array();
 536
 537                 // Fields that can just be stuffed in the siteInfo object
 538                 $normalFields = array( 'sitename', 'base', 'generator', 'case' );
 539
 540                 while ( $this->reader->read() ) {
 541                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 542                                         $this->reader->name == 'siteinfo' ) {
 543                                 break;
 544                         }
 545
 546                         $tag = $this->reader->name;
 547
 548                         if ( $tag == 'namespace' ) {
 549                                 $this->foreignNamespaces[ $this->nodeAttribute( 'key' ) ] =
 550                                         $this->nodeContents();
 551                         } elseif ( in_array( $tag, $normalFields ) ) {
 552                                 $siteInfo[$tag] = $this->nodeContents();
 553                         }
 554                 }
 555
 556                 $siteInfo['_namespaces'] = $this->foreignNamespaces;
 557                 $this->siteInfoCallback( $siteInfo );
 558         }
 559
 560         private function handleLogItem() {
 561                 $this->debug( "Enter log item handler." );
 562                 $logInfo = array();
 563
 564                 // Fields that can just be stuffed in the pageInfo object
 565                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 566                                         'logtitle', 'params' );
 567
 568                 while ( $this->reader->read() ) {
 569                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 570                                         $this->reader->name == 'logitem' ) {
 571                                 break;
 572                         }
 573
 574                         $tag = $this->reader->name;
 575
 576                         if ( !Hooks::run( 'ImportHandleLogItemXMLTag', array(
 577                                 $this, $logInfo
 578                         ) ) ) {
 579                                 // Do nothing
 580                         } elseif ( in_array( $tag, $normalFields ) ) {
 581                                 $logInfo[$tag] = $this->nodeContents();
 582                         } elseif ( $tag == 'contributor' ) {
 583                                 $logInfo['contributor'] = $this->handleContributor();
 584                         } elseif ( $tag != '#text' ) {
 585                                 $this->warn( "Unhandled log-item XML tag $tag" );
 586                         }
 587                 }
 588
 589                 $this->processLogItem( $logInfo );
 590         }
 591
 592         /**
 593          * @param array $logInfo
 594          * @return bool|mixed
 595          */
 596         private function processLogItem( $logInfo ) {
 597                 $revision = new WikiRevision( $this->config );
 598
 599                 $revision->setID( $logInfo['id'] );
 600                 $revision->setType( $logInfo['type'] );
 601                 $revision->setAction( $logInfo['action'] );
 602                 $revision->setTimestamp( $logInfo['timestamp'] );
 603                 $revision->setParams( $logInfo['params'] );
 604                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 605                 $revision->setNoUpdates( $this->mNoUpdates );
 606
 607                 if ( isset( $logInfo['comment'] ) ) {
 608                         $revision->setComment( $logInfo['comment'] );
 609                 }
 610
 611                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 612                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 613                 }
 614                 if ( isset( $logInfo['contributor']['username'] ) ) {
 615                         $revision->setUserName( $logInfo['contributor']['username'] );
 616                 }
 617
 618                 return $this->logItemCallback( $revision );
 619         }
 620
 621         private function handlePage() {
 622                 // Handle page data.
 623                 $this->debug( "Enter page handler." );
 624                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 625
 626                 // Fields that can just be stuffed in the pageInfo object
 627                 $normalFields = array( 'title', 'ns', 'id', 'redirect', 'restrictions' );
 628
 629                 $skip = false;
 630                 $badTitle = false;
 631
 632                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 633                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 634                                         $this->reader->name == 'page' ) {
 635                                 break;
 636                         }
 637
 638                         $skip = false;
 639
 640                         $tag = $this->reader->name;
 641
 642                         if ( $badTitle ) {
 643                                 // The title is invalid, bail out of this page
 644                                 $skip = true;
 645                         } elseif ( !Hooks::run( 'ImportHandlePageXMLTag', array( $this,
 646                                                 &$pageInfo ) ) ) {
 647                                 // Do nothing
 648                         } elseif ( in_array( $tag, $normalFields ) ) {
 649                                 // An XML snippet:
 650                                 // <page>
 651                                 //     <id>123</id>
 652                                 //     <title>Page</title>
 653                                 //     <redirect title="NewTitle"/>
 654                                 //     ...
 655                                 // Because the redirect tag is built differently, we need special handling for that case.
 656                                 if ( $tag == 'redirect' ) {
 657                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 658                                 } else {
 659                                         $pageInfo[$tag] = $this->nodeContents();
 660                                 }
 661                         } elseif ( $tag == 'revision' || $tag == 'upload' ) {
 662                                 if ( !isset( $title ) ) {
 663                                         $title = $this->processTitle( $pageInfo['title'],
 664                                                 isset( $pageInfo['ns'] ) ? $pageInfo['ns'] : null );
 665
 666                                         if ( !$title ) {
 667                                                 $badTitle = true;
 668                                                 $skip = true;
 669                                         }
 670
 671                                         $this->pageCallback( $title );
 672                                         list( $pageInfo['_title'], $foreignTitle ) = $title;
 673                                 }
 674
 675                                 if ( $title ) {
 676                                         if ( $tag == 'revision' ) {
 677                                                 $this->handleRevision( $pageInfo );
 678                                         } else {
 679                                                 $this->handleUpload( $pageInfo );
 680                                         }
 681                                 }
 682                         } elseif ( $tag != '#text' ) {
 683                                 $this->warn( "Unhandled page XML tag $tag" );
 684                                 $skip = true;
 685                         }
 686                 }
 687
 688                 $this->pageOutCallback( $pageInfo['_title'], $foreignTitle,
 689                                         $pageInfo['revisionCount'],
 690                                         $pageInfo['successfulRevisionCount'],
 691                                         $pageInfo );
 692         }
 693
 694         /**
 695          * @param array $pageInfo
 696          */
 697         private function handleRevision( &$pageInfo ) {
 698                 $this->debug( "Enter revision handler" );
 699                 $revisionInfo = array();
 700
 701                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' );
 702
 703                 $skip = false;
 704
 705                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 706                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 707                                         $this->reader->name == 'revision' ) {
 708                                 break;
 709                         }
 710
 711                         $tag = $this->reader->name;
 712
 713                         if ( !Hooks::run( 'ImportHandleRevisionXMLTag', array(
 714                                 $this, $pageInfo, $revisionInfo
 715                         ) ) ) {
 716                                 // Do nothing
 717                         } elseif ( in_array( $tag, $normalFields ) ) {
 718                                 $revisionInfo[$tag] = $this->nodeContents();
 719                         } elseif ( $tag == 'contributor' ) {
 720                                 $revisionInfo['contributor'] = $this->handleContributor();
 721                         } elseif ( $tag != '#text' ) {
 722                                 $this->warn( "Unhandled revision XML tag $tag" );
 723                                 $skip = true;
 724                         }
 725                 }
 726
 727                 $pageInfo['revisionCount']++;
 728                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 729                         $pageInfo['successfulRevisionCount']++;
 730                 }
 731         }
 732
 733         /**
 734          * @param array $pageInfo
 735          * @param array $revisionInfo
 736          * @return bool|mixed
 737          */
 738         private function processRevision( $pageInfo, $revisionInfo ) {
 739                 $revision = new WikiRevision( $this->config );
 740
 741                 if ( isset( $revisionInfo['id'] ) ) {
 742                         $revision->setID( $revisionInfo['id'] );
 743                 }
 744                 if ( isset( $revisionInfo['model'] ) ) {
 745                         $revision->setModel( $revisionInfo['model'] );
 746                 }
 747                 if ( isset( $revisionInfo['format'] ) ) {
 748                         $revision->setFormat( $revisionInfo['format'] );
 749                 }
 750                 $revision->setTitle( $pageInfo['_title'] );
 751
 752                 if ( isset( $revisionInfo['text'] ) ) {
 753                         $handler = $revision->getContentHandler();
 754                         $text = $handler->importTransform(
 755                                 $revisionInfo['text'],
 756                                 $revision->getFormat() );
 757
 758                         $revision->setText( $text );
 759                 }
 760                 if ( isset( $revisionInfo['timestamp'] ) ) {
 761                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 762                 } else {
 763                         $revision->setTimestamp( wfTimestampNow() );
 764                 }
 765
 766                 if ( isset( $revisionInfo['comment'] ) ) {
 767                         $revision->setComment( $revisionInfo['comment'] );
 768                 }
 769
 770                 if ( isset( $revisionInfo['minor'] ) ) {
 771                         $revision->setMinor( true );
 772                 }
 773                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 774                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 775                 }
 776                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 777                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 778                 }
 779                 $revision->setNoUpdates( $this->mNoUpdates );
 780
 781                 return $this->revisionCallback( $revision );
 782         }
 783
 784         /**
 785          * @param array $pageInfo
 786          * @return mixed
 787          */
 788         private function handleUpload( &$pageInfo ) {
 789                 $this->debug( "Enter upload handler" );
 790                 $uploadInfo = array();
 791
 792                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 793                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 794
 795                 $skip = false;
 796
 797                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 798                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 799                                         $this->reader->name == 'upload' ) {
 800                                 break;
 801                         }
 802
 803                         $tag = $this->reader->name;
 804
 805                         if ( !Hooks::run( 'ImportHandleUploadXMLTag', array(
 806                                 $this, $pageInfo
 807                         ) ) ) {
 808                                 // Do nothing
 809                         } elseif ( in_array( $tag, $normalFields ) ) {
 810                                 $uploadInfo[$tag] = $this->nodeContents();
 811                         } elseif ( $tag == 'contributor' ) {
 812                                 $uploadInfo['contributor'] = $this->handleContributor();
 813                         } elseif ( $tag == 'contents' ) {
 814                                 $contents = $this->nodeContents();
 815                                 $encoding = $this->reader->getAttribute( 'encoding' );
 816                                 if ( $encoding === 'base64' ) {
 817                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 818                                         $uploadInfo['isTempSrc'] = true;
 819                                 }
 820                         } elseif ( $tag != '#text' ) {
 821                                 $this->warn( "Unhandled upload XML tag $tag" );
 822                                 $skip = true;
 823                         }
 824                 }
 825
 826                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 827                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 828                         if ( file_exists( $path ) ) {
 829                                 $uploadInfo['fileSrc'] = $path;
 830                                 $uploadInfo['isTempSrc'] = false;
 831                         }
 832                 }
 833
 834                 if ( $this->mImportUploads ) {
 835                         return $this->processUpload( $pageInfo, $uploadInfo );
 836                 }
 837         }
 838
 839         /**
 840          * @param string $contents
 841          * @return string
 842          */
 843         private function dumpTemp( $contents ) {
 844                 $filename = tempnam( wfTempDir(), 'importupload' );
 845                 file_put_contents( $filename, $contents );
 846                 return $filename;
 847         }
 848
 849         /**
 850          * @param array $pageInfo
 851          * @param array $uploadInfo
 852          * @return mixed
 853          */
 854         private function processUpload( $pageInfo, $uploadInfo ) {
 855                 $revision = new WikiRevision( $this->config );
 856                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 857
 858                 $revision->setTitle( $pageInfo['_title'] );
 859                 $revision->setID( $pageInfo['id'] );
 860                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 861                 $revision->setText( $text );
 862                 $revision->setFilename( $uploadInfo['filename'] );
 863                 if ( isset( $uploadInfo['archivename'] ) ) {
 864                         $revision->setArchiveName( $uploadInfo['archivename'] );
 865                 }
 866                 $revision->setSrc( $uploadInfo['src'] );
 867                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 868                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 869                                 !empty( $uploadInfo['isTempSrc'] ) );
 870                 }
 871                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 872                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 873                 }
 874                 $revision->setSize( intval( $uploadInfo['size'] ) );
 875                 $revision->setComment( $uploadInfo['comment'] );
 876
 877                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 878                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 879                 }
 880                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 881                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 882                 }
 883                 $revision->setNoUpdates( $this->mNoUpdates );
 884
 885                 return call_user_func( $this->mUploadCallback, $revision );
 886         }
 887
 888         /**
 889          * @return array
 890          */
 891         private function handleContributor() {
 892                 $fields = array( 'id', 'ip', 'username' );
 893                 $info = array();
 894
 895                 while ( $this->reader->read() ) {
 896                         if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
 897                                         $this->reader->name == 'contributor' ) {
 898                                 break;
 899                         }
 900
 901                         $tag = $this->reader->name;
 902
 903                         if ( in_array( $tag, $fields ) ) {
 904                                 $info[$tag] = $this->nodeContents();
 905                         }
 906                 }
 907
 908                 return $info;
 909         }
 910
 911         /**
 912          * @param string $text
 913          * @param string|null $ns
 914          * @return array|bool
 915          */
 916         private function processTitle( $text, $ns = null ) {
 917                 if ( is_null( $this->foreignNamespaces ) ) {
 918                         $foreignTitleFactory = new NaiveForeignTitleFactory();
 919                 } else {
 920                         $foreignTitleFactory = new NamespaceAwareForeignTitleFactory(
 921                                 $this->foreignNamespaces );
 922                 }
 923
 924                 $foreignTitle = $foreignTitleFactory->createForeignTitle( $text,
 925                         intval( $ns ) );
 926
 927                 $title = $this->importTitleFactory->createTitleFromForeignTitle(
 928                         $foreignTitle );
 929
 930                 $commandLineMode = $this->config->get( 'CommandLineMode' );
 931                 if ( is_null( $title ) ) {
 932                         # Invalid page title? Ignore the page
 933                         $this->notice( 'import-error-invalid', $foreignTitle->getFullText() );
 934                         return false;
 935                 } elseif ( $title->isExternal() ) {
 936                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
 937                         return false;
 938                 } elseif ( !$title->canExist() ) {
 939                         $this->notice( 'import-error-special', $title->getPrefixedText() );
 940                         return false;
 941                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
 942                         # Do not import if the importing wiki user cannot edit this page
 943                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
 944                         return false;
 945                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
 946                         # Do not import if the importing wiki user cannot create this page
 947                         $this->notice( 'import-error-create', $title->getPrefixedText() );
 948                         return false;
 949                 }
 950
 951                 return array( $title, $foreignTitle );
 952         }
 953 }
 954
 955 /** This is a horrible hack used to keep source compatibility */
 956 class UploadSourceAdapter {
 957         /** @var array */
 958         public static $sourceRegistrations = array();
 959
 960         /** @var string */
 961         private $mSource;
 962
 963         /** @var string */
 964         private $mBuffer;
 965
 966         /** @var int */
 967         private $mPosition;
 968
 969         /**
 970          * @param ImportSource $source
 971          * @return string
 972          */
 973         static function registerSource( ImportSource $source ) {
 974                 $id = wfRandomString();
 975
 976                 self::$sourceRegistrations[$id] = $source;
 977
 978                 return $id;
 979         }
 980
 981         /**
 982          * @param string $path
 983          * @param string $mode
 984          * @param array $options
 985          * @param string $opened_path
 986          * @return bool
 987          */
 988         function stream_open( $path, $mode, $options, &$opened_path ) {
 989                 $url = parse_url( $path );
 990                 $id = $url['host'];
 991
 992                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
 993                         return false;
 994                 }
 995
 996                 $this->mSource = self::$sourceRegistrations[$id];
 997
 998                 return true;
 999         }
1000
1001         /**
1002          * @param int $count
1003          * @return string
1004          */
1005         function stream_read( $count ) {
1006                 $return = '';
1007                 $leave = false;
1008
1009                 while ( !$leave && !$this->mSource->atEnd() &&
1010                                 strlen( $this->mBuffer ) < $count ) {
1011                         $read = $this->mSource->readChunk();
1012
1013                         if ( !strlen( $read ) ) {
1014                                 $leave = true;
1015                         }
1016
1017                         $this->mBuffer .= $read;
1018                 }
1019
1020                 if ( strlen( $this->mBuffer ) ) {
1021                         $return = substr( $this->mBuffer, 0, $count );
1022                         $this->mBuffer = substr( $this->mBuffer, $count );
1023                 }
1024
1025                 $this->mPosition += strlen( $return );
1026
1027                 return $return;
1028         }
1029
1030         /**
1031          * @param string $data
1032          * @return bool
1033          */
1034         function stream_write( $data ) {
1035                 return false;
1036         }
1037
1038         /**
1039          * @return mixed
1040          */
1041         function stream_tell() {
1042                 return $this->mPosition;
1043         }
1044
1045         /**
1046          * @return bool
1047          */
1048         function stream_eof() {
1049                 return $this->mSource->atEnd();
1050         }
1051
1052         /**
1053          * @return array
1054          */
1055         function url_stat() {
1056                 $result = array();
1057
1058                 $result['dev'] = $result[0] = 0;
1059                 $result['ino'] = $result[1] = 0;
1060                 $result['mode'] = $result[2] = 0;
1061                 $result['nlink'] = $result[3] = 0;
1062                 $result['uid'] = $result[4] = 0;
1063                 $result['gid'] = $result[5] = 0;
1064                 $result['rdev'] = $result[6] = 0;
1065                 $result['size'] = $result[7] = 0;
1066                 $result['atime'] = $result[8] = 0;
1067                 $result['mtime'] = $result[9] = 0;
1068                 $result['ctime'] = $result[10] = 0;
1069                 $result['blksize'] = $result[11] = 0;
1070                 $result['blocks'] = $result[12] = 0;
1071
1072                 return $result;
1073         }
1074 }
1075
1076 /**
1077  * @todo document (e.g. one-sentence class description).
1078  * @ingroup SpecialPage
1079  */
1080 class WikiRevision {
1081         /** @todo Unused? */
1082         public $importer = null;
1083
1084         /** @var Title */
1085         public $title = null;
1086
1087         /** @var int */
1088         public $id = 0;
1089
1090         /** @var string */
1091         public $timestamp = "20010115000000";
1092
1093         /**
1094          * @var int
1095          * @todo Can't find any uses. Public, because that's suspicious. Get clarity. */
1096         public $user = 0;
1097
1098         /** @var string */
1099         public $user_text = "";
1100
1101         /** @var string */
1102         public $model = null;
1103
1104         /** @var string */
1105         public $format = null;
1106
1107         /** @var string */
1108         public $text = "";
1109
1110         /** @var int */
1111         protected $size;
1112
1113         /** @var Content */
1114         public $content = null;
1115
1116         /** @var ContentHandler */
1117         protected $contentHandler = null;
1118
1119         /** @var string */
1120         public $comment = "";
1121
1122         /** @var bool */
1123         public $minor = false;
1124
1125         /** @var string */
1126         public $type = "";
1127
1128         /** @var string */
1129         public $action = "";
1130
1131         /** @var string */
1132         public $params = "";
1133
1134         /** @var string */
1135         public $fileSrc = '';
1136
1137         /** @var bool|string */
1138         public $sha1base36 = false;
1139
1140         /**
1141          * @var bool
1142          * @todo Unused?
1143          */
1144         public $isTemp = false;
1145
1146         /** @var string */
1147         public $archiveName = '';
1148
1149         protected $filename;
1150
1151         /** @var mixed */
1152         protected $src;
1153
1154         /** @todo Unused? */
1155         public $fileIsTemp;
1156
1157         /** @var bool */
1158         private $mNoUpdates = false;
1159
1160         /** @var Config $config */
1161         private $config;
1162
1163         public function __construct( Config $config ) {
1164                 $this->config = $config;
1165         }
1166
1167         /**
1168          * @param Title $title
1169          * @throws MWException
1170          */
1171         function setTitle( $title ) {
1172                 if ( is_object( $title ) ) {
1173                         $this->title = $title;
1174                 } elseif ( is_null( $title ) ) {
1175                         throw new MWException( "WikiRevision given a null title in import. "
1176                                 . "You may need to adjust \$wgLegalTitleChars." );
1177                 } else {
1178                         throw new MWException( "WikiRevision given non-object title in import." );
1179                 }
1180         }
1181
1182         /**
1183          * @param int $id
1184          */
1185         function setID( $id ) {
1186                 $this->id = $id;
1187         }
1188
1189         /**
1190          * @param string $ts
1191          */
1192         function setTimestamp( $ts ) {
1193                 # 2003-08-05T18:30:02Z
1194                 $this->timestamp = wfTimestamp( TS_MW, $ts );
1195         }
1196
1197         /**
1198          * @param string $user
1199          */
1200         function setUsername( $user ) {
1201                 $this->user_text = $user;
1202         }
1203
1204         /**
1205          * @param string $ip
1206          */
1207         function setUserIP( $ip ) {
1208                 $this->user_text = $ip;
1209         }
1210
1211         /**
1212          * @param string $model
1213          */
1214         function setModel( $model ) {
1215                 $this->model = $model;
1216         }
1217
1218         /**
1219          * @param string $format
1220          */
1221         function setFormat( $format ) {
1222                 $this->format = $format;
1223         }
1224
1225         /**
1226          * @param string $text
1227          */
1228         function setText( $text ) {
1229                 $this->text = $text;
1230         }
1231
1232         /**
1233          * @param string $text
1234          */
1235         function setComment( $text ) {
1236                 $this->comment = $text;
1237         }
1238
1239         /**
1240          * @param bool $minor
1241          */
1242         function setMinor( $minor ) {
1243                 $this->minor = (bool)$minor;
1244         }
1245
1246         /**
1247          * @param mixed $src
1248          */
1249         function setSrc( $src ) {
1250                 $this->src = $src;
1251         }
1252
1253         /**
1254          * @param string $src
1255          * @param bool $isTemp
1256          */
1257         function setFileSrc( $src, $isTemp ) {
1258                 $this->fileSrc = $src;
1259                 $this->fileIsTemp = $isTemp;
1260         }
1261
1262         /**
1263          * @param string $sha1base36
1264          */
1265         function setSha1Base36( $sha1base36 ) {
1266                 $this->sha1base36 = $sha1base36;
1267         }
1268
1269         /**
1270          * @param string $filename
1271          */
1272         function setFilename( $filename ) {
1273                 $this->filename = $filename;
1274         }
1275
1276         /**
1277          * @param string $archiveName
1278          */
1279         function setArchiveName( $archiveName ) {
1280                 $this->archiveName = $archiveName;
1281         }
1282
1283         /**
1284          * @param int $size
1285          */
1286         function setSize( $size ) {
1287                 $this->size = intval( $size );
1288         }
1289
1290         /**
1291          * @param string $type
1292          */
1293         function setType( $type ) {
1294                 $this->type = $type;
1295         }
1296
1297         /**
1298          * @param string $action
1299          */
1300         function setAction( $action ) {
1301                 $this->action = $action;
1302         }
1303
1304         /**
1305          * @param array $params
1306          */
1307         function setParams( $params ) {
1308                 $this->params = $params;
1309         }
1310
1311         /**
1312          * @param bool $noupdates
1313          */
1314         public function setNoUpdates( $noupdates ) {
1315                 $this->mNoUpdates = $noupdates;
1316         }
1317
1318         /**
1319          * @return Title
1320          */
1321         function getTitle() {
1322                 return $this->title;
1323         }
1324
1325         /**
1326          * @return int
1327          */
1328         function getID() {
1329                 return $this->id;
1330         }
1331
1332         /**
1333          * @return string
1334          */
1335         function getTimestamp() {
1336                 return $this->timestamp;
1337         }
1338
1339         /**
1340          * @return string
1341          */
1342         function getUser() {
1343                 return $this->user_text;
1344         }
1345
1346         /**
1347          * @return string
1348          *
1349          * @deprecated Since 1.21, use getContent() instead.
1350          */
1351         function getText() {
1352                 ContentHandler::deprecated( __METHOD__, '1.21' );
1353
1354                 return $this->text;
1355         }
1356
1357         /**
1358          * @return ContentHandler
1359          */
1360         function getContentHandler() {
1361                 if ( is_null( $this->contentHandler ) ) {
1362                         $this->contentHandler = ContentHandler::getForModelID( $this->getModel() );
1363                 }
1364
1365                 return $this->contentHandler;
1366         }
1367
1368         /**
1369          * @return Content
1370          */
1371         function getContent() {
1372                 if ( is_null( $this->content ) ) {
1373                         $handler = $this->getContentHandler();
1374                         $this->content = $handler->unserializeContent( $this->text, $this->getFormat() );
1375                 }
1376
1377                 return $this->content;
1378         }
1379
1380         /**
1381          * @return string
1382          */
1383         function getModel() {
1384                 if ( is_null( $this->model ) ) {
1385                         $this->model = $this->getTitle()->getContentModel();
1386                 }
1387
1388                 return $this->model;
1389         }
1390
1391         /**
1392          * @return string
1393          */
1394         function getFormat() {
1395                 if ( is_null( $this->format ) ) {
1396                         $this->format = $this->getContentHandler()->getDefaultFormat();
1397                 }
1398
1399                 return $this->format;
1400         }
1401
1402         /**
1403          * @return string
1404          */
1405         function getComment() {
1406                 return $this->comment;
1407         }
1408
1409         /**
1410          * @return bool
1411          */
1412         function getMinor() {
1413                 return $this->minor;
1414         }
1415
1416         /**
1417          * @return mixed
1418          */
1419         function getSrc() {
1420                 return $this->src;
1421         }
1422
1423         /**
1424          * @return bool|string
1425          */
1426         function getSha1() {
1427                 if ( $this->sha1base36 ) {
1428                         return wfBaseConvert( $this->sha1base36, 36, 16 );
1429                 }
1430                 return false;
1431         }
1432
1433         /**
1434          * @return string
1435          */
1436         function getFileSrc() {
1437                 return $this->fileSrc;
1438         }
1439
1440         /**
1441          * @return bool
1442          */
1443         function isTempSrc() {
1444                 return $this->isTemp;
1445         }
1446
1447         /**
1448          * @return mixed
1449          */
1450         function getFilename() {
1451                 return $this->filename;
1452         }
1453
1454         /**
1455          * @return string
1456          */
1457         function getArchiveName() {
1458                 return $this->archiveName;
1459         }
1460
1461         /**
1462          * @return mixed
1463          */
1464         function getSize() {
1465                 return $this->size;
1466         }
1467
1468         /**
1469          * @return string
1470          */
1471         function getType() {
1472                 return $this->type;
1473         }
1474
1475         /**
1476          * @return string
1477          */
1478         function getAction() {
1479                 return $this->action;
1480         }
1481
1482         /**
1483          * @return string
1484          */
1485         function getParams() {
1486                 return $this->params;
1487         }
1488
1489         /**
1490          * @return bool
1491          */
1492         function importOldRevision() {
1493                 $dbw = wfGetDB( DB_MASTER );
1494
1495                 # Sneak a single revision into place
1496                 $user = User::newFromName( $this->getUser() );
1497                 if ( $user ) {
1498                         $userId = intval( $user->getId() );
1499                         $userText = $user->getName();
1500                         $userObj = $user;
1501                 } else {
1502                         $userId = 0;
1503                         $userText = $this->getUser();
1504                         $userObj = new User;
1505                 }
1506
1507                 // avoid memory leak...?
1508                 $linkCache = LinkCache::singleton();
1509                 $linkCache->clear();
1510
1511                 $page = WikiPage::factory( $this->title );
1512                 $page->loadPageData( 'fromdbmaster' );
1513                 if ( !$page->exists() ) {
1514                         # must create the page...
1515                         $pageId = $page->insertOn( $dbw );
1516                         $created = true;
1517                         $oldcountable = null;
1518                 } else {
1519                         $pageId = $page->getId();
1520                         $created = false;
1521
1522                         $prior = $dbw->selectField( 'revision', '1',
1523                                 array( 'rev_page' => $pageId,
1524                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1525                                         'rev_user_text' => $userText,
1526                                         'rev_comment' => $this->getComment() ),
1527                                 __METHOD__
1528                         );
1529                         if ( $prior ) {
1530                                 // @todo FIXME: This could fail slightly for multiple matches :P
1531                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1532                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1533                                 return false;
1534                         }
1535                         $oldcountable = $page->isCountable();
1536                 }
1537
1538                 # @todo FIXME: Use original rev_id optionally (better for backups)
1539                 # Insert the row
1540                 $revision = new Revision( array(
1541                         'title' => $this->title,
1542                         'page' => $pageId,
1543                         'content_model' => $this->getModel(),
1544                         'content_format' => $this->getFormat(),
1545                         //XXX: just set 'content' => $this->getContent()?
1546                         'text' => $this->getContent()->serialize( $this->getFormat() ),
1547                         'comment' => $this->getComment(),
1548                         'user' => $userId,
1549                         'user_text' => $userText,
1550                         'timestamp' => $this->timestamp,
1551                         'minor_edit' => $this->minor,
1552                         ) );
1553                 $revision->insertOn( $dbw );
1554                 $changed = $page->updateIfNewerOn( $dbw, $revision );
1555
1556                 if ( $changed !== false && !$this->mNoUpdates ) {
1557                         wfDebug( __METHOD__ . ": running updates\n" );
1558                         $page->doEditUpdates(
1559                                 $revision,
1560                                 $userObj,
1561                                 array( 'created' => $created, 'oldcountable' => $oldcountable )
1562                         );
1563                 }
1564
1565                 return true;
1566         }
1567
1568         function importLogItem() {
1569                 $dbw = wfGetDB( DB_MASTER );
1570                 # @todo FIXME: This will not record autoblocks
1571                 if ( !$this->getTitle() ) {
1572                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1573                                 $this->timestamp . "\n" );
1574                         return;
1575                 }
1576                 # Check if it exists already
1577                 // @todo FIXME: Use original log ID (better for backups)
1578                 $prior = $dbw->selectField( 'logging', '1',
1579                         array( 'log_type' => $this->getType(),
1580                                 'log_action' => $this->getAction(),
1581                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1582                                 'log_namespace' => $this->getTitle()->getNamespace(),
1583                                 'log_title' => $this->getTitle()->getDBkey(),
1584                                 'log_comment' => $this->getComment(),
1585                                 #'log_user_text' => $this->user_text,
1586                                 'log_params' => $this->params ),
1587                         __METHOD__
1588                 );
1589                 // @todo FIXME: This could fail slightly for multiple matches :P
1590                 if ( $prior ) {
1591                         wfDebug( __METHOD__
1592                                 . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp "
1593                                 . $this->timestamp . "\n" );
1594                         return;
1595                 }
1596                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1597                 $data = array(
1598                         'log_id' => $log_id,
1599                         'log_type' => $this->type,
1600                         'log_action' => $this->action,
1601                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1602                         'log_user' => User::idFromName( $this->user_text ),
1603                         #'log_user_text' => $this->user_text,
1604                         'log_namespace' => $this->getTitle()->getNamespace(),
1605                         'log_title' => $this->getTitle()->getDBkey(),
1606                         'log_comment' => $this->getComment(),
1607                         'log_params' => $this->params
1608                 );
1609                 $dbw->insert( 'logging', $data, __METHOD__ );
1610         }
1611
1612         /**
1613          * @return bool
1614          */
1615         function importUpload() {
1616                 # Construct a file
1617                 $archiveName = $this->getArchiveName();
1618                 if ( $archiveName ) {
1619                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1620                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1621                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1622                 } else {
1623                         $file = wfLocalFile( $this->getTitle() );
1624                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1625                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1626                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1627                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1628                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1629                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1630                         }
1631                 }
1632                 if ( !$file ) {
1633                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1634                         return false;
1635                 }
1636
1637                 # Get the file source or download if necessary
1638                 $source = $this->getFileSrc();
1639                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1640                 if ( !$source ) {
1641                         $source = $this->downloadSource();
1642                         $flags |= File::DELETE_SOURCE;
1643                 }
1644                 if ( !$source ) {
1645                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1646                         return false;
1647                 }
1648                 $sha1 = $this->getSha1();
1649                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1650                         if ( $flags & File::DELETE_SOURCE ) {
1651                                 # Broken file; delete it if it is a temporary file
1652                                 unlink( $source );
1653                         }
1654                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1655                         return false;
1656                 }
1657
1658                 $user = User::newFromName( $this->user_text );
1659
1660                 # Do the actual upload
1661                 if ( $archiveName ) {
1662                         $status = $file->uploadOld( $source, $archiveName,
1663                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1664                 } else {
1665                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1666                                 $flags, false, $this->getTimestamp(), $user );
1667                 }
1668
1669                 if ( $status->isGood() ) {
1670                         wfDebug( __METHOD__ . ": Successful\n" );
1671                         return true;
1672                 } else {
1673                         wfDebug( __METHOD__ . ': failed: ' . $status->getHTML() . "\n" );
1674                         return false;
1675                 }
1676         }
1677
1678         /**
1679          * @return bool|string
1680          */
1681         function downloadSource() {
1682                 if ( !$this->config->get( 'EnableUploads' ) ) {
1683                         return false;
1684                 }
1685
1686                 $tempo = tempnam( wfTempDir(), 'download' );
1687                 $f = fopen( $tempo, 'wb' );
1688                 if ( !$f ) {
1689                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1690                         return false;
1691                 }
1692
1693                 // @todo FIXME!
1694                 $src = $this->getSrc();
1695                 $data = Http::get( $src );
1696                 if ( !$data ) {
1697                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1698                         fclose( $f );
1699                         unlink( $tempo );
1700                         return false;
1701                 }
1702
1703                 fwrite( $f, $data );
1704                 fclose( $f );
1705
1706                 return $tempo;
1707         }
1708
1709 }
1710
1711 /**
1712  * Source interface for XML import.
1713  */
1714 interface ImportSource {
1715
1716         /**
1717          * Indicates whether the end of the input has been reached.
1718          * Will return true after a finite number of calls to readChunk.
1719          *
1720          * @return bool true if there is no more input, false otherwise.
1721          */
1722         function atEnd();
1723
1724         /**
1725          * Return a chunk of the input, as a (possibly empty) string.
1726          * When the end of input is reached, readChunk() returns false.
1727          * If atEnd() returns false, readChunk() will return a string.
1728          * If atEnd() returns true, readChunk() will return false.
1729          *
1730          * @return bool|string
1731          */
1732         function readChunk();
1733 }
1734
1735 /**
1736  * Used for importing XML dumps where the content of the dump is in a string.
1737  * This class is ineffecient, and should only be used for small dumps.
1738  * For larger dumps, ImportStreamSource should be used instead.
1739  *
1740  * @ingroup SpecialPage
1741  */
1742 class ImportStringSource implements ImportSource {
1743         function __construct( $string ) {
1744                 $this->mString = $string;
1745                 $this->mRead = false;
1746         }
1747
1748         /**
1749          * @return bool
1750          */
1751         function atEnd() {
1752                 return $this->mRead;
1753         }
1754
1755         /**
1756          * @return bool|string
1757          */
1758         function readChunk() {
1759                 if ( $this->atEnd() ) {
1760                         return false;
1761                 }
1762                 $this->mRead = true;
1763                 return $this->mString;
1764         }
1765 }
1766
1767 /**
1768  * Imports a XML dump from a file (either from file upload, files on disk, or HTTP)
1769  * @ingroup SpecialPage
1770  */
1771 class ImportStreamSource implements ImportSource {
1772         function __construct( $handle ) {
1773                 $this->mHandle = $handle;
1774         }
1775
1776         /**
1777          * @return bool
1778          */
1779         function atEnd() {
1780                 return feof( $this->mHandle );
1781         }
1782
1783         /**
1784          * @return string
1785          */
1786         function readChunk() {
1787                 return fread( $this->mHandle, 32768 );
1788         }
1789
1790         /**
1791          * @param string $filename
1792          * @return Status
1793          */
1794         static function newFromFile( $filename ) {
1795                 wfSuppressWarnings();
1796                 $file = fopen( $filename, 'rt' );
1797                 wfRestoreWarnings();
1798                 if ( !$file ) {
1799                         return Status::newFatal( "importcantopen" );
1800                 }
1801                 return Status::newGood( new ImportStreamSource( $file ) );
1802         }
1803
1804         /**
1805          * @param string $fieldname
1806          * @return Status
1807          */
1808         static function newFromUpload( $fieldname = "xmlimport" ) {
1809                 $upload =& $_FILES[$fieldname];
1810
1811                 if ( $upload === null || !$upload['name'] ) {
1812                         return Status::newFatal( 'importnofile' );
1813                 }
1814                 if ( !empty( $upload['error'] ) ) {
1815                         switch ( $upload['error'] ) {
1816                                 case 1:
1817                                         # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1818                                         return Status::newFatal( 'importuploaderrorsize' );
1819                                 case 2:
1820                                         # The uploaded file exceeds the MAX_FILE_SIZE directive that
1821                                         # was specified in the HTML form.
1822                                         return Status::newFatal( 'importuploaderrorsize' );
1823                                 case 3:
1824                                         # The uploaded file was only partially uploaded
1825                                         return Status::newFatal( 'importuploaderrorpartial' );
1826                                 case 6:
1827                                         # Missing a temporary folder.
1828                                         return Status::newFatal( 'importuploaderrortemp' );
1829                                 # case else: # Currently impossible
1830                         }
1831
1832                 }
1833                 $fname = $upload['tmp_name'];
1834                 if ( is_uploaded_file( $fname ) ) {
1835                         return ImportStreamSource::newFromFile( $fname );
1836                 } else {
1837                         return Status::newFatal( 'importnofile' );
1838                 }
1839         }
1840
1841         /**
1842          * @param string $url
1843          * @param string $method
1844          * @return Status
1845          */
1846         static function newFromURL( $url, $method = 'GET' ) {
1847                 wfDebug( __METHOD__ . ": opening $url\n" );
1848                 # Use the standard HTTP fetch function; it times out
1849                 # quicker and sorts out user-agent problems which might
1850                 # otherwise prevent importing from large sites, such
1851                 # as the Wikimedia cluster, etc.
1852                 $data = Http::request( $method, $url, array( 'followRedirects' => true ) );
1853                 if ( $data !== false ) {
1854                         $file = tmpfile();
1855                         fwrite( $file, $data );
1856                         fflush( $file );
1857                         fseek( $file, 0 );
1858                         return Status::newGood( new ImportStreamSource( $file ) );
1859                 } else {
1860                         return Status::newFatal( 'importcantopen' );
1861                 }
1862         }
1863
1864         /**
1865          * @param string $interwiki
1866          * @param string $page
1867          * @param bool $history
1868          * @param bool $templates
1869          * @param int $pageLinkDepth
1870          * @return Status
1871          */
1872         public static function newFromInterwiki( $interwiki, $page, $history = false,
1873                 $templates = false, $pageLinkDepth = 0
1874         ) {
1875                 if ( $page == '' ) {
1876                         return Status::newFatal( 'import-noarticle' );
1877                 }
1878                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1879                 if ( is_null( $link ) || !$link->isExternal() ) {
1880                         return Status::newFatal( 'importbadinterwiki' );
1881                 } else {
1882                         $params = array();
1883                         if ( $history ) {
1884                                 $params['history'] = 1;
1885                         }
1886                         if ( $templates ) {
1887                                 $params['templates'] = 1;
1888                         }
1889                         if ( $pageLinkDepth ) {
1890                                 $params['pagelink-depth'] = $pageLinkDepth;
1891                         }
1892                         $url = $link->getFullURL( $params );
1893                         # For interwikis, use POST to avoid redirects.
1894                         return ImportStreamSource::newFromURL( $url, "POST" );
1895                 }
1896         }
1897 }