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 $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
  36         private $mSiteInfoCallback, $mTargetNamespace, $mTargetRootPage, $mPageOutCallback;
  37         private $mNoticeCallback, $mDebug;
  38         private $mImportUploads, $mImageBasePath;
  39         private $mNoUpdates = false;
  40
  41         /**
  42          * Creates an ImportXMLReader drawing from the source provided
  43          * @param string $source
  44          */
  45         function __construct( $source ) {
  46                 $this->reader = new XMLReader();
  47
  48                 stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  49                 $id = UploadSourceAdapter::registerSource( $source );
  50                 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
  51                         $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  52                 } else {
  53                         $this->reader->open( "uploadsource://$id" );
  54                 }
  55
  56                 // Default callbacks
  57                 $this->setRevisionCallback( array( $this, "importRevision" ) );
  58                 $this->setUploadCallback( array( $this, 'importUpload' ) );
  59                 $this->setLogItemCallback( array( $this, 'importLogItem' ) );
  60                 $this->setPageOutCallback( array( $this, 'finishImportPage' ) );
  61         }
  62
  63         /**
  64          * @return null|XMLReader
  65          */
  66         public function getReader() {
  67                 return $this->reader;
  68         }
  69
  70         public function throwXmlError( $err ) {
  71                 $this->debug( "FAILURE: $err" );
  72                 wfDebug( "WikiImporter XML error: $err\n" );
  73         }
  74
  75         public function debug( $data ) {
  76                 if ( $this->mDebug ) {
  77                         wfDebug( "IMPORT: $data\n" );
  78                 }
  79         }
  80
  81         public function warn( $data ) {
  82                 wfDebug( "IMPORT: $data\n" );
  83         }
  84
  85         public function notice( $msg /*, $param, ...*/ ) {
  86                 $params = func_get_args();
  87                 array_shift( $params );
  88
  89                 if ( is_callable( $this->mNoticeCallback ) ) {
  90                         call_user_func( $this->mNoticeCallback, $msg, $params );
  91                 } else { # No ImportReporter -> CLI
  92                         echo wfMessage( $msg, $params )->text() . "\n";
  93                 }
  94         }
  95
  96         /**
  97          * Set debug mode...
  98          * @param bool $debug
  99          */
 100         function setDebug( $debug ) {
 101                 $this->mDebug = $debug;
 102         }
 103
 104         /**
 105          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
 106          * @param bool $noupdates
 107          */
 108         function setNoUpdates( $noupdates ) {
 109                 $this->mNoUpdates = $noupdates;
 110         }
 111
 112         /**
 113          * Set a callback that displays notice messages
 114          *
 115          * @param callable $callback
 116          * @return callable
 117          */
 118         public function setNoticeCallback( $callback ) {
 119                 return wfSetVar( $this->mNoticeCallback, $callback );
 120         }
 121
 122         /**
 123          * Sets the action to perform as each new page in the stream is reached.
 124          * @param callable $callback
 125          * @return callable
 126          */
 127         public function setPageCallback( $callback ) {
 128                 $previous = $this->mPageCallback;
 129                 $this->mPageCallback = $callback;
 130                 return $previous;
 131         }
 132
 133         /**
 134          * Sets the action to perform as each page in the stream is completed.
 135          * Callback accepts the page title (as a Title object), a second object
 136          * with the original title form (in case it's been overridden into a
 137          * local namespace), and a count of revisions.
 138          *
 139          * @param callable $callback
 140          * @return callable
 141          */
 142         public function setPageOutCallback( $callback ) {
 143                 $previous = $this->mPageOutCallback;
 144                 $this->mPageOutCallback = $callback;
 145                 return $previous;
 146         }
 147
 148         /**
 149          * Sets the action to perform as each page revision is reached.
 150          * @param callable $callback
 151          * @return callable
 152          */
 153         public function setRevisionCallback( $callback ) {
 154                 $previous = $this->mRevisionCallback;
 155                 $this->mRevisionCallback = $callback;
 156                 return $previous;
 157         }
 158
 159         /**
 160          * Sets the action to perform as each file upload version is reached.
 161          * @param callable $callback
 162          * @return callable
 163          */
 164         public function setUploadCallback( $callback ) {
 165                 $previous = $this->mUploadCallback;
 166                 $this->mUploadCallback = $callback;
 167                 return $previous;
 168         }
 169
 170         /**
 171          * Sets the action to perform as each log item reached.
 172          * @param callable $callback
 173          * @return callable
 174          */
 175         public function setLogItemCallback( $callback ) {
 176                 $previous = $this->mLogItemCallback;
 177                 $this->mLogItemCallback = $callback;
 178                 return $previous;
 179         }
 180
 181         /**
 182          * Sets the action to perform when site info is encountered
 183          * @param callable $callback
 184          * @return callable
 185          */
 186         public function setSiteInfoCallback( $callback ) {
 187                 $previous = $this->mSiteInfoCallback;
 188                 $this->mSiteInfoCallback = $callback;
 189                 return $previous;
 190         }
 191
 192         /**
 193          * Set a target namespace to override the defaults
 194          * @param null|int $namespace
 195          * @return bool
 196          */
 197         public function setTargetNamespace( $namespace ) {
 198                 if ( is_null( $namespace ) ) {
 199                         // Don't override namespaces
 200                         $this->mTargetNamespace = null;
 201                 } elseif ( $namespace >= 0 ) {
 202                         // @todo FIXME: Check for validity
 203                         $this->mTargetNamespace = intval( $namespace );
 204                 } else {
 205                         return false;
 206                 }
 207         }
 208
 209         /**
 210          * Set a target root page under which all pages are imported
 211          * @param null|string $rootpage
 212          * @return Status
 213          */
 214         public function setTargetRootPage( $rootpage ) {
 215                 $status = Status::newGood();
 216                 if ( is_null( $rootpage ) ) {
 217                         // No rootpage
 218                         $this->mTargetRootPage = null;
 219                 } elseif ( $rootpage !== '' ) {
 220                         $rootpage = rtrim( $rootpage, '/' ); //avoid double slashes
 221                         $title = Title::newFromText( $rootpage, !is_null( $this->mTargetNamespace )
 222                                 ? $this->mTargetNamespace
 223                                 : NS_MAIN
 224                         );
 225
 226                         if ( !$title || $title->isExternal() ) {
 227                                 $status->fatal( 'import-rootpage-invalid' );
 228                         } else {
 229                                 if ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 230                                         global $wgContLang;
 231
 232                                         $displayNSText = $title->getNamespace() == NS_MAIN
 233                                                 ? wfMessage( 'blanknamespace' )->text()
 234                                                 : $wgContLang->getNsText( $title->getNamespace() );
 235                                         $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 236                                 } else {
 237                                         // set namespace to 'all', so the namespace check in processTitle() can passed
 238                                         $this->setTargetNamespace( null );
 239                                         $this->mTargetRootPage = $title->getPrefixedDBkey();
 240                                 }
 241                         }
 242                 }
 243                 return $status;
 244         }
 245
 246         /**
 247          * @param string $dir
 248          */
 249         public function setImageBasePath( $dir ) {
 250                 $this->mImageBasePath = $dir;
 251         }
 252
 253         /**
 254          * @param bool $import
 255          */
 256         public function setImportUploads( $import ) {
 257                 $this->mImportUploads = $import;
 258         }
 259
 260         /**
 261          * Default per-revision callback, performs the import.
 262          * @param WikiRevision $revision
 263          * @return bool
 264          */
 265         public function importRevision( $revision ) {
 266                 if ( !$revision->getContent()->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
 267                         $this->notice( 'import-error-bad-location',
 268                                 $revision->getTitle()->getPrefixedText(),
 269                                 $revision->getID(),
 270                                 $revision->getModel(),
 271                                 $revision->getFormat() );
 272
 273                         return false;
 274                 }
 275
 276                 try {
 277                         $dbw = wfGetDB( DB_MASTER );
 278                         return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
 279                 } catch ( MWContentSerializationException $ex ) {
 280                         $this->notice( 'import-error-unserialize',
 281                                 $revision->getTitle()->getPrefixedText(),
 282                                 $revision->getID(),
 283                                 $revision->getModel(),
 284                                 $revision->getFormat() );
 285                 }
 286
 287                 return false;
 288         }
 289
 290         /**
 291          * Default per-revision callback, performs the import.
 292          * @param WikiRevision $revision
 293          * @return bool
 294          */
 295         public function importLogItem( $revision ) {
 296                 $dbw = wfGetDB( DB_MASTER );
 297                 return $dbw->deadlockLoop( array( $revision, 'importLogItem' ) );
 298         }
 299
 300         /**
 301          * Dummy for now...
 302          * @param WikiRevision $revision
 303          * @return bool
 304          */
 305         public function importUpload( $revision ) {
 306                 $dbw = wfGetDB( DB_MASTER );
 307                 return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
 308         }
 309
 310         /**
 311          * Mostly for hook use
 312          * @param Title $title
 313          * @param string $origTitle
 314          * @param int $revCount
 315          * @param int $sRevCount
 316          * @param array $pageInfo
 317          * @return bool
 318          */
 319         public function finishImportPage( $title, $origTitle, $revCount, $sRevCount, $pageInfo ) {
 320                 $args = func_get_args();
 321                 return wfRunHooks( 'AfterImportPage', $args );
 322         }
 323
 324         /**
 325          * Alternate per-revision callback, for debugging.
 326          * @param WikiRevision $revision
 327          */
 328         public function debugRevisionHandler( &$revision ) {
 329                 $this->debug( "Got revision:" );
 330                 if ( is_object( $revision->title ) ) {
 331                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 332                 } else {
 333                         $this->debug( "-- Title: <invalid>" );
 334                 }
 335                 $this->debug( "-- User: " . $revision->user_text );
 336                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 337                 $this->debug( "-- Comment: " . $revision->comment );
 338                 $this->debug( "-- Text: " . $revision->text );
 339         }
 340
 341         /**
 342          * Notify the callback function when a new "<page>" is reached.
 343          * @param Title $title
 344          */
 345         function pageCallback( $title ) {
 346                 if ( isset( $this->mPageCallback ) ) {
 347                         call_user_func( $this->mPageCallback, $title );
 348                 }
 349         }
 350
 351         /**
 352          * Notify the callback function when a "</page>" is closed.
 353          * @param Title $title
 354          * @param Title $origTitle
 355          * @param int $revCount
 356          * @param int $sucCount Number of revisions for which callback returned true
 357          * @param array $pageInfo Associative array of page information
 358          */
 359         private function pageOutCallback( $title, $origTitle, $revCount, $sucCount, $pageInfo ) {
 360                 if ( isset( $this->mPageOutCallback ) ) {
 361                         $args = func_get_args();
 362                         call_user_func_array( $this->mPageOutCallback, $args );
 363                 }
 364         }
 365
 366         /**
 367          * Notify the callback function of a revision
 368          * @param WikiRevision $revision
 369          * @return bool|mixed
 370          */
 371         private function revisionCallback( $revision ) {
 372                 if ( isset( $this->mRevisionCallback ) ) {
 373                         return call_user_func_array( $this->mRevisionCallback,
 374                                         array( $revision, $this ) );
 375                 } else {
 376                         return false;
 377                 }
 378         }
 379
 380         /**
 381          * Notify the callback function of a new log item
 382          * @param WikiRevision $revision
 383          * @return bool|mixed
 384          */
 385         private function logItemCallback( $revision ) {
 386                 if ( isset( $this->mLogItemCallback ) ) {
 387                         return call_user_func_array( $this->mLogItemCallback,
 388                                         array( $revision, $this ) );
 389                 } else {
 390                         return false;
 391                 }
 392         }
 393
 394         /**
 395          * Shouldn't something like this be built-in to XMLReader?
 396          * Fetches text contents of the current element, assuming
 397          * no sub-elements or such scary things.
 398          * @return string
 399          * @access private
 400          */
 401         public function nodeContents() {
 402                 if ( $this->reader->isEmptyElement ) {
 403                         return "";
 404                 }
 405                 $buffer = "";
 406                 while ( $this->reader->read() ) {
 407                         switch ( $this->reader->nodeType ) {
 408                         case XmlReader::TEXT:
 409                         case XmlReader::SIGNIFICANT_WHITESPACE:
 410                                 $buffer .= $this->reader->value;
 411                                 break;
 412                         case XmlReader::END_ELEMENT:
 413                                 return $buffer;
 414                         }
 415                 }
 416
 417                 $this->reader->close();
 418                 return '';
 419         }
 420
 421         # --------------
 422
 423         /** Left in for debugging */
 424         private function dumpElement() {
 425                 static $lookup = null;
 426                 if ( !$lookup ) {
 427                         $xmlReaderConstants = array(
 428                                 "NONE",
 429                                 "ELEMENT",
 430                                 "ATTRIBUTE",
 431                                 "TEXT",
 432                                 "CDATA",
 433                                 "ENTITY_REF",
 434                                 "ENTITY",
 435                                 "PI",
 436                                 "COMMENT",
 437                                 "DOC",
 438                                 "DOC_TYPE",
 439                                 "DOC_FRAGMENT",
 440                                 "NOTATION",
 441                                 "WHITESPACE",
 442                                 "SIGNIFICANT_WHITESPACE",
 443                                 "END_ELEMENT",
 444                                 "END_ENTITY",
 445                                 "XML_DECLARATION",
 446                         );
 447                         $lookup = array();
 448
 449                         foreach ( $xmlReaderConstants as $name ) {
 450                                 $lookup[constant( "XmlReader::$name" )] = $name;
 451                         }
 452                 }
 453
 454                 print var_dump(
 455                         $lookup[$this->reader->nodeType],
 456                         $this->reader->name,
 457                         $this->reader->value
 458                 ) . "\n\n";
 459         }
 460
 461         /**
 462          * Primary entry point
 463          * @throws MWException
 464          * @return bool
 465          */
 466         public function doImport() {
 467
 468                 // Calls to reader->read need to be wrapped in calls to
 469                 // libxml_disable_entity_loader() to avoid local file
 470                 // inclusion attacks (bug 46932).
 471                 $oldDisable = libxml_disable_entity_loader( true );
 472                 $this->reader->read();
 473
 474                 if ( $this->reader->name != 'mediawiki' ) {
 475                         libxml_disable_entity_loader( $oldDisable );
 476                         throw new MWException( "Expected <mediawiki> tag, got " .
 477                                 $this->reader->name );
 478                 }
 479                 $this->debug( "<mediawiki> tag is correct." );
 480
 481                 $this->debug( "Starting primary dump processing loop." );
 482
 483                 $keepReading = $this->reader->read();
 484                 $skip = false;
 485                 while ( $keepReading ) {
 486                         $tag = $this->reader->name;
 487                         $type = $this->reader->nodeType;
 488
 489                         if ( !wfRunHooks( 'ImportHandleToplevelXMLTag', array( $this ) ) ) {
 490                                 // Do nothing
 491                         } elseif ( $tag == 'mediawiki' && $type == XmlReader::END_ELEMENT ) {
 492                                 break;
 493                         } elseif ( $tag == 'siteinfo' ) {
 494                                 $this->handleSiteInfo();
 495                         } elseif ( $tag == 'page' ) {
 496                                 $this->handlePage();
 497                         } elseif ( $tag == 'logitem' ) {
 498                                 $this->handleLogItem();
 499                         } elseif ( $tag != '#text' ) {
 500                                 $this->warn( "Unhandled top-level XML tag $tag" );
 501
 502                                 $skip = true;
 503                         }
 504
 505                         if ( $skip ) {
 506                                 $keepReading = $this->reader->next();
 507                                 $skip = false;
 508                                 $this->debug( "Skip" );
 509                         } else {
 510                                 $keepReading = $this->reader->read();
 511                         }
 512                 }
 513
 514                 libxml_disable_entity_loader( $oldDisable );
 515                 return true;
 516         }
 517
 518         /**
 519          * @return bool
 520          * @throws MWException
 521          */
 522         private function handleSiteInfo() {
 523                 // Site info is useful, but not actually used for dump imports.
 524                 // Includes a quick short-circuit to save performance.
 525                 if ( ! $this->mSiteInfoCallback ) {
 526                         $this->reader->next();
 527                         return true;
 528                 }
 529                 throw new MWException( "SiteInfo tag is not yet handled, do not set mSiteInfoCallback" );
 530         }
 531
 532         private function handleLogItem() {
 533                 $this->debug( "Enter log item handler." );
 534                 $logInfo = array();
 535
 536                 // Fields that can just be stuffed in the pageInfo object
 537                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 538                                         'logtitle', 'params' );
 539
 540                 while ( $this->reader->read() ) {
 541                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 542                                         $this->reader->name == 'logitem' ) {
 543                                 break;
 544                         }
 545
 546                         $tag = $this->reader->name;
 547
 548                         if ( !wfRunHooks( 'ImportHandleLogItemXMLTag', array(
 549                                 $this, $logInfo
 550                         ) ) ) {
 551                                 // Do nothing
 552                         } elseif ( in_array( $tag, $normalFields ) ) {
 553                                 $logInfo[$tag] = $this->nodeContents();
 554                         } elseif ( $tag == 'contributor' ) {
 555                                 $logInfo['contributor'] = $this->handleContributor();
 556                         } elseif ( $tag != '#text' ) {
 557                                 $this->warn( "Unhandled log-item XML tag $tag" );
 558                         }
 559                 }
 560
 561                 $this->processLogItem( $logInfo );
 562         }
 563
 564         /**
 565          * @param array $logInfo
 566          * @return bool|mixed
 567          */
 568         private function processLogItem( $logInfo ) {
 569                 $revision = new WikiRevision;
 570
 571                 $revision->setID( $logInfo['id'] );
 572                 $revision->setType( $logInfo['type'] );
 573                 $revision->setAction( $logInfo['action'] );
 574                 $revision->setTimestamp( $logInfo['timestamp'] );
 575                 $revision->setParams( $logInfo['params'] );
 576                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 577                 $revision->setNoUpdates( $this->mNoUpdates );
 578
 579                 if ( isset( $logInfo['comment'] ) ) {
 580                         $revision->setComment( $logInfo['comment'] );
 581                 }
 582
 583                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 584                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 585                 }
 586                 if ( isset( $logInfo['contributor']['username'] ) ) {
 587                         $revision->setUserName( $logInfo['contributor']['username'] );
 588                 }
 589
 590                 return $this->logItemCallback( $revision );
 591         }
 592
 593         private function handlePage() {
 594                 // Handle page data.
 595                 $this->debug( "Enter page handler." );
 596                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 597
 598                 // Fields that can just be stuffed in the pageInfo object
 599                 $normalFields = array( 'title', 'id', 'redirect', 'restrictions' );
 600
 601                 $skip = false;
 602                 $badTitle = false;
 603
 604                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 605                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 606                                         $this->reader->name == 'page' ) {
 607                                 break;
 608                         }
 609
 610                         $tag = $this->reader->name;
 611
 612                         if ( $badTitle ) {
 613                                 // The title is invalid, bail out of this page
 614                                 $skip = true;
 615                         } elseif ( !wfRunHooks( 'ImportHandlePageXMLTag', array( $this,
 616                                                 &$pageInfo ) ) ) {
 617                                 // Do nothing
 618                         } elseif ( in_array( $tag, $normalFields ) ) {
 619                                 $pageInfo[$tag] = $this->nodeContents();
 620                                 if ( $tag == 'title' ) {
 621                                         $title = $this->processTitle( $pageInfo['title'] );
 622
 623                                         if ( !$title ) {
 624                                                 $badTitle = true;
 625                                                 $skip = true;
 626                                         }
 627
 628                                         $this->pageCallback( $title );
 629                                         list( $pageInfo['_title'], $origTitle ) = $title;
 630                                 }
 631                         } elseif ( $tag == 'revision' ) {
 632                                 $this->handleRevision( $pageInfo );
 633                         } elseif ( $tag == 'upload' ) {
 634                                 $this->handleUpload( $pageInfo );
 635                         } elseif ( $tag != '#text' ) {
 636                                 $this->warn( "Unhandled page XML tag $tag" );
 637                                 $skip = true;
 638                         }
 639                 }
 640
 641                 $this->pageOutCallback( $pageInfo['_title'], $origTitle,
 642                                         $pageInfo['revisionCount'],
 643                                         $pageInfo['successfulRevisionCount'],
 644                                         $pageInfo );
 645         }
 646
 647         /**
 648          * @param array $pageInfo
 649          */
 650         private function handleRevision( &$pageInfo ) {
 651                 $this->debug( "Enter revision handler" );
 652                 $revisionInfo = array();
 653
 654                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' );
 655
 656                 $skip = false;
 657
 658                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 659                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 660                                         $this->reader->name == 'revision' ) {
 661                                 break;
 662                         }
 663
 664                         $tag = $this->reader->name;
 665
 666                         if ( !wfRunHooks( 'ImportHandleRevisionXMLTag', array(
 667                                 $this, $pageInfo, $revisionInfo
 668                         ) ) ) {
 669                                 // Do nothing
 670                         } elseif ( in_array( $tag, $normalFields ) ) {
 671                                 $revisionInfo[$tag] = $this->nodeContents();
 672                         } elseif ( $tag == 'contributor' ) {
 673                                 $revisionInfo['contributor'] = $this->handleContributor();
 674                         } elseif ( $tag != '#text' ) {
 675                                 $this->warn( "Unhandled revision XML tag $tag" );
 676                                 $skip = true;
 677                         }
 678                 }
 679
 680                 $pageInfo['revisionCount']++;
 681                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 682                         $pageInfo['successfulRevisionCount']++;
 683                 }
 684         }
 685
 686         /**
 687          * @param array $pageInfo
 688          * @param array $revisionInfo
 689          * @return bool|mixed
 690          */
 691         private function processRevision( $pageInfo, $revisionInfo ) {
 692                 $revision = new WikiRevision;
 693
 694                 if ( isset( $revisionInfo['id'] ) ) {
 695                         $revision->setID( $revisionInfo['id'] );
 696                 }
 697                 if ( isset( $revisionInfo['text'] ) ) {
 698                         $revision->setText( $revisionInfo['text'] );
 699                 }
 700                 if ( isset( $revisionInfo['model'] ) ) {
 701                         $revision->setModel( $revisionInfo['model'] );
 702                 }
 703                 if ( isset( $revisionInfo['format'] ) ) {
 704                         $revision->setFormat( $revisionInfo['format'] );
 705                 }
 706                 $revision->setTitle( $pageInfo['_title'] );
 707
 708                 if ( isset( $revisionInfo['timestamp'] ) ) {
 709                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 710                 } else {
 711                         $revision->setTimestamp( wfTimestampNow() );
 712                 }
 713
 714                 if ( isset( $revisionInfo['comment'] ) ) {
 715                         $revision->setComment( $revisionInfo['comment'] );
 716                 }
 717
 718                 if ( isset( $revisionInfo['minor'] ) ) {
 719                         $revision->setMinor( true );
 720                 }
 721                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 722                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 723                 }
 724                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 725                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 726                 }
 727                 $revision->setNoUpdates( $this->mNoUpdates );
 728
 729                 return $this->revisionCallback( $revision );
 730         }
 731
 732         /**
 733          * @param array $pageInfo
 734          * @return mixed
 735          */
 736         private function handleUpload( &$pageInfo ) {
 737                 $this->debug( "Enter upload handler" );
 738                 $uploadInfo = array();
 739
 740                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 741                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 742
 743                 $skip = false;
 744
 745                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 746                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 747                                         $this->reader->name == 'upload' ) {
 748                                 break;
 749                         }
 750
 751                         $tag = $this->reader->name;
 752
 753                         if ( !wfRunHooks( 'ImportHandleUploadXMLTag', array(
 754                                 $this, $pageInfo
 755                         ) ) ) {
 756                                 // Do nothing
 757                         } elseif ( in_array( $tag, $normalFields ) ) {
 758                                 $uploadInfo[$tag] = $this->nodeContents();
 759                         } elseif ( $tag == 'contributor' ) {
 760                                 $uploadInfo['contributor'] = $this->handleContributor();
 761                         } elseif ( $tag == 'contents' ) {
 762                                 $contents = $this->nodeContents();
 763                                 $encoding = $this->reader->getAttribute( 'encoding' );
 764                                 if ( $encoding === 'base64' ) {
 765                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 766                                         $uploadInfo['isTempSrc'] = true;
 767                                 }
 768                         } elseif ( $tag != '#text' ) {
 769                                 $this->warn( "Unhandled upload XML tag $tag" );
 770                                 $skip = true;
 771                         }
 772                 }
 773
 774                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 775                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 776                         if ( file_exists( $path ) ) {
 777                                 $uploadInfo['fileSrc'] = $path;
 778                                 $uploadInfo['isTempSrc'] = false;
 779                         }
 780                 }
 781
 782                 if ( $this->mImportUploads ) {
 783                         return $this->processUpload( $pageInfo, $uploadInfo );
 784                 }
 785         }
 786
 787         /**
 788          * @param string $contents
 789          * @return string
 790          */
 791         private function dumpTemp( $contents ) {
 792                 $filename = tempnam( wfTempDir(), 'importupload' );
 793                 file_put_contents( $filename, $contents );
 794                 return $filename;
 795         }
 796
 797         /**
 798          * @param array $pageInfo
 799          * @param array $uploadInfo
 800          * @return mixed
 801          */
 802         private function processUpload( $pageInfo, $uploadInfo ) {
 803                 $revision = new WikiRevision;
 804                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 805
 806                 $revision->setTitle( $pageInfo['_title'] );
 807                 $revision->setID( $pageInfo['id'] );
 808                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 809                 $revision->setText( $text );
 810                 $revision->setFilename( $uploadInfo['filename'] );
 811                 if ( isset( $uploadInfo['archivename'] ) ) {
 812                         $revision->setArchiveName( $uploadInfo['archivename'] );
 813                 }
 814                 $revision->setSrc( $uploadInfo['src'] );
 815                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 816                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 817                                 !empty( $uploadInfo['isTempSrc'] ) );
 818                 }
 819                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 820                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 821                 }
 822                 $revision->setSize( intval( $uploadInfo['size'] ) );
 823                 $revision->setComment( $uploadInfo['comment'] );
 824
 825                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 826                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 827                 }
 828                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 829                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 830                 }
 831                 $revision->setNoUpdates( $this->mNoUpdates );
 832
 833                 return call_user_func( $this->mUploadCallback, $revision );
 834         }
 835
 836         /**
 837          * @return array
 838          */
 839         private function handleContributor() {
 840                 $fields = array( 'id', 'ip', 'username' );
 841                 $info = array();
 842
 843                 while ( $this->reader->read() ) {
 844                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 845                                         $this->reader->name == 'contributor' ) {
 846                                 break;
 847                         }
 848
 849                         $tag = $this->reader->name;
 850
 851                         if ( in_array( $tag, $fields ) ) {
 852                                 $info[$tag] = $this->nodeContents();
 853                         }
 854                 }
 855
 856                 return $info;
 857         }
 858
 859         /**
 860          * @param string $text
 861          * @return array|bool
 862          */
 863         private function processTitle( $text ) {
 864                 global $wgCommandLineMode;
 865
 866                 $workTitle = $text;
 867                 $origTitle = Title::newFromText( $workTitle );
 868
 869                 if ( !is_null( $this->mTargetNamespace ) && !is_null( $origTitle ) ) {
 870                         # makeTitleSafe, because $origTitle can have a interwiki (different setting of interwiki map)
 871                         # and than dbKey can begin with a lowercase char
 872                         $title = Title::makeTitleSafe( $this->mTargetNamespace,
 873                                 $origTitle->getDBkey() );
 874                 } else {
 875                         if ( !is_null( $this->mTargetRootPage ) ) {
 876                                 $workTitle = $this->mTargetRootPage . '/' . $workTitle;
 877                         }
 878                         $title = Title::newFromText( $workTitle );
 879                 }
 880
 881                 if ( is_null( $title ) ) {
 882                         # Invalid page title? Ignore the page
 883                         $this->notice( 'import-error-invalid', $workTitle );
 884                         return false;
 885                 } elseif ( $title->isExternal() ) {
 886                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
 887                         return false;
 888                 } elseif ( !$title->canExist() ) {
 889                         $this->notice( 'import-error-special', $title->getPrefixedText() );
 890                         return false;
 891                 } elseif ( !$title->userCan( 'edit' ) && !$wgCommandLineMode ) {
 892                         # Do not import if the importing wiki user cannot edit this page
 893                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
 894                         return false;
 895                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$wgCommandLineMode ) {
 896                         # Do not import if the importing wiki user cannot create this page
 897                         $this->notice( 'import-error-create', $title->getPrefixedText() );
 898                         return false;
 899                 }
 900
 901                 return array( $title, $origTitle );
 902         }
 903 }
 904
 905 /** This is a horrible hack used to keep source compatibility */
 906 class UploadSourceAdapter {
 907         /** @var array */
 908         private static $sourceRegistrations = array();
 909
 910         /** @var string */
 911         private $mSource;
 912
 913         /** @var string */
 914         private $mBuffer;
 915
 916         /** @var int */
 917         private $mPosition;
 918
 919         /**
 920          * @param string $source
 921          * @return string
 922          */
 923         static function registerSource( $source ) {
 924                 $id = wfRandomString();
 925
 926                 self::$sourceRegistrations[$id] = $source;
 927
 928                 return $id;
 929         }
 930
 931         /**
 932          * @param string $path
 933          * @param string $mode
 934          * @param array $options
 935          * @param string $opened_path
 936          * @return bool
 937          */
 938         function stream_open( $path, $mode, $options, &$opened_path ) {
 939                 $url = parse_url( $path );
 940                 $id = $url['host'];
 941
 942                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
 943                         return false;
 944                 }
 945
 946                 $this->mSource = self::$sourceRegistrations[$id];
 947
 948                 return true;
 949         }
 950
 951         /**
 952          * @param int $count
 953          * @return string
 954          */
 955         function stream_read( $count ) {
 956                 $return = '';
 957                 $leave = false;
 958
 959                 while ( !$leave && !$this->mSource->atEnd() &&
 960                                 strlen( $this->mBuffer ) < $count ) {
 961                         $read = $this->mSource->readChunk();
 962
 963                         if ( !strlen( $read ) ) {
 964                                 $leave = true;
 965                         }
 966
 967                         $this->mBuffer .= $read;
 968                 }
 969
 970                 if ( strlen( $this->mBuffer ) ) {
 971                         $return = substr( $this->mBuffer, 0, $count );
 972                         $this->mBuffer = substr( $this->mBuffer, $count );
 973                 }
 974
 975                 $this->mPosition += strlen( $return );
 976
 977                 return $return;
 978         }
 979
 980         /**
 981          * @param string $data
 982          * @return bool
 983          */
 984         function stream_write( $data ) {
 985                 return false;
 986         }
 987
 988         /**
 989          * @return mixed
 990          */
 991         function stream_tell() {
 992                 return $this->mPosition;
 993         }
 994
 995         /**
 996          * @return bool
 997          */
 998         function stream_eof() {
 999                 return $this->mSource->atEnd();
1000         }
1001
1002         /**
1003          * @return array
1004          */
1005         function url_stat() {
1006                 $result = array();
1007
1008                 $result['dev'] = $result[0] = 0;
1009                 $result['ino'] = $result[1] = 0;
1010                 $result['mode'] = $result[2] = 0;
1011                 $result['nlink'] = $result[3] = 0;
1012                 $result['uid'] = $result[4] = 0;
1013                 $result['gid'] = $result[5] = 0;
1014                 $result['rdev'] = $result[6] = 0;
1015                 $result['size'] = $result[7] = 0;
1016                 $result['atime'] = $result[8] = 0;
1017                 $result['mtime'] = $result[9] = 0;
1018                 $result['ctime'] = $result[10] = 0;
1019                 $result['blksize'] = $result[11] = 0;
1020                 $result['blocks'] = $result[12] = 0;
1021
1022                 return $result;
1023         }
1024 }
1025
1026 class XMLReader2 extends XMLReader {
1027
1028         /**
1029          * @return bool|string
1030          */
1031         function nodeContents() {
1032                 if ( $this->isEmptyElement ) {
1033                         return "";
1034                 }
1035                 $buffer = "";
1036                 while ( $this->read() ) {
1037                         switch ( $this->nodeType ) {
1038                         case XmlReader::TEXT:
1039                         case XmlReader::SIGNIFICANT_WHITESPACE:
1040                                 $buffer .= $this->value;
1041                                 break;
1042                         case XmlReader::END_ELEMENT:
1043                                 return $buffer;
1044                         }
1045                 }
1046                 return $this->close();
1047         }
1048 }
1049
1050 /**
1051  * @todo document (e.g. one-sentence class description).
1052  * @ingroup SpecialPage
1053  */
1054 class WikiRevision {
1055         /** @todo Unused? */
1056         private $importer = null;
1057
1058         /** @var Title */
1059         public $title = null;
1060
1061         /** @var int */
1062         private $id = 0;
1063
1064         /** @var string */
1065         public $timestamp = "20010115000000";
1066
1067         /**
1068          * @var int
1069          * @todo Can't find any uses. Public, because that's suspicious. Get clarity. */
1070         public $user = 0;
1071
1072         /** @var string */
1073         public $user_text = "";
1074
1075         /** @var string */
1076         protected $model = null;
1077
1078         /** @var string */
1079         protected $format = null;
1080
1081         /** @var string */
1082         public $text = "";
1083
1084         /** @var int */
1085         protected $size;
1086
1087         /** @var Content */
1088         protected $content = null;
1089
1090         /** @var string */
1091         public $comment = "";
1092
1093         /** @var bool */
1094         protected $minor = false;
1095
1096         /** @var string */
1097         protected $type = "";
1098
1099         /** @var string */
1100         protected $action = "";
1101
1102         /** @var string */
1103         protected $params = "";
1104
1105         /** @var string */
1106         protected $fileSrc = '';
1107
1108         /** @var bool|string */
1109         protected $sha1base36 = false;
1110
1111         /**
1112          * @var bool
1113          * @todo Unused?
1114          */
1115         private $isTemp = false;
1116
1117         /** @var string */
1118         protected $archiveName = '';
1119
1120         protected $filename;
1121
1122         /** @var mixed */
1123         protected $src;
1124
1125         /** @todo Unused? */
1126         private $fileIsTemp;
1127
1128         /** @var bool */
1129         private $mNoUpdates = false;
1130
1131         /**
1132          * @param Title $title
1133          * @throws MWException
1134          */
1135         function setTitle( $title ) {
1136                 if ( is_object( $title ) ) {
1137                         $this->title = $title;
1138                 } elseif ( is_null( $title ) ) {
1139                         throw new MWException( "WikiRevision given a null title in import. "
1140                                 . "You may need to adjust \$wgLegalTitleChars." );
1141                 } else {
1142                         throw new MWException( "WikiRevision given non-object title in import." );
1143                 }
1144         }
1145
1146         /**
1147          * @param int $id
1148          */
1149         function setID( $id ) {
1150                 $this->id = $id;
1151         }
1152
1153         /**
1154          * @param string $ts
1155          */
1156         function setTimestamp( $ts ) {
1157                 # 2003-08-05T18:30:02Z
1158                 $this->timestamp = wfTimestamp( TS_MW, $ts );
1159         }
1160
1161         /**
1162          * @param string $user
1163          */
1164         function setUsername( $user ) {
1165                 $this->user_text = $user;
1166         }
1167
1168         /**
1169          * @param string $ip
1170          */
1171         function setUserIP( $ip ) {
1172                 $this->user_text = $ip;
1173         }
1174
1175         /**
1176          * @param string $model
1177          */
1178         function setModel( $model ) {
1179                 $this->model = $model;
1180         }
1181
1182         /**
1183          * @param string $format
1184          */
1185         function setFormat( $format ) {
1186                 $this->format = $format;
1187         }
1188
1189         /**
1190          * @param string $text
1191          */
1192         function setText( $text ) {
1193                 $this->text = $text;
1194         }
1195
1196         /**
1197          * @param string $text
1198          */
1199         function setComment( $text ) {
1200                 $this->comment = $text;
1201         }
1202
1203         /**
1204          * @param bool $minor
1205          */
1206         function setMinor( $minor ) {
1207                 $this->minor = (bool)$minor;
1208         }
1209
1210         /**
1211          * @param mixed $src
1212          */
1213         function setSrc( $src ) {
1214                 $this->src = $src;
1215         }
1216
1217         /**
1218          * @param string $src
1219          * @param bool $isTemp
1220          */
1221         function setFileSrc( $src, $isTemp ) {
1222                 $this->fileSrc = $src;
1223                 $this->fileIsTemp = $isTemp;
1224         }
1225
1226         /**
1227          * @param string $sha1base36
1228          */
1229         function setSha1Base36( $sha1base36 ) {
1230                 $this->sha1base36 = $sha1base36;
1231         }
1232
1233         /**
1234          * @param string $filename
1235          */
1236         function setFilename( $filename ) {
1237                 $this->filename = $filename;
1238         }
1239
1240         /**
1241          * @param string $archiveName
1242          */
1243         function setArchiveName( $archiveName ) {
1244                 $this->archiveName = $archiveName;
1245         }
1246
1247         /**
1248          * @param int $size
1249          */
1250         function setSize( $size ) {
1251                 $this->size = intval( $size );
1252         }
1253
1254         /**
1255          * @param string $type
1256          */
1257         function setType( $type ) {
1258                 $this->type = $type;
1259         }
1260
1261         /**
1262          * @param string $action
1263          */
1264         function setAction( $action ) {
1265                 $this->action = $action;
1266         }
1267
1268         /**
1269          * @param array $params
1270          */
1271         function setParams( $params ) {
1272                 $this->params = $params;
1273         }
1274
1275         /**
1276          * @param bool $noupdates
1277          */
1278         public function setNoUpdates( $noupdates ) {
1279                 $this->mNoUpdates = $noupdates;
1280         }
1281
1282         /**
1283          * @return Title
1284          */
1285         function getTitle() {
1286                 return $this->title;
1287         }
1288
1289         /**
1290          * @return int
1291          */
1292         function getID() {
1293                 return $this->id;
1294         }
1295
1296         /**
1297          * @return string
1298          */
1299         function getTimestamp() {
1300                 return $this->timestamp;
1301         }
1302
1303         /**
1304          * @return string
1305          */
1306         function getUser() {
1307                 return $this->user_text;
1308         }
1309
1310         /**
1311          * @return string
1312          *
1313          * @deprecated Since 1.21, use getContent() instead.
1314          */
1315         function getText() {
1316                 ContentHandler::deprecated( __METHOD__, '1.21' );
1317
1318                 return $this->text;
1319         }
1320
1321         /**
1322          * @return Content
1323          */
1324         function getContent() {
1325                 if ( is_null( $this->content ) ) {
1326                         $this->content =
1327                                 ContentHandler::makeContent(
1328                                         $this->text,
1329                                         $this->getTitle(),
1330                                         $this->getModel(),
1331                                         $this->getFormat()
1332                                 );
1333                 }
1334
1335                 return $this->content;
1336         }
1337
1338         /**
1339          * @return string
1340          */
1341         function getModel() {
1342                 if ( is_null( $this->model ) ) {
1343                         $this->model = $this->getTitle()->getContentModel();
1344                 }
1345
1346                 return $this->model;
1347         }
1348
1349         /**
1350          * @return string
1351          */
1352         function getFormat() {
1353                 if ( is_null( $this->model ) ) {
1354                         $this->format = ContentHandler::getForTitle( $this->getTitle() )->getDefaultFormat();
1355                 }
1356
1357                 return $this->format;
1358         }
1359
1360         /**
1361          * @return string
1362          */
1363         function getComment() {
1364                 return $this->comment;
1365         }
1366
1367         /**
1368          * @return bool
1369          */
1370         function getMinor() {
1371                 return $this->minor;
1372         }
1373
1374         /**
1375          * @return mixed
1376          */
1377         function getSrc() {
1378                 return $this->src;
1379         }
1380
1381         /**
1382          * @return bool|string
1383          */
1384         function getSha1() {
1385                 if ( $this->sha1base36 ) {
1386                         return wfBaseConvert( $this->sha1base36, 36, 16 );
1387                 }
1388                 return false;
1389         }
1390
1391         /**
1392          * @return string
1393          */
1394         function getFileSrc() {
1395                 return $this->fileSrc;
1396         }
1397
1398         /**
1399          * @return bool
1400          */
1401         function isTempSrc() {
1402                 return $this->isTemp;
1403         }
1404
1405         /**
1406          * @return mixed
1407          */
1408         function getFilename() {
1409                 return $this->filename;
1410         }
1411
1412         /**
1413          * @return string
1414          */
1415         function getArchiveName() {
1416                 return $this->archiveName;
1417         }
1418
1419         /**
1420          * @return mixed
1421          */
1422         function getSize() {
1423                 return $this->size;
1424         }
1425
1426         /**
1427          * @return string
1428          */
1429         function getType() {
1430                 return $this->type;
1431         }
1432
1433         /**
1434          * @return string
1435          */
1436         function getAction() {
1437                 return $this->action;
1438         }
1439
1440         /**
1441          * @return string
1442          */
1443         function getParams() {
1444                 return $this->params;
1445         }
1446
1447         /**
1448          * @return bool
1449          */
1450         function importOldRevision() {
1451                 $dbw = wfGetDB( DB_MASTER );
1452
1453                 # Sneak a single revision into place
1454                 $user = User::newFromName( $this->getUser() );
1455                 if ( $user ) {
1456                         $userId = intval( $user->getId() );
1457                         $userText = $user->getName();
1458                         $userObj = $user;
1459                 } else {
1460                         $userId = 0;
1461                         $userText = $this->getUser();
1462                         $userObj = new User;
1463                 }
1464
1465                 // avoid memory leak...?
1466                 $linkCache = LinkCache::singleton();
1467                 $linkCache->clear();
1468
1469                 $page = WikiPage::factory( $this->title );
1470                 if ( !$page->exists() ) {
1471                         # must create the page...
1472                         $pageId = $page->insertOn( $dbw );
1473                         $created = true;
1474                         $oldcountable = null;
1475                 } else {
1476                         $pageId = $page->getId();
1477                         $created = false;
1478
1479                         $prior = $dbw->selectField( 'revision', '1',
1480                                 array( 'rev_page' => $pageId,
1481                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1482                                         'rev_user_text' => $userText,
1483                                         'rev_comment' => $this->getComment() ),
1484                                 __METHOD__
1485                         );
1486                         if ( $prior ) {
1487                                 // @todo FIXME: This could fail slightly for multiple matches :P
1488                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1489                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1490                                 return false;
1491                         }
1492                         $oldcountable = $page->isCountable();
1493                 }
1494
1495                 # @todo FIXME: Use original rev_id optionally (better for backups)
1496                 # Insert the row
1497                 $revision = new Revision( array(
1498                         'title' => $this->title,
1499                         'page' => $pageId,
1500                         'content_model' => $this->getModel(),
1501                         'content_format' => $this->getFormat(),
1502                         //XXX: just set 'content' => $this->getContent()?
1503                         'text' => $this->getContent()->serialize( $this->getFormat() ),
1504                         'comment' => $this->getComment(),
1505                         'user' => $userId,
1506                         'user_text' => $userText,
1507                         'timestamp' => $this->timestamp,
1508                         'minor_edit' => $this->minor,
1509                         ) );
1510                 $revision->insertOn( $dbw );
1511                 $changed = $page->updateIfNewerOn( $dbw, $revision );
1512
1513                 if ( $changed !== false && !$this->mNoUpdates ) {
1514                         wfDebug( __METHOD__ . ": running updates\n" );
1515                         $page->doEditUpdates(
1516                                 $revision,
1517                                 $userObj,
1518                                 array( 'created' => $created, 'oldcountable' => $oldcountable )
1519                         );
1520                 }
1521
1522                 return true;
1523         }
1524
1525         /**
1526          * @return mixed
1527          */
1528         function importLogItem() {
1529                 $dbw = wfGetDB( DB_MASTER );
1530                 # @todo FIXME: This will not record autoblocks
1531                 if ( !$this->getTitle() ) {
1532                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1533                                 $this->timestamp . "\n" );
1534                         return;
1535                 }
1536                 # Check if it exists already
1537                 // @todo FIXME: Use original log ID (better for backups)
1538                 $prior = $dbw->selectField( 'logging', '1',
1539                         array( 'log_type' => $this->getType(),
1540                                 'log_action' => $this->getAction(),
1541                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1542                                 'log_namespace' => $this->getTitle()->getNamespace(),
1543                                 'log_title' => $this->getTitle()->getDBkey(),
1544                                 'log_comment' => $this->getComment(),
1545                                 #'log_user_text' => $this->user_text,
1546                                 'log_params' => $this->params ),
1547                         __METHOD__
1548                 );
1549                 // @todo FIXME: This could fail slightly for multiple matches :P
1550                 if ( $prior ) {
1551                         wfDebug( __METHOD__
1552                                 . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp "
1553                                 . $this->timestamp . "\n" );
1554                         return;
1555                 }
1556                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1557                 $data = array(
1558                         'log_id' => $log_id,
1559                         'log_type' => $this->type,
1560                         'log_action' => $this->action,
1561                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1562                         'log_user' => User::idFromName( $this->user_text ),
1563                         #'log_user_text' => $this->user_text,
1564                         'log_namespace' => $this->getTitle()->getNamespace(),
1565                         'log_title' => $this->getTitle()->getDBkey(),
1566                         'log_comment' => $this->getComment(),
1567                         'log_params' => $this->params
1568                 );
1569                 $dbw->insert( 'logging', $data, __METHOD__ );
1570         }
1571
1572         /**
1573          * @return bool
1574          */
1575         function importUpload() {
1576                 # Construct a file
1577                 $archiveName = $this->getArchiveName();
1578                 if ( $archiveName ) {
1579                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1580                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1581                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1582                 } else {
1583                         $file = wfLocalFile( $this->getTitle() );
1584                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1585                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1586                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1587                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1588                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1589                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1590                         }
1591                 }
1592                 if ( !$file ) {
1593                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1594                         return false;
1595                 }
1596
1597                 # Get the file source or download if necessary
1598                 $source = $this->getFileSrc();
1599                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1600                 if ( !$source ) {
1601                         $source = $this->downloadSource();
1602                         $flags |= File::DELETE_SOURCE;
1603                 }
1604                 if ( !$source ) {
1605                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1606                         return false;
1607                 }
1608                 $sha1 = $this->getSha1();
1609                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1610                         if ( $flags & File::DELETE_SOURCE ) {
1611                                 # Broken file; delete it if it is a temporary file
1612                                 unlink( $source );
1613                         }
1614                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1615                         return false;
1616                 }
1617
1618                 $user = User::newFromName( $this->user_text );
1619
1620                 # Do the actual upload
1621                 if ( $archiveName ) {
1622                         $status = $file->uploadOld( $source, $archiveName,
1623                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1624                 } else {
1625                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1626                                 $flags, false, $this->getTimestamp(), $user );
1627                 }
1628
1629                 if ( $status->isGood() ) {
1630                         wfDebug( __METHOD__ . ": Successful\n" );
1631                         return true;
1632                 } else {
1633                         wfDebug( __METHOD__ . ': failed: ' . $status->getXml() . "\n" );
1634                         return false;
1635                 }
1636         }
1637
1638         /**
1639          * @return bool|string
1640          */
1641         function downloadSource() {
1642                 global $wgEnableUploads;
1643                 if ( !$wgEnableUploads ) {
1644                         return false;
1645                 }
1646
1647                 $tempo = tempnam( wfTempDir(), 'download' );
1648                 $f = fopen( $tempo, 'wb' );
1649                 if ( !$f ) {
1650                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1651                         return false;
1652                 }
1653
1654                 // @todo FIXME!
1655                 $src = $this->getSrc();
1656                 $data = Http::get( $src );
1657                 if ( !$data ) {
1658                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1659                         fclose( $f );
1660                         unlink( $tempo );
1661                         return false;
1662                 }
1663
1664                 fwrite( $f, $data );
1665                 fclose( $f );
1666
1667                 return $tempo;
1668         }
1669
1670 }
1671
1672 /**
1673  * @todo document (e.g. one-sentence class description).
1674  * @ingroup SpecialPage
1675  */
1676 class ImportStringSource {
1677         function __construct( $string ) {
1678                 $this->mString = $string;
1679                 $this->mRead = false;
1680         }
1681
1682         /**
1683          * @return bool
1684          */
1685         function atEnd() {
1686                 return $this->mRead;
1687         }
1688
1689         /**
1690          * @return bool|string
1691          */
1692         function readChunk() {
1693                 if ( $this->atEnd() ) {
1694                         return false;
1695                 }
1696                 $this->mRead = true;
1697                 return $this->mString;
1698         }
1699 }
1700
1701 /**
1702  * @todo document (e.g. one-sentence class description).
1703  * @ingroup SpecialPage
1704  */
1705 class ImportStreamSource {
1706         function __construct( $handle ) {
1707                 $this->mHandle = $handle;
1708         }
1709
1710         /**
1711          * @return bool
1712          */
1713         function atEnd() {
1714                 return feof( $this->mHandle );
1715         }
1716
1717         /**
1718          * @return string
1719          */
1720         function readChunk() {
1721                 return fread( $this->mHandle, 32768 );
1722         }
1723
1724         /**
1725          * @param string $filename
1726          * @return Status
1727          */
1728         static function newFromFile( $filename ) {
1729                 wfSuppressWarnings();
1730                 $file = fopen( $filename, 'rt' );
1731                 wfRestoreWarnings();
1732                 if ( !$file ) {
1733                         return Status::newFatal( "importcantopen" );
1734                 }
1735                 return Status::newGood( new ImportStreamSource( $file ) );
1736         }
1737
1738         /**
1739          * @param string $fieldname
1740          * @return Status
1741          */
1742         static function newFromUpload( $fieldname = "xmlimport" ) {
1743                 $upload =& $_FILES[$fieldname];
1744
1745                 if ( $upload === null || !$upload['name'] ) {
1746                         return Status::newFatal( 'importnofile' );
1747                 }
1748                 if ( !empty( $upload['error'] ) ) {
1749                         switch ( $upload['error'] ) {
1750                                 case 1:
1751                                         # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1752                                         return Status::newFatal( 'importuploaderrorsize' );
1753                                 case 2:
1754                                         # The uploaded file exceeds the MAX_FILE_SIZE directive that
1755                                         # was specified in the HTML form.
1756                                         return Status::newFatal( 'importuploaderrorsize' );
1757                                 case 3:
1758                                         # The uploaded file was only partially uploaded
1759                                         return Status::newFatal( 'importuploaderrorpartial' );
1760                                 case 6:
1761                                         # Missing a temporary folder.
1762                                         return Status::newFatal( 'importuploaderrortemp' );
1763                                 # case else: # Currently impossible
1764                         }
1765
1766                 }
1767                 $fname = $upload['tmp_name'];
1768                 if ( is_uploaded_file( $fname ) ) {
1769                         return ImportStreamSource::newFromFile( $fname );
1770                 } else {
1771                         return Status::newFatal( 'importnofile' );
1772                 }
1773         }
1774
1775         /**
1776          * @param string $url
1777          * @param string $method
1778          * @return Status
1779          */
1780         static function newFromURL( $url, $method = 'GET' ) {
1781                 wfDebug( __METHOD__ . ": opening $url\n" );
1782                 # Use the standard HTTP fetch function; it times out
1783                 # quicker and sorts out user-agent problems which might
1784                 # otherwise prevent importing from large sites, such
1785                 # as the Wikimedia cluster, etc.
1786                 $data = Http::request( $method, $url, array( 'followRedirects' => true ) );
1787                 if ( $data !== false ) {
1788                         $file = tmpfile();
1789                         fwrite( $file, $data );
1790                         fflush( $file );
1791                         fseek( $file, 0 );
1792                         return Status::newGood( new ImportStreamSource( $file ) );
1793                 } else {
1794                         return Status::newFatal( 'importcantopen' );
1795                 }
1796         }
1797
1798         /**
1799          * @param string $interwiki
1800          * @param string $page
1801          * @param bool $history
1802          * @param bool $templates
1803          * @param int $pageLinkDepth
1804          * @return Status
1805          */
1806         public static function newFromInterwiki( $interwiki, $page, $history = false,
1807                 $templates = false, $pageLinkDepth = 0
1808         ) {
1809                 if ( $page == '' ) {
1810                         return Status::newFatal( 'import-noarticle' );
1811                 }
1812                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1813                 if ( is_null( $link ) || !$link->isExternal() ) {
1814                         return Status::newFatal( 'importbadinterwiki' );
1815                 } else {
1816                         $params = array();
1817                         if ( $history ) {
1818                                 $params['history'] = 1;
1819                         }
1820                         if ( $templates ) {
1821                                 $params['templates'] = 1;
1822                         }
1823                         if ( $pageLinkDepth ) {
1824                                 $params['pagelink-depth'] = $pageLinkDepth;
1825                         }
1826                         $url = $link->getFullURL( $params );
1827                         # For interwikis, use POST to avoid redirects.
1828                         return ImportStreamSource::newFromURL( $url, "POST" );
1829                 }
1830         }
1831 }