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