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                 if ( !in_array(  'uploadsource', stream_get_wrappers() ) ) {
  49                         stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  50                 }
  51                 $id = UploadSourceAdapter::registerSource( $source );
  52                 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
  53                         $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  54                 } else {
  55                         $this->reader->open( "uploadsource://$id" );
  56                 }
  57
  58                 // Default callbacks
  59                 $this->setRevisionCallback( array( $this, "importRevision" ) );
  60                 $this->setUploadCallback( array( $this, 'importUpload' ) );
  61                 $this->setLogItemCallback( array( $this, 'importLogItem' ) );
  62                 $this->setPageOutCallback( array( $this, 'finishImportPage' ) );
  63         }
  64
  65         /**
  66          * @return null|XMLReader
  67          */
  68         public function getReader() {
  69                 return $this->reader;
  70         }
  71
  72         public function throwXmlError( $err ) {
  73                 $this->debug( "FAILURE: $err" );
  74                 wfDebug( "WikiImporter XML error: $err\n" );
  75         }
  76
  77         public function debug( $data ) {
  78                 if ( $this->mDebug ) {
  79                         wfDebug( "IMPORT: $data\n" );
  80                 }
  81         }
  82
  83         public function warn( $data ) {
  84                 wfDebug( "IMPORT: $data\n" );
  85         }
  86
  87         public function notice( $msg /*, $param, ...*/ ) {
  88                 $params = func_get_args();
  89                 array_shift( $params );
  90
  91                 if ( is_callable( $this->mNoticeCallback ) ) {
  92                         call_user_func( $this->mNoticeCallback, $msg, $params );
  93                 } else { # No ImportReporter -> CLI
  94                         echo wfMessage( $msg, $params )->text() . "\n";
  95                 }
  96         }
  97
  98         /**
  99          * Set debug mode...
 100          * @param bool $debug
 101          */
 102         function setDebug( $debug ) {
 103                 $this->mDebug = $debug;
 104         }
 105
 106         /**
 107          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
 108          * @param bool $noupdates
 109          */
 110         function setNoUpdates( $noupdates ) {
 111                 $this->mNoUpdates = $noupdates;
 112         }
 113
 114         /**
 115          * Set a callback that displays notice messages
 116          *
 117          * @param callable $callback
 118          * @return callable
 119          */
 120         public function setNoticeCallback( $callback ) {
 121                 return wfSetVar( $this->mNoticeCallback, $callback );
 122         }
 123
 124         /**
 125          * Sets the action to perform as each new page in the stream is reached.
 126          * @param callable $callback
 127          * @return callable
 128          */
 129         public function setPageCallback( $callback ) {
 130                 $previous = $this->mPageCallback;
 131                 $this->mPageCallback = $callback;
 132                 return $previous;
 133         }
 134
 135         /**
 136          * Sets the action to perform as each page in the stream is completed.
 137          * Callback accepts the page title (as a Title object), a second object
 138          * with the original title form (in case it's been overridden into a
 139          * local namespace), and a count of revisions.
 140          *
 141          * @param callable $callback
 142          * @return callable
 143          */
 144         public function setPageOutCallback( $callback ) {
 145                 $previous = $this->mPageOutCallback;
 146                 $this->mPageOutCallback = $callback;
 147                 return $previous;
 148         }
 149
 150         /**
 151          * Sets the action to perform as each page revision is reached.
 152          * @param callable $callback
 153          * @return callable
 154          */
 155         public function setRevisionCallback( $callback ) {
 156                 $previous = $this->mRevisionCallback;
 157                 $this->mRevisionCallback = $callback;
 158                 return $previous;
 159         }
 160
 161         /**
 162          * Sets the action to perform as each file upload version is reached.
 163          * @param callable $callback
 164          * @return callable
 165          */
 166         public function setUploadCallback( $callback ) {
 167                 $previous = $this->mUploadCallback;
 168                 $this->mUploadCallback = $callback;
 169                 return $previous;
 170         }
 171
 172         /**
 173          * Sets the action to perform as each log item reached.
 174          * @param callable $callback
 175          * @return callable
 176          */
 177         public function setLogItemCallback( $callback ) {
 178                 $previous = $this->mLogItemCallback;
 179                 $this->mLogItemCallback = $callback;
 180                 return $previous;
 181         }
 182
 183         /**
 184          * Sets the action to perform when site info is encountered
 185          * @param callable $callback
 186          * @return callable
 187          */
 188         public function setSiteInfoCallback( $callback ) {
 189                 $previous = $this->mSiteInfoCallback;
 190                 $this->mSiteInfoCallback = $callback;
 191                 return $previous;
 192         }
 193
 194         /**
 195          * Set a target namespace to override the defaults
 196          * @param null|int $namespace
 197          * @return bool
 198          */
 199         public function setTargetNamespace( $namespace ) {
 200                 if ( is_null( $namespace ) ) {
 201                         // Don't override namespaces
 202                         $this->mTargetNamespace = null;
 203                 } elseif ( $namespace >= 0 ) {
 204                         // @todo FIXME: Check for validity
 205                         $this->mTargetNamespace = intval( $namespace );
 206                 } else {
 207                         return false;
 208                 }
 209         }
 210
 211         /**
 212          * Set a target root page under which all pages are imported
 213          * @param null|string $rootpage
 214          * @return Status
 215          */
 216         public function setTargetRootPage( $rootpage ) {
 217                 $status = Status::newGood();
 218                 if ( is_null( $rootpage ) ) {
 219                         // No rootpage
 220                         $this->mTargetRootPage = null;
 221                 } elseif ( $rootpage !== '' ) {
 222                         $rootpage = rtrim( $rootpage, '/' ); //avoid double slashes
 223                         $title = Title::newFromText( $rootpage, !is_null( $this->mTargetNamespace )
 224                                 ? $this->mTargetNamespace
 225                                 : NS_MAIN
 226                         );
 227
 228                         if ( !$title || $title->isExternal() ) {
 229                                 $status->fatal( 'import-rootpage-invalid' );
 230                         } else {
 231                                 if ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 232                                         global $wgContLang;
 233
 234                                         $displayNSText = $title->getNamespace() == NS_MAIN
 235                                                 ? wfMessage( 'blanknamespace' )->text()
 236                                                 : $wgContLang->getNsText( $title->getNamespace() );
 237                                         $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 238                                 } else {
 239                                         // set namespace to 'all', so the namespace check in processTitle() can passed
 240                                         $this->setTargetNamespace( null );
 241                                         $this->mTargetRootPage = $title->getPrefixedDBkey();
 242                                 }
 243                         }
 244                 }
 245                 return $status;
 246         }
 247
 248         /**
 249          * @param string $dir
 250          */
 251         public function setImageBasePath( $dir ) {
 252                 $this->mImageBasePath = $dir;
 253         }
 254
 255         /**
 256          * @param bool $import
 257          */
 258         public function setImportUploads( $import ) {
 259                 $this->mImportUploads = $import;
 260         }
 261
 262         /**
 263          * Default per-revision callback, performs the import.
 264          * @param WikiRevision $revision
 265          * @return bool
 266          */
 267         public function importRevision( $revision ) {
 268                 if ( !$revision->getContent()->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
 269                         $this->notice( 'import-error-bad-location',
 270                                 $revision->getTitle()->getPrefixedText(),
 271                                 $revision->getID(),
 272                                 $revision->getModel(),
 273                                 $revision->getFormat() );
 274
 275                         return false;
 276                 }
 277
 278                 try {
 279                         $dbw = wfGetDB( DB_MASTER );
 280                         return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
 281                 } catch ( MWContentSerializationException $ex ) {
 282                         $this->notice( 'import-error-unserialize',
 283                                 $revision->getTitle()->getPrefixedText(),
 284                                 $revision->getID(),
 285                                 $revision->getModel(),
 286                                 $revision->getFormat() );
 287                 }
 288
 289                 return false;
 290         }
 291
 292         /**
 293          * Default per-revision callback, performs the import.
 294          * @param WikiRevision $revision
 295          * @return bool
 296          */
 297         public function importLogItem( $revision ) {
 298                 $dbw = wfGetDB( DB_MASTER );
 299                 return $dbw->deadlockLoop( array( $revision, 'importLogItem' ) );
 300         }
 301
 302         /**
 303          * Dummy for now...
 304          * @param WikiRevision $revision
 305          * @return bool
 306          */
 307         public function importUpload( $revision ) {
 308                 $dbw = wfGetDB( DB_MASTER );
 309                 return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
 310         }
 311
 312         /**
 313          * Mostly for hook use
 314          * @param Title $title
 315          * @param string $origTitle
 316          * @param int $revCount
 317          * @param int $sRevCount
 318          * @param array $pageInfo
 319          * @return bool
 320          */
 321         public function finishImportPage( $title, $origTitle, $revCount, $sRevCount, $pageInfo ) {
 322                 $args = func_get_args();
 323                 return wfRunHooks( 'AfterImportPage', $args );
 324         }
 325
 326         /**
 327          * Alternate per-revision callback, for debugging.
 328          * @param WikiRevision $revision
 329          */
 330         public function debugRevisionHandler( &$revision ) {
 331                 $this->debug( "Got revision:" );
 332                 if ( is_object( $revision->title ) ) {
 333                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 334                 } else {
 335                         $this->debug( "-- Title: <invalid>" );
 336                 }
 337                 $this->debug( "-- User: " . $revision->user_text );
 338                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 339                 $this->debug( "-- Comment: " . $revision->comment );
 340                 $this->debug( "-- Text: " . $revision->text );
 341         }
 342
 343         /**
 344          * Notify the callback function when a new "<page>" is reached.
 345          * @param Title $title
 346          */
 347         function pageCallback( $title ) {
 348                 if ( isset( $this->mPageCallback ) ) {
 349                         call_user_func( $this->mPageCallback, $title );
 350                 }
 351         }
 352
 353         /**
 354          * Notify the callback function when a "</page>" is closed.
 355          * @param Title $title
 356          * @param Title $origTitle
 357          * @param int $revCount
 358          * @param int $sucCount Number of revisions for which callback returned true
 359          * @param array $pageInfo Associative array of page information
 360          */
 361         private function pageOutCallback( $title, $origTitle, $revCount, $sucCount, $pageInfo ) {
 362                 if ( isset( $this->mPageOutCallback ) ) {
 363                         $args = func_get_args();
 364                         call_user_func_array( $this->mPageOutCallback, $args );
 365                 }
 366         }
 367
 368         /**
 369          * Notify the callback function of a revision
 370          * @param WikiRevision $revision
 371          * @return bool|mixed
 372          */
 373         private function revisionCallback( $revision ) {
 374                 if ( isset( $this->mRevisionCallback ) ) {
 375                         return call_user_func_array( $this->mRevisionCallback,
 376                                         array( $revision, $this ) );
 377                 } else {
 378                         return false;
 379                 }
 380         }
 381
 382         /**
 383          * Notify the callback function of a new log item
 384          * @param WikiRevision $revision
 385          * @return bool|mixed
 386          */
 387         private function logItemCallback( $revision ) {
 388                 if ( isset( $this->mLogItemCallback ) ) {
 389                         return call_user_func_array( $this->mLogItemCallback,
 390                                         array( $revision, $this ) );
 391                 } else {
 392                         return false;
 393                 }
 394         }
 395
 396         /**
 397          * Shouldn't something like this be built-in to XMLReader?
 398          * Fetches text contents of the current element, assuming
 399          * no sub-elements or such scary things.
 400          * @return string
 401          * @access private
 402          */
 403         public function nodeContents() {
 404                 if ( $this->reader->isEmptyElement ) {
 405                         return "";
 406                 }
 407                 $buffer = "";
 408                 while ( $this->reader->read() ) {
 409                         switch ( $this->reader->nodeType ) {
 410                         case XmlReader::TEXT:
 411                         case XmlReader::SIGNIFICANT_WHITESPACE:
 412                                 $buffer .= $this->reader->value;
 413                                 break;
 414                         case XmlReader::END_ELEMENT:
 415                                 return $buffer;
 416                         }
 417                 }
 418
 419                 $this->reader->close();
 420                 return '';
 421         }
 422
 423         # --------------
 424
 425         /** Left in for debugging */
 426         private function dumpElement() {
 427                 static $lookup = null;
 428                 if ( !$lookup ) {
 429                         $xmlReaderConstants = array(
 430                                 "NONE",
 431                                 "ELEMENT",
 432                                 "ATTRIBUTE",
 433                                 "TEXT",
 434                                 "CDATA",
 435                                 "ENTITY_REF",
 436                                 "ENTITY",
 437                                 "PI",
 438                                 "COMMENT",
 439                                 "DOC",
 440                                 "DOC_TYPE",
 441                                 "DOC_FRAGMENT",
 442                                 "NOTATION",
 443                                 "WHITESPACE",
 444                                 "SIGNIFICANT_WHITESPACE",
 445                                 "END_ELEMENT",
 446                                 "END_ENTITY",
 447                                 "XML_DECLARATION",
 448                         );
 449                         $lookup = array();
 450
 451                         foreach ( $xmlReaderConstants as $name ) {
 452                                 $lookup[constant( "XmlReader::$name" )] = $name;
 453                         }
 454                 }
 455
 456                 print var_dump(
 457                         $lookup[$this->reader->nodeType],
 458                         $this->reader->name,
 459                         $this->reader->value
 460                 ) . "\n\n";
 461         }
 462
 463         /**
 464          * Primary entry point
 465          * @throws MWException
 466          * @return bool
 467          */
 468         public function doImport() {
 469
 470                 // Calls to reader->read need to be wrapped in calls to
 471                 // libxml_disable_entity_loader() to avoid local file
 472                 // inclusion attacks (bug 46932).
 473                 $oldDisable = libxml_disable_entity_loader( true );
 474                 $this->reader->read();
 475
 476                 if ( $this->reader->name != 'mediawiki' ) {
 477                         libxml_disable_entity_loader( $oldDisable );
 478                         throw new MWException( "Expected <mediawiki> tag, got " .
 479                                 $this->reader->name );
 480                 }
 481                 $this->debug( "<mediawiki> tag is correct." );
 482
 483                 $this->debug( "Starting primary dump processing loop." );
 484
 485                 $keepReading = $this->reader->read();
 486                 $skip = false;
 487                 while ( $keepReading ) {
 488                         $tag = $this->reader->name;
 489                         $type = $this->reader->nodeType;
 490
 491                         if ( !wfRunHooks( 'ImportHandleToplevelXMLTag', array( $this ) ) ) {
 492                                 // Do nothing
 493                         } elseif ( $tag == 'mediawiki' && $type == XmlReader::END_ELEMENT ) {
 494                                 break;
 495                         } elseif ( $tag == 'siteinfo' ) {
 496                                 $this->handleSiteInfo();
 497                         } elseif ( $tag == 'page' ) {
 498                                 $this->handlePage();
 499                         } elseif ( $tag == 'logitem' ) {
 500                                 $this->handleLogItem();
 501                         } elseif ( $tag != '#text' ) {
 502                                 $this->warn( "Unhandled top-level XML tag $tag" );
 503
 504                                 $skip = true;
 505                         }
 506
 507                         if ( $skip ) {
 508                                 $keepReading = $this->reader->next();
 509                                 $skip = false;
 510                                 $this->debug( "Skip" );
 511                         } else {
 512                                 $keepReading = $this->reader->read();
 513                         }
 514                 }
 515
 516                 libxml_disable_entity_loader( $oldDisable );
 517                 return true;
 518         }
 519
 520         /**
 521          * @return bool
 522          * @throws MWException
 523          */
 524         private function handleSiteInfo() {
 525                 // Site info is useful, but not actually used for dump imports.
 526                 // Includes a quick short-circuit to save performance.
 527                 if ( ! $this->mSiteInfoCallback ) {
 528                         $this->reader->next();
 529                         return true;
 530                 }
 531                 throw new MWException( "SiteInfo tag is not yet handled, do not set mSiteInfoCallback" );
 532         }
 533
 534         private function handleLogItem() {
 535                 $this->debug( "Enter log item handler." );
 536                 $logInfo = array();
 537
 538                 // Fields that can just be stuffed in the pageInfo object
 539                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 540                                         'logtitle', 'params' );
 541
 542                 while ( $this->reader->read() ) {
 543                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 544                                         $this->reader->name == 'logitem' ) {
 545                                 break;
 546                         }
 547
 548                         $tag = $this->reader->name;
 549
 550                         if ( !wfRunHooks( 'ImportHandleLogItemXMLTag', array(
 551                                 $this, $logInfo
 552                         ) ) ) {
 553                                 // Do nothing
 554                         } elseif ( in_array( $tag, $normalFields ) ) {
 555                                 $logInfo[$tag] = $this->nodeContents();
 556                         } elseif ( $tag == 'contributor' ) {
 557                                 $logInfo['contributor'] = $this->handleContributor();
 558                         } elseif ( $tag != '#text' ) {
 559                                 $this->warn( "Unhandled log-item XML tag $tag" );
 560                         }
 561                 }
 562
 563                 $this->processLogItem( $logInfo );
 564         }
 565
 566         /**
 567          * @param array $logInfo
 568          * @return bool|mixed
 569          */
 570         private function processLogItem( $logInfo ) {
 571                 $revision = new WikiRevision;
 572
 573                 $revision->setID( $logInfo['id'] );
 574                 $revision->setType( $logInfo['type'] );
 575                 $revision->setAction( $logInfo['action'] );
 576                 $revision->setTimestamp( $logInfo['timestamp'] );
 577                 $revision->setParams( $logInfo['params'] );
 578                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 579                 $revision->setNoUpdates( $this->mNoUpdates );
 580
 581                 if ( isset( $logInfo['comment'] ) ) {
 582                         $revision->setComment( $logInfo['comment'] );
 583                 }
 584
 585                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 586                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 587                 }
 588                 if ( isset( $logInfo['contributor']['username'] ) ) {
 589                         $revision->setUserName( $logInfo['contributor']['username'] );
 590                 }
 591
 592                 return $this->logItemCallback( $revision );
 593         }
 594
 595         private function handlePage() {
 596                 // Handle page data.
 597                 $this->debug( "Enter page handler." );
 598                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 599
 600                 // Fields that can just be stuffed in the pageInfo object
 601                 $normalFields = array( 'title', 'id', 'redirect', 'restrictions' );
 602
 603                 $skip = false;
 604                 $badTitle = false;
 605
 606                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 607                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 608                                         $this->reader->name == 'page' ) {
 609                                 break;
 610                         }
 611
 612                         $tag = $this->reader->name;
 613
 614                         if ( $badTitle ) {
 615                                 // The title is invalid, bail out of this page
 616                                 $skip = true;
 617                         } elseif ( !wfRunHooks( 'ImportHandlePageXMLTag', array( $this,
 618                                                 &$pageInfo ) ) ) {
 619                                 // Do nothing
 620                         } elseif ( in_array( $tag, $normalFields ) ) {
 621                                 $pageInfo[$tag] = $this->nodeContents();
 622                                 if ( $tag == 'title' ) {
 623                                         $title = $this->processTitle( $pageInfo['title'] );
 624
 625                                         if ( !$title ) {
 626                                                 $badTitle = true;
 627                                                 $skip = true;
 628                                         }
 629
 630                                         $this->pageCallback( $title );
 631                                         list( $pageInfo['_title'], $origTitle ) = $title;
 632                                 }
 633                         } elseif ( $tag == 'revision' ) {
 634                                 $this->handleRevision( $pageInfo );
 635                         } elseif ( $tag == 'upload' ) {
 636                                 $this->handleUpload( $pageInfo );
 637                         } elseif ( $tag != '#text' ) {
 638                                 $this->warn( "Unhandled page XML tag $tag" );
 639                                 $skip = true;
 640                         }
 641                 }
 642
 643                 $this->pageOutCallback( $pageInfo['_title'], $origTitle,
 644                                         $pageInfo['revisionCount'],
 645                                         $pageInfo['successfulRevisionCount'],
 646                                         $pageInfo );
 647         }
 648
 649         /**
 650          * @param array $pageInfo
 651          */
 652         private function handleRevision( &$pageInfo ) {
 653                 $this->debug( "Enter revision handler" );
 654                 $revisionInfo = array();
 655
 656                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' );
 657
 658                 $skip = false;
 659
 660                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 661                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 662                                         $this->reader->name == 'revision' ) {
 663                                 break;
 664                         }
 665
 666                         $tag = $this->reader->name;
 667
 668                         if ( !wfRunHooks( 'ImportHandleRevisionXMLTag', array(
 669                                 $this, $pageInfo, $revisionInfo
 670                         ) ) ) {
 671                                 // Do nothing
 672                         } elseif ( in_array( $tag, $normalFields ) ) {
 673                                 $revisionInfo[$tag] = $this->nodeContents();
 674                         } elseif ( $tag == 'contributor' ) {
 675                                 $revisionInfo['contributor'] = $this->handleContributor();
 676                         } elseif ( $tag != '#text' ) {
 677                                 $this->warn( "Unhandled revision XML tag $tag" );
 678                                 $skip = true;
 679                         }
 680                 }
 681
 682                 $pageInfo['revisionCount']++;
 683                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 684                         $pageInfo['successfulRevisionCount']++;
 685                 }
 686         }
 687
 688         /**
 689          * @param array $pageInfo
 690          * @param array $revisionInfo
 691          * @return bool|mixed
 692          */
 693         private function processRevision( $pageInfo, $revisionInfo ) {
 694                 $revision = new WikiRevision;
 695
 696                 if ( isset( $revisionInfo['id'] ) ) {
 697                         $revision->setID( $revisionInfo['id'] );
 698                 }
 699                 if ( isset( $revisionInfo['text'] ) ) {
 700                         $revision->setText( $revisionInfo['text'] );
 701                 }
 702                 if ( isset( $revisionInfo['model'] ) ) {
 703                         $revision->setModel( $revisionInfo['model'] );
 704                 }
 705                 if ( isset( $revisionInfo['format'] ) ) {
 706                         $revision->setFormat( $revisionInfo['format'] );
 707                 }
 708                 $revision->setTitle( $pageInfo['_title'] );
 709
 710                 if ( isset( $revisionInfo['timestamp'] ) ) {
 711                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 712                 } else {
 713                         $revision->setTimestamp( wfTimestampNow() );
 714                 }
 715
 716                 if ( isset( $revisionInfo['comment'] ) ) {
 717                         $revision->setComment( $revisionInfo['comment'] );
 718                 }
 719
 720                 if ( isset( $revisionInfo['minor'] ) ) {
 721                         $revision->setMinor( true );
 722                 }
 723                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 724                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 725                 }
 726                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 727                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 728                 }
 729                 $revision->setNoUpdates( $this->mNoUpdates );
 730
 731                 return $this->revisionCallback( $revision );
 732         }
 733
 734         /**
 735          * @param array $pageInfo
 736          * @return mixed
 737          */
 738         private function handleUpload( &$pageInfo ) {
 739                 $this->debug( "Enter upload handler" );
 740                 $uploadInfo = array();
 741
 742                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 743                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 744
 745                 $skip = false;
 746
 747                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 748                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 749                                         $this->reader->name == 'upload' ) {
 750                                 break;
 751                         }
 752
 753                         $tag = $this->reader->name;
 754
 755                         if ( !wfRunHooks( 'ImportHandleUploadXMLTag', array(
 756                                 $this, $pageInfo
 757                         ) ) ) {
 758                                 // Do nothing
 759                         } elseif ( in_array( $tag, $normalFields ) ) {
 760                                 $uploadInfo[$tag] = $this->nodeContents();
 761                         } elseif ( $tag == 'contributor' ) {
 762                                 $uploadInfo['contributor'] = $this->handleContributor();
 763                         } elseif ( $tag == 'contents' ) {
 764                                 $contents = $this->nodeContents();
 765                                 $encoding = $this->reader->getAttribute( 'encoding' );
 766                                 if ( $encoding === 'base64' ) {
 767                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 768                                         $uploadInfo['isTempSrc'] = true;
 769                                 }
 770                         } elseif ( $tag != '#text' ) {
 771                                 $this->warn( "Unhandled upload XML tag $tag" );
 772                                 $skip = true;
 773                         }
 774                 }
 775
 776                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 777                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 778                         if ( file_exists( $path ) ) {
 779                                 $uploadInfo['fileSrc'] = $path;
 780                                 $uploadInfo['isTempSrc'] = false;
 781                         }
 782                 }
 783
 784                 if ( $this->mImportUploads ) {
 785                         return $this->processUpload( $pageInfo, $uploadInfo );
 786                 }
 787         }
 788
 789         /**
 790          * @param string $contents
 791          * @return string
 792          */
 793         private function dumpTemp( $contents ) {
 794                 $filename = tempnam( wfTempDir(), 'importupload' );
 795                 file_put_contents( $filename, $contents );
 796                 return $filename;
 797         }
 798
 799         /**
 800          * @param array $pageInfo
 801          * @param array $uploadInfo
 802          * @return mixed
 803          */
 804         private function processUpload( $pageInfo, $uploadInfo ) {
 805                 $revision = new WikiRevision;
 806                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 807
 808                 $revision->setTitle( $pageInfo['_title'] );
 809                 $revision->setID( $pageInfo['id'] );
 810                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 811                 $revision->setText( $text );
 812                 $revision->setFilename( $uploadInfo['filename'] );
 813                 if ( isset( $uploadInfo['archivename'] ) ) {
 814                         $revision->setArchiveName( $uploadInfo['archivename'] );
 815                 }
 816                 $revision->setSrc( $uploadInfo['src'] );
 817                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 818                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 819                                 !empty( $uploadInfo['isTempSrc'] ) );
 820                 }
 821                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 822                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 823                 }
 824                 $revision->setSize( intval( $uploadInfo['size'] ) );
 825                 $revision->setComment( $uploadInfo['comment'] );
 826
 827                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 828                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 829                 }
 830                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 831                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 832                 }
 833                 $revision->setNoUpdates( $this->mNoUpdates );
 834
 835                 return call_user_func( $this->mUploadCallback, $revision );
 836         }
 837
 838         /**
 839          * @return array
 840          */
 841         private function handleContributor() {
 842                 $fields = array( 'id', 'ip', 'username' );
 843                 $info = array();
 844
 845                 while ( $this->reader->read() ) {
 846                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 847                                         $this->reader->name == 'contributor' ) {
 848                                 break;
 849                         }
 850
 851                         $tag = $this->reader->name;
 852
 853                         if ( in_array( $tag, $fields ) ) {
 854                                 $info[$tag] = $this->nodeContents();
 855                         }
 856                 }
 857
 858                 return $info;
 859         }
 860
 861         /**
 862          * @param string $text
 863          * @return array|bool
 864          */
 865         private function processTitle( $text ) {
 866                 global $wgCommandLineMode;
 867
 868                 $workTitle = $text;
 869                 $origTitle = Title::newFromText( $workTitle );
 870
 871                 if ( !is_null( $this->mTargetNamespace ) && !is_null( $origTitle ) ) {
 872                         # makeTitleSafe, because $origTitle can have a interwiki (different setting of interwiki map)
 873                         # and than dbKey can begin with a lowercase char
 874                         $title = Title::makeTitleSafe( $this->mTargetNamespace,
 875                                 $origTitle->getDBkey() );
 876                 } else {
 877                         if ( !is_null( $this->mTargetRootPage ) ) {
 878                                 $workTitle = $this->mTargetRootPage . '/' . $workTitle;
 879                         }
 880                         $title = Title::newFromText( $workTitle );
 881                 }
 882
 883                 if ( is_null( $title ) ) {
 884                         # Invalid page title? Ignore the page
 885                         $this->notice( 'import-error-invalid', $workTitle );
 886                         return false;
 887                 } elseif ( $title->isExternal() ) {
 888                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
 889                         return false;
 890                 } elseif ( !$title->canExist() ) {
 891                         $this->notice( 'import-error-special', $title->getPrefixedText() );
 892                         return false;
 893                 } elseif ( !$title->userCan( 'edit' ) && !$wgCommandLineMode ) {
 894                         # Do not import if the importing wiki user cannot edit this page
 895                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
 896                         return false;
 897                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$wgCommandLineMode ) {
 898                         # Do not import if the importing wiki user cannot create this page
 899                         $this->notice( 'import-error-create', $title->getPrefixedText() );
 900                         return false;
 901                 }
 902
 903                 return array( $title, $origTitle );
 904         }
 905 }
 906
 907 /** This is a horrible hack used to keep source compatibility */
 908 class UploadSourceAdapter {
 909         /** @var array */
 910         private static $sourceRegistrations = array();
 911
 912         /** @var string */
 913         private $mSource;
 914
 915         /** @var string */
 916         private $mBuffer;
 917
 918         /** @var int */
 919         private $mPosition;
 920
 921         /**
 922          * @param string $source
 923          * @return string
 924          */
 925         static function registerSource( $source ) {
 926                 $id = wfRandomString();
 927
 928                 self::$sourceRegistrations[$id] = $source;
 929
 930                 return $id;
 931         }
 932
 933         /**
 934          * @param string $path
 935          * @param string $mode
 936          * @param array $options
 937          * @param string $opened_path
 938          * @return bool
 939          */
 940         function stream_open( $path, $mode, $options, &$opened_path ) {
 941                 $url = parse_url( $path );
 942                 $id = $url['host'];
 943
 944                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
 945                         return false;
 946                 }
 947
 948                 $this->mSource = self::$sourceRegistrations[$id];
 949
 950                 return true;
 951         }
 952
 953         /**
 954          * @param int $count
 955          * @return string
 956          */
 957         function stream_read( $count ) {
 958                 $return = '';
 959                 $leave = false;
 960
 961                 while ( !$leave && !$this->mSource->atEnd() &&
 962                                 strlen( $this->mBuffer ) < $count ) {
 963                         $read = $this->mSource->readChunk();
 964
 965                         if ( !strlen( $read ) ) {
 966                                 $leave = true;
 967                         }
 968
 969                         $this->mBuffer .= $read;
 970                 }
 971
 972                 if ( strlen( $this->mBuffer ) ) {
 973                         $return = substr( $this->mBuffer, 0, $count );
 974                         $this->mBuffer = substr( $this->mBuffer, $count );
 975                 }
 976
 977                 $this->mPosition += strlen( $return );
 978
 979                 return $return;
 980         }
 981
 982         /**
 983          * @param string $data
 984          * @return bool
 985          */
 986         function stream_write( $data ) {
 987                 return false;
 988         }
 989
 990         /**
 991          * @return mixed
 992          */
 993         function stream_tell() {
 994                 return $this->mPosition;
 995         }
 996
 997         /**
 998          * @return bool
 999          */
1000         function stream_eof() {
1001                 return $this->mSource->atEnd();
1002         }
1003
1004         /**
1005          * @return array
1006          */
1007         function url_stat() {
1008                 $result = array();
1009
1010                 $result['dev'] = $result[0] = 0;
1011                 $result['ino'] = $result[1] = 0;
1012                 $result['mode'] = $result[2] = 0;
1013                 $result['nlink'] = $result[3] = 0;
1014                 $result['uid'] = $result[4] = 0;
1015                 $result['gid'] = $result[5] = 0;
1016                 $result['rdev'] = $result[6] = 0;
1017                 $result['size'] = $result[7] = 0;
1018                 $result['atime'] = $result[8] = 0;
1019                 $result['mtime'] = $result[9] = 0;
1020                 $result['ctime'] = $result[10] = 0;
1021                 $result['blksize'] = $result[11] = 0;
1022                 $result['blocks'] = $result[12] = 0;
1023
1024                 return $result;
1025         }
1026 }
1027
1028 class XMLReader2 extends XMLReader {
1029
1030         /**
1031          * @return bool|string
1032          */
1033         function nodeContents() {
1034                 if ( $this->isEmptyElement ) {
1035                         return "";
1036                 }
1037                 $buffer = "";
1038                 while ( $this->read() ) {
1039                         switch ( $this->nodeType ) {
1040                         case XmlReader::TEXT:
1041                         case XmlReader::SIGNIFICANT_WHITESPACE:
1042                                 $buffer .= $this->value;
1043                                 break;
1044                         case XmlReader::END_ELEMENT:
1045                                 return $buffer;
1046                         }
1047                 }
1048                 return $this->close();
1049         }
1050 }
1051
1052 /**
1053  * @todo document (e.g. one-sentence class description).
1054  * @ingroup SpecialPage
1055  */
1056 class WikiRevision {
1057         /** @todo Unused? */
1058         private $importer = null;
1059
1060         /** @var Title */
1061         public $title = null;
1062
1063         /** @var int */
1064         private $id = 0;
1065
1066         /** @var string */
1067         public $timestamp = "20010115000000";
1068
1069         /**
1070          * @var int
1071          * @todo Can't find any uses. Public, because that's suspicious. Get clarity. */
1072         public $user = 0;
1073
1074         /** @var string */
1075         public $user_text = "";
1076
1077         /** @var string */
1078         protected $model = null;
1079
1080         /** @var string */
1081         protected $format = null;
1082
1083         /** @var string */
1084         public $text = "";
1085
1086         /** @var int */
1087         protected $size;
1088
1089         /** @var Content */
1090         protected $content = null;
1091
1092         /** @var string */
1093         public $comment = "";
1094
1095         /** @var bool */
1096         protected $minor = false;
1097
1098         /** @var string */
1099         protected $type = "";
1100
1101         /** @var string */
1102         protected $action = "";
1103
1104         /** @var string */
1105         protected $params = "";
1106
1107         /** @var string */
1108         protected $fileSrc = '';
1109
1110         /** @var bool|string */
1111         protected $sha1base36 = false;
1112
1113         /**
1114          * @var bool
1115          * @todo Unused?
1116          */
1117         private $isTemp = false;
1118
1119         /** @var string */
1120         protected $archiveName = '';
1121
1122         protected $filename;
1123
1124         /** @var mixed */
1125         protected $src;
1126
1127         /** @todo Unused? */
1128         private $fileIsTemp;
1129
1130         /** @var bool */
1131         private $mNoUpdates = false;
1132
1133         /**
1134          * @param Title $title
1135          * @throws MWException
1136          */
1137         function setTitle( $title ) {
1138                 if ( is_object( $title ) ) {
1139                         $this->title = $title;
1140                 } elseif ( is_null( $title ) ) {
1141                         throw new MWException( "WikiRevision given a null title in import. "
1142                                 . "You may need to adjust \$wgLegalTitleChars." );
1143                 } else {
1144                         throw new MWException( "WikiRevision given non-object title in import." );
1145                 }
1146         }
1147
1148         /**
1149          * @param int $id
1150          */
1151         function setID( $id ) {
1152                 $this->id = $id;
1153         }
1154
1155         /**
1156          * @param string $ts
1157          */
1158         function setTimestamp( $ts ) {
1159                 # 2003-08-05T18:30:02Z
1160                 $this->timestamp = wfTimestamp( TS_MW, $ts );
1161         }
1162
1163         /**
1164          * @param string $user
1165          */
1166         function setUsername( $user ) {
1167                 $this->user_text = $user;
1168         }
1169
1170         /**
1171          * @param string $ip
1172          */
1173         function setUserIP( $ip ) {
1174                 $this->user_text = $ip;
1175         }
1176
1177         /**
1178          * @param string $model
1179          */
1180         function setModel( $model ) {
1181                 $this->model = $model;
1182         }
1183
1184         /**
1185          * @param string $format
1186          */
1187         function setFormat( $format ) {
1188                 $this->format = $format;
1189         }
1190
1191         /**
1192          * @param string $text
1193          */
1194         function setText( $text ) {
1195                 $this->text = $text;
1196         }
1197
1198         /**
1199          * @param string $text
1200          */
1201         function setComment( $text ) {
1202                 $this->comment = $text;
1203         }
1204
1205         /**
1206          * @param bool $minor
1207          */
1208         function setMinor( $minor ) {
1209                 $this->minor = (bool)$minor;
1210         }
1211
1212         /**
1213          * @param mixed $src
1214          */
1215         function setSrc( $src ) {
1216                 $this->src = $src;
1217         }
1218
1219         /**
1220          * @param string $src
1221          * @param bool $isTemp
1222          */
1223         function setFileSrc( $src, $isTemp ) {
1224                 $this->fileSrc = $src;
1225                 $this->fileIsTemp = $isTemp;
1226         }
1227
1228         /**
1229          * @param string $sha1base36
1230          */
1231         function setSha1Base36( $sha1base36 ) {
1232                 $this->sha1base36 = $sha1base36;
1233         }
1234
1235         /**
1236          * @param string $filename
1237          */
1238         function setFilename( $filename ) {
1239                 $this->filename = $filename;
1240         }
1241
1242         /**
1243          * @param string $archiveName
1244          */
1245         function setArchiveName( $archiveName ) {
1246                 $this->archiveName = $archiveName;
1247         }
1248
1249         /**
1250          * @param int $size
1251          */
1252         function setSize( $size ) {
1253                 $this->size = intval( $size );
1254         }
1255
1256         /**
1257          * @param string $type
1258          */
1259         function setType( $type ) {
1260                 $this->type = $type;
1261         }
1262
1263         /**
1264          * @param string $action
1265          */
1266         function setAction( $action ) {
1267                 $this->action = $action;
1268         }
1269
1270         /**
1271          * @param array $params
1272          */
1273         function setParams( $params ) {
1274                 $this->params = $params;
1275         }
1276
1277         /**
1278          * @param bool $noupdates
1279          */
1280         public function setNoUpdates( $noupdates ) {
1281                 $this->mNoUpdates = $noupdates;
1282         }
1283
1284         /**
1285          * @return Title
1286          */
1287         function getTitle() {
1288                 return $this->title;
1289         }
1290
1291         /**
1292          * @return int
1293          */
1294         function getID() {
1295                 return $this->id;
1296         }
1297
1298         /**
1299          * @return string
1300          */
1301         function getTimestamp() {
1302                 return $this->timestamp;
1303         }
1304
1305         /**
1306          * @return string
1307          */
1308         function getUser() {
1309                 return $this->user_text;
1310         }
1311
1312         /**
1313          * @return string
1314          *
1315          * @deprecated Since 1.21, use getContent() instead.
1316          */
1317         function getText() {
1318                 ContentHandler::deprecated( __METHOD__, '1.21' );
1319
1320                 return $this->text;
1321         }
1322
1323         /**
1324          * @return Content
1325          */
1326         function getContent() {
1327                 if ( is_null( $this->content ) ) {
1328                         $this->content =
1329                                 ContentHandler::makeContent(
1330                                         $this->text,
1331                                         $this->getTitle(),
1332                                         $this->getModel(),
1333                                         $this->getFormat()
1334                                 );
1335                 }
1336
1337                 return $this->content;
1338         }
1339
1340         /**
1341          * @return string
1342          */
1343         function getModel() {
1344                 if ( is_null( $this->model ) ) {
1345                         $this->model = $this->getTitle()->getContentModel();
1346                 }
1347
1348                 return $this->model;
1349         }
1350
1351         /**
1352          * @return string
1353          */
1354         function getFormat() {
1355                 if ( is_null( $this->model ) ) {
1356                         $this->format = ContentHandler::getForTitle( $this->getTitle() )->getDefaultFormat();
1357                 }
1358
1359                 return $this->format;
1360         }
1361
1362         /**
1363          * @return string
1364          */
1365         function getComment() {
1366                 return $this->comment;
1367         }
1368
1369         /**
1370          * @return bool
1371          */
1372         function getMinor() {
1373                 return $this->minor;
1374         }
1375
1376         /**
1377          * @return mixed
1378          */
1379         function getSrc() {
1380                 return $this->src;
1381         }
1382
1383         /**
1384          * @return bool|string
1385          */
1386         function getSha1() {
1387                 if ( $this->sha1base36 ) {
1388                         return wfBaseConvert( $this->sha1base36, 36, 16 );
1389                 }
1390                 return false;
1391         }
1392
1393         /**
1394          * @return string
1395          */
1396         function getFileSrc() {
1397                 return $this->fileSrc;
1398         }
1399
1400         /**
1401          * @return bool
1402          */
1403         function isTempSrc() {
1404                 return $this->isTemp;
1405         }
1406
1407         /**
1408          * @return mixed
1409          */
1410         function getFilename() {
1411                 return $this->filename;
1412         }
1413
1414         /**
1415          * @return string
1416          */
1417         function getArchiveName() {
1418                 return $this->archiveName;
1419         }
1420
1421         /**
1422          * @return mixed
1423          */
1424         function getSize() {
1425                 return $this->size;
1426         }
1427
1428         /**
1429          * @return string
1430          */
1431         function getType() {
1432                 return $this->type;
1433         }
1434
1435         /**
1436          * @return string
1437          */
1438         function getAction() {
1439                 return $this->action;
1440         }
1441
1442         /**
1443          * @return string
1444          */
1445         function getParams() {
1446                 return $this->params;
1447         }
1448
1449         /**
1450          * @return bool
1451          */
1452         function importOldRevision() {
1453                 $dbw = wfGetDB( DB_MASTER );
1454
1455                 # Sneak a single revision into place
1456                 $user = User::newFromName( $this->getUser() );
1457                 if ( $user ) {
1458                         $userId = intval( $user->getId() );
1459                         $userText = $user->getName();
1460                         $userObj = $user;
1461                 } else {
1462                         $userId = 0;
1463                         $userText = $this->getUser();
1464                         $userObj = new User;
1465                 }
1466
1467                 // avoid memory leak...?
1468                 $linkCache = LinkCache::singleton();
1469                 $linkCache->clear();
1470
1471                 $page = WikiPage::factory( $this->title );
1472                 if ( !$page->exists() ) {
1473                         # must create the page...
1474                         $pageId = $page->insertOn( $dbw );
1475                         $created = true;
1476                         $oldcountable = null;
1477                 } else {
1478                         $pageId = $page->getId();
1479                         $created = false;
1480
1481                         $prior = $dbw->selectField( 'revision', '1',
1482                                 array( 'rev_page' => $pageId,
1483                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1484                                         'rev_user_text' => $userText,
1485                                         'rev_comment' => $this->getComment() ),
1486                                 __METHOD__
1487                         );
1488                         if ( $prior ) {
1489                                 // @todo FIXME: This could fail slightly for multiple matches :P
1490                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1491                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1492                                 return false;
1493                         }
1494                         $oldcountable = $page->isCountable();
1495                 }
1496
1497                 # @todo FIXME: Use original rev_id optionally (better for backups)
1498                 # Insert the row
1499                 $revision = new Revision( array(
1500                         'title' => $this->title,
1501                         'page' => $pageId,
1502                         'content_model' => $this->getModel(),
1503                         'content_format' => $this->getFormat(),
1504                         //XXX: just set 'content' => $this->getContent()?
1505                         'text' => $this->getContent()->serialize( $this->getFormat() ),
1506                         'comment' => $this->getComment(),
1507                         'user' => $userId,
1508                         'user_text' => $userText,
1509                         'timestamp' => $this->timestamp,
1510                         'minor_edit' => $this->minor,
1511                         ) );
1512                 $revision->insertOn( $dbw );
1513                 $changed = $page->updateIfNewerOn( $dbw, $revision );
1514
1515                 if ( $changed !== false && !$this->mNoUpdates ) {
1516                         wfDebug( __METHOD__ . ": running updates\n" );
1517                         $page->doEditUpdates(
1518                                 $revision,
1519                                 $userObj,
1520                                 array( 'created' => $created, 'oldcountable' => $oldcountable )
1521                         );
1522                 }
1523
1524                 return true;
1525         }
1526
1527         /**
1528          * @return mixed
1529          */
1530         function importLogItem() {
1531                 $dbw = wfGetDB( DB_MASTER );
1532                 # @todo FIXME: This will not record autoblocks
1533                 if ( !$this->getTitle() ) {
1534                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1535                                 $this->timestamp . "\n" );
1536                         return;
1537                 }
1538                 # Check if it exists already
1539                 // @todo FIXME: Use original log ID (better for backups)
1540                 $prior = $dbw->selectField( 'logging', '1',
1541                         array( 'log_type' => $this->getType(),
1542                                 'log_action' => $this->getAction(),
1543                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1544                                 'log_namespace' => $this->getTitle()->getNamespace(),
1545                                 'log_title' => $this->getTitle()->getDBkey(),
1546                                 'log_comment' => $this->getComment(),
1547                                 #'log_user_text' => $this->user_text,
1548                                 'log_params' => $this->params ),
1549                         __METHOD__
1550                 );
1551                 // @todo FIXME: This could fail slightly for multiple matches :P
1552                 if ( $prior ) {
1553                         wfDebug( __METHOD__
1554                                 . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp "
1555                                 . $this->timestamp . "\n" );
1556                         return;
1557                 }
1558                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1559                 $data = array(
1560                         'log_id' => $log_id,
1561                         'log_type' => $this->type,
1562                         'log_action' => $this->action,
1563                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1564                         'log_user' => User::idFromName( $this->user_text ),
1565                         #'log_user_text' => $this->user_text,
1566                         'log_namespace' => $this->getTitle()->getNamespace(),
1567                         'log_title' => $this->getTitle()->getDBkey(),
1568                         'log_comment' => $this->getComment(),
1569                         'log_params' => $this->params
1570                 );
1571                 $dbw->insert( 'logging', $data, __METHOD__ );
1572         }
1573
1574         /**
1575          * @return bool
1576          */
1577         function importUpload() {
1578                 # Construct a file
1579                 $archiveName = $this->getArchiveName();
1580                 if ( $archiveName ) {
1581                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1582                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1583                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1584                 } else {
1585                         $file = wfLocalFile( $this->getTitle() );
1586                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1587                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1588                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1589                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1590                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1591                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1592                         }
1593                 }
1594                 if ( !$file ) {
1595                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1596                         return false;
1597                 }
1598
1599                 # Get the file source or download if necessary
1600                 $source = $this->getFileSrc();
1601                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1602                 if ( !$source ) {
1603                         $source = $this->downloadSource();
1604                         $flags |= File::DELETE_SOURCE;
1605                 }
1606                 if ( !$source ) {
1607                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1608                         return false;
1609                 }
1610                 $sha1 = $this->getSha1();
1611                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1612                         if ( $flags & File::DELETE_SOURCE ) {
1613                                 # Broken file; delete it if it is a temporary file
1614                                 unlink( $source );
1615                         }
1616                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1617                         return false;
1618                 }
1619
1620                 $user = User::newFromName( $this->user_text );
1621
1622                 # Do the actual upload
1623                 if ( $archiveName ) {
1624                         $status = $file->uploadOld( $source, $archiveName,
1625                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1626                 } else {
1627                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1628                                 $flags, false, $this->getTimestamp(), $user );
1629                 }
1630
1631                 if ( $status->isGood() ) {
1632                         wfDebug( __METHOD__ . ": Successful\n" );
1633                         return true;
1634                 } else {
1635                         wfDebug( __METHOD__ . ': failed: ' . $status->getXml() . "\n" );
1636                         return false;
1637                 }
1638         }
1639
1640         /**
1641          * @return bool|string
1642          */
1643         function downloadSource() {
1644                 global $wgEnableUploads;
1645                 if ( !$wgEnableUploads ) {
1646                         return false;
1647                 }
1648
1649                 $tempo = tempnam( wfTempDir(), 'download' );
1650                 $f = fopen( $tempo, 'wb' );
1651                 if ( !$f ) {
1652                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1653                         return false;
1654                 }
1655
1656                 // @todo FIXME!
1657                 $src = $this->getSrc();
1658                 $data = Http::get( $src );
1659                 if ( !$data ) {
1660                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1661                         fclose( $f );
1662                         unlink( $tempo );
1663                         return false;
1664                 }
1665
1666                 fwrite( $f, $data );
1667                 fclose( $f );
1668
1669                 return $tempo;
1670         }
1671
1672 }
1673
1674 /**
1675  * @todo document (e.g. one-sentence class description).
1676  * @ingroup SpecialPage
1677  */
1678 class ImportStringSource {
1679         function __construct( $string ) {
1680                 $this->mString = $string;
1681                 $this->mRead = false;
1682         }
1683
1684         /**
1685          * @return bool
1686          */
1687         function atEnd() {
1688                 return $this->mRead;
1689         }
1690
1691         /**
1692          * @return bool|string
1693          */
1694         function readChunk() {
1695                 if ( $this->atEnd() ) {
1696                         return false;
1697                 }
1698                 $this->mRead = true;
1699                 return $this->mString;
1700         }
1701 }
1702
1703 /**
1704  * @todo document (e.g. one-sentence class description).
1705  * @ingroup SpecialPage
1706  */
1707 class ImportStreamSource {
1708         function __construct( $handle ) {
1709                 $this->mHandle = $handle;
1710         }
1711
1712         /**
1713          * @return bool
1714          */
1715         function atEnd() {
1716                 return feof( $this->mHandle );
1717         }
1718
1719         /**
1720          * @return string
1721          */
1722         function readChunk() {
1723                 return fread( $this->mHandle, 32768 );
1724         }
1725
1726         /**
1727          * @param string $filename
1728          * @return Status
1729          */
1730         static function newFromFile( $filename ) {
1731                 wfSuppressWarnings();
1732                 $file = fopen( $filename, 'rt' );
1733                 wfRestoreWarnings();
1734                 if ( !$file ) {
1735                         return Status::newFatal( "importcantopen" );
1736                 }
1737                 return Status::newGood( new ImportStreamSource( $file ) );
1738         }
1739
1740         /**
1741          * @param string $fieldname
1742          * @return Status
1743          */
1744         static function newFromUpload( $fieldname = "xmlimport" ) {
1745                 $upload =& $_FILES[$fieldname];
1746
1747                 if ( $upload === null || !$upload['name'] ) {
1748                         return Status::newFatal( 'importnofile' );
1749                 }
1750                 if ( !empty( $upload['error'] ) ) {
1751                         switch ( $upload['error'] ) {
1752                                 case 1:
1753                                         # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1754                                         return Status::newFatal( 'importuploaderrorsize' );
1755                                 case 2:
1756                                         # The uploaded file exceeds the MAX_FILE_SIZE directive that
1757                                         # was specified in the HTML form.
1758                                         return Status::newFatal( 'importuploaderrorsize' );
1759                                 case 3:
1760                                         # The uploaded file was only partially uploaded
1761                                         return Status::newFatal( 'importuploaderrorpartial' );
1762                                 case 6:
1763                                         # Missing a temporary folder.
1764                                         return Status::newFatal( 'importuploaderrortemp' );
1765                                 # case else: # Currently impossible
1766                         }
1767
1768                 }
1769                 $fname = $upload['tmp_name'];
1770                 if ( is_uploaded_file( $fname ) ) {
1771                         return ImportStreamSource::newFromFile( $fname );
1772                 } else {
1773                         return Status::newFatal( 'importnofile' );
1774                 }
1775         }
1776
1777         /**
1778          * @param string $url
1779          * @param string $method
1780          * @return Status
1781          */
1782         static function newFromURL( $url, $method = 'GET' ) {
1783                 wfDebug( __METHOD__ . ": opening $url\n" );
1784                 # Use the standard HTTP fetch function; it times out
1785                 # quicker and sorts out user-agent problems which might
1786                 # otherwise prevent importing from large sites, such
1787                 # as the Wikimedia cluster, etc.
1788                 $data = Http::request( $method, $url, array( 'followRedirects' => true ) );
1789                 if ( $data !== false ) {
1790                         $file = tmpfile();
1791                         fwrite( $file, $data );
1792                         fflush( $file );
1793                         fseek( $file, 0 );
1794                         return Status::newGood( new ImportStreamSource( $file ) );
1795                 } else {
1796                         return Status::newFatal( 'importcantopen' );
1797                 }
1798         }
1799
1800         /**
1801          * @param string $interwiki
1802          * @param string $page
1803          * @param bool $history
1804          * @param bool $templates
1805          * @param int $pageLinkDepth
1806          * @return Status
1807          */
1808         public static function newFromInterwiki( $interwiki, $page, $history = false,
1809                 $templates = false, $pageLinkDepth = 0
1810         ) {
1811                 if ( $page == '' ) {
1812                         return Status::newFatal( 'import-noarticle' );
1813                 }
1814                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1815                 if ( is_null( $link ) || !$link->isExternal() ) {
1816                         return Status::newFatal( 'importbadinterwiki' );
1817                 } else {
1818                         $params = array();
1819                         if ( $history ) {
1820                                 $params['history'] = 1;
1821                         }
1822                         if ( $templates ) {
1823                                 $params['templates'] = 1;
1824                         }
1825                         if ( $pageLinkDepth ) {
1826                                 $params['pagelink-depth'] = $pageLinkDepth;
1827                         }
1828                         $url = $link->getFullURL( $params );
1829                         # For interwikis, use POST to avoid redirects.
1830                         return ImportStreamSource::newFromURL( $url, "POST" );
1831                 }
1832         }
1833 }