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 ImportStreamSource $source
  44          */
  45         function __construct( ImportStreamSource $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->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          * Retrieves the contents of the named attribute of the current element.
 398          * @param string $attr the name of the attribute
 399          * @return string the value of the attribute or an empty string if it is not set in the current element.
 400          */
 401         public function nodeAttribute( $attr ) {
 402                 return $this->reader->getAttribute( $attr );
 403         }
 404
 405         /**
 406          * Shouldn't something like this be built-in to XMLReader?
 407          * Fetches text contents of the current element, assuming
 408          * no sub-elements or such scary things.
 409          * @return string
 410          * @access private
 411          */
 412         public function nodeContents() {
 413                 if ( $this->reader->isEmptyElement ) {
 414                         return "";
 415                 }
 416                 $buffer = "";
 417                 while ( $this->reader->read() ) {
 418                         switch ( $this->reader->nodeType ) {
 419                         case XmlReader::TEXT:
 420                         case XmlReader::SIGNIFICANT_WHITESPACE:
 421                                 $buffer .= $this->reader->value;
 422                                 break;
 423                         case XmlReader::END_ELEMENT:
 424                                 return $buffer;
 425                         }
 426                 }
 427
 428                 $this->reader->close();
 429                 return '';
 430         }
 431
 432         # --------------
 433
 434         /** Left in for debugging */
 435         private function dumpElement() {
 436                 static $lookup = null;
 437                 if ( !$lookup ) {
 438                         $xmlReaderConstants = array(
 439                                 "NONE",
 440                                 "ELEMENT",
 441                                 "ATTRIBUTE",
 442                                 "TEXT",
 443                                 "CDATA",
 444                                 "ENTITY_REF",
 445                                 "ENTITY",
 446                                 "PI",
 447                                 "COMMENT",
 448                                 "DOC",
 449                                 "DOC_TYPE",
 450                                 "DOC_FRAGMENT",
 451                                 "NOTATION",
 452                                 "WHITESPACE",
 453                                 "SIGNIFICANT_WHITESPACE",
 454                                 "END_ELEMENT",
 455                                 "END_ENTITY",
 456                                 "XML_DECLARATION",
 457                         );
 458                         $lookup = array();
 459
 460                         foreach ( $xmlReaderConstants as $name ) {
 461                                 $lookup[constant( "XmlReader::$name" )] = $name;
 462                         }
 463                 }
 464
 465                 print var_dump(
 466                         $lookup[$this->reader->nodeType],
 467                         $this->reader->name,
 468                         $this->reader->value
 469                 ) . "\n\n";
 470         }
 471
 472         /**
 473          * Primary entry point
 474          * @throws MWException
 475          * @return bool
 476          */
 477         public function doImport() {
 478
 479                 // Calls to reader->read need to be wrapped in calls to
 480                 // libxml_disable_entity_loader() to avoid local file
 481                 // inclusion attacks (bug 46932).
 482                 $oldDisable = libxml_disable_entity_loader( true );
 483                 $this->reader->read();
 484
 485                 if ( $this->reader->name != 'mediawiki' ) {
 486                         libxml_disable_entity_loader( $oldDisable );
 487                         throw new MWException( "Expected <mediawiki> tag, got " .
 488                                 $this->reader->name );
 489                 }
 490                 $this->debug( "<mediawiki> tag is correct." );
 491
 492                 $this->debug( "Starting primary dump processing loop." );
 493
 494                 $keepReading = $this->reader->read();
 495                 $skip = false;
 496                 while ( $keepReading ) {
 497                         $tag = $this->reader->name;
 498                         $type = $this->reader->nodeType;
 499
 500                         if ( !wfRunHooks( 'ImportHandleToplevelXMLTag', array( $this ) ) ) {
 501                                 // Do nothing
 502                         } elseif ( $tag == 'mediawiki' && $type == XmlReader::END_ELEMENT ) {
 503                                 break;
 504                         } elseif ( $tag == 'siteinfo' ) {
 505                                 $this->handleSiteInfo();
 506                         } elseif ( $tag == 'page' ) {
 507                                 $this->handlePage();
 508                         } elseif ( $tag == 'logitem' ) {
 509                                 $this->handleLogItem();
 510                         } elseif ( $tag != '#text' ) {
 511                                 $this->warn( "Unhandled top-level XML tag $tag" );
 512
 513                                 $skip = true;
 514                         }
 515
 516                         if ( $skip ) {
 517                                 $keepReading = $this->reader->next();
 518                                 $skip = false;
 519                                 $this->debug( "Skip" );
 520                         } else {
 521                                 $keepReading = $this->reader->read();
 522                         }
 523                 }
 524
 525                 libxml_disable_entity_loader( $oldDisable );
 526                 return true;
 527         }
 528
 529         /**
 530          * @return bool
 531          * @throws MWException
 532          */
 533         private function handleSiteInfo() {
 534                 // Site info is useful, but not actually used for dump imports.
 535                 // Includes a quick short-circuit to save performance.
 536                 if ( ! $this->mSiteInfoCallback ) {
 537                         $this->reader->next();
 538                         return true;
 539                 }
 540                 throw new MWException( "SiteInfo tag is not yet handled, do not set mSiteInfoCallback" );
 541         }
 542
 543         private function handleLogItem() {
 544                 $this->debug( "Enter log item handler." );
 545                 $logInfo = array();
 546
 547                 // Fields that can just be stuffed in the pageInfo object
 548                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 549                                         'logtitle', 'params' );
 550
 551                 while ( $this->reader->read() ) {
 552                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 553                                         $this->reader->name == 'logitem' ) {
 554                                 break;
 555                         }
 556
 557                         $tag = $this->reader->name;
 558
 559                         if ( !wfRunHooks( 'ImportHandleLogItemXMLTag', array(
 560                                 $this, $logInfo
 561                         ) ) ) {
 562                                 // Do nothing
 563                         } elseif ( in_array( $tag, $normalFields ) ) {
 564                                 $logInfo[$tag] = $this->nodeContents();
 565                         } elseif ( $tag == 'contributor' ) {
 566                                 $logInfo['contributor'] = $this->handleContributor();
 567                         } elseif ( $tag != '#text' ) {
 568                                 $this->warn( "Unhandled log-item XML tag $tag" );
 569                         }
 570                 }
 571
 572                 $this->processLogItem( $logInfo );
 573         }
 574
 575         /**
 576          * @param array $logInfo
 577          * @return bool|mixed
 578          */
 579         private function processLogItem( $logInfo ) {
 580                 $revision = new WikiRevision;
 581
 582                 $revision->setID( $logInfo['id'] );
 583                 $revision->setType( $logInfo['type'] );
 584                 $revision->setAction( $logInfo['action'] );
 585                 $revision->setTimestamp( $logInfo['timestamp'] );
 586                 $revision->setParams( $logInfo['params'] );
 587                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 588                 $revision->setNoUpdates( $this->mNoUpdates );
 589
 590                 if ( isset( $logInfo['comment'] ) ) {
 591                         $revision->setComment( $logInfo['comment'] );
 592                 }
 593
 594                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 595                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 596                 }
 597                 if ( isset( $logInfo['contributor']['username'] ) ) {
 598                         $revision->setUserName( $logInfo['contributor']['username'] );
 599                 }
 600
 601                 return $this->logItemCallback( $revision );
 602         }
 603
 604         private function handlePage() {
 605                 // Handle page data.
 606                 $this->debug( "Enter page handler." );
 607                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 608
 609                 // Fields that can just be stuffed in the pageInfo object
 610                 $normalFields = array( 'title', 'id', 'redirect', 'restrictions' );
 611
 612                 $skip = false;
 613                 $badTitle = false;
 614
 615                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 616                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 617                                         $this->reader->name == 'page' ) {
 618                                 break;
 619                         }
 620
 621                         $tag = $this->reader->name;
 622
 623                         if ( $badTitle ) {
 624                                 // The title is invalid, bail out of this page
 625                                 $skip = true;
 626                         } elseif ( !wfRunHooks( 'ImportHandlePageXMLTag', array( $this,
 627                                                 &$pageInfo ) ) ) {
 628                                 // Do nothing
 629                         } elseif ( in_array( $tag, $normalFields ) ) {
 630                                 // An XML snippet:
 631                                 // <page>
 632                                 //     <id>123</id>
 633                                 //     <title>Page</title>
 634                                 //     <redirect title="NewTitle"/>
 635                                 //     ...
 636                                 // Because the redirect tag is built differently, we need special handling for that case.
 637                                 if ( $tag == 'redirect' ) {
 638                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 639                                 } else {
 640                                         $pageInfo[$tag] = $this->nodeContents();
 641                                         if ( $tag == 'title' ) {
 642                                                 $title = $this->processTitle( $pageInfo['title'] );
 643
 644                                                 if ( !$title ) {
 645                                                         $badTitle = true;
 646                                                         $skip = true;
 647                                                 }
 648
 649                                                 $this->pageCallback( $title );
 650                                                 list( $pageInfo['_title'], $origTitle ) = $title;
 651                                         }
 652                                 }
 653                         } elseif ( $tag == 'revision' ) {
 654                                 $this->handleRevision( $pageInfo );
 655                         } elseif ( $tag == 'upload' ) {
 656                                 $this->handleUpload( $pageInfo );
 657                         } elseif ( $tag != '#text' ) {
 658                                 $this->warn( "Unhandled page XML tag $tag" );
 659                                 $skip = true;
 660                         }
 661                 }
 662
 663                 $this->pageOutCallback( $pageInfo['_title'], $origTitle,
 664                                         $pageInfo['revisionCount'],
 665                                         $pageInfo['successfulRevisionCount'],
 666                                         $pageInfo );
 667         }
 668
 669         /**
 670          * @param array $pageInfo
 671          */
 672         private function handleRevision( &$pageInfo ) {
 673                 $this->debug( "Enter revision handler" );
 674                 $revisionInfo = array();
 675
 676                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' );
 677
 678                 $skip = false;
 679
 680                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 681                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 682                                         $this->reader->name == 'revision' ) {
 683                                 break;
 684                         }
 685
 686                         $tag = $this->reader->name;
 687
 688                         if ( !wfRunHooks( 'ImportHandleRevisionXMLTag', array(
 689                                 $this, $pageInfo, $revisionInfo
 690                         ) ) ) {
 691                                 // Do nothing
 692                         } elseif ( in_array( $tag, $normalFields ) ) {
 693                                 $revisionInfo[$tag] = $this->nodeContents();
 694                         } elseif ( $tag == 'contributor' ) {
 695                                 $revisionInfo['contributor'] = $this->handleContributor();
 696                         } elseif ( $tag != '#text' ) {
 697                                 $this->warn( "Unhandled revision XML tag $tag" );
 698                                 $skip = true;
 699                         }
 700                 }
 701
 702                 $pageInfo['revisionCount']++;
 703                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 704                         $pageInfo['successfulRevisionCount']++;
 705                 }
 706         }
 707
 708         /**
 709          * @param array $pageInfo
 710          * @param array $revisionInfo
 711          * @return bool|mixed
 712          */
 713         private function processRevision( $pageInfo, $revisionInfo ) {
 714                 $revision = new WikiRevision;
 715
 716                 if ( isset( $revisionInfo['id'] ) ) {
 717                         $revision->setID( $revisionInfo['id'] );
 718                 }
 719                 if ( isset( $revisionInfo['model'] ) ) {
 720                         $revision->setModel( $revisionInfo['model'] );
 721                 }
 722                 if ( isset( $revisionInfo['format'] ) ) {
 723                         $revision->setFormat( $revisionInfo['format'] );
 724                 }
 725                 $revision->setTitle( $pageInfo['_title'] );
 726
 727                 if ( isset( $revisionInfo['text'] ) ) {
 728                         $handler = $revision->getContentHandler();
 729                         $text = $handler->importTransform(
 730                                 $revisionInfo['text'],
 731                                 $revision->getFormat() );
 732
 733                         $revision->setText( $text );
 734                 }
 735                 if ( isset( $revisionInfo['timestamp'] ) ) {
 736                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 737                 } else {
 738                         $revision->setTimestamp( wfTimestampNow() );
 739                 }
 740
 741                 if ( isset( $revisionInfo['comment'] ) ) {
 742                         $revision->setComment( $revisionInfo['comment'] );
 743                 }
 744
 745                 if ( isset( $revisionInfo['minor'] ) ) {
 746                         $revision->setMinor( true );
 747                 }
 748                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 749                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 750                 }
 751                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 752                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 753                 }
 754                 $revision->setNoUpdates( $this->mNoUpdates );
 755
 756                 return $this->revisionCallback( $revision );
 757         }
 758
 759         /**
 760          * @param array $pageInfo
 761          * @return mixed
 762          */
 763         private function handleUpload( &$pageInfo ) {
 764                 $this->debug( "Enter upload handler" );
 765                 $uploadInfo = array();
 766
 767                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 768                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 769
 770                 $skip = false;
 771
 772                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 773                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 774                                         $this->reader->name == 'upload' ) {
 775                                 break;
 776                         }
 777
 778                         $tag = $this->reader->name;
 779
 780                         if ( !wfRunHooks( 'ImportHandleUploadXMLTag', array(
 781                                 $this, $pageInfo
 782                         ) ) ) {
 783                                 // Do nothing
 784                         } elseif ( in_array( $tag, $normalFields ) ) {
 785                                 $uploadInfo[$tag] = $this->nodeContents();
 786                         } elseif ( $tag == 'contributor' ) {
 787                                 $uploadInfo['contributor'] = $this->handleContributor();
 788                         } elseif ( $tag == 'contents' ) {
 789                                 $contents = $this->nodeContents();
 790                                 $encoding = $this->reader->getAttribute( 'encoding' );
 791                                 if ( $encoding === 'base64' ) {
 792                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 793                                         $uploadInfo['isTempSrc'] = true;
 794                                 }
 795                         } elseif ( $tag != '#text' ) {
 796                                 $this->warn( "Unhandled upload XML tag $tag" );
 797                                 $skip = true;
 798                         }
 799                 }
 800
 801                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 802                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 803                         if ( file_exists( $path ) ) {
 804                                 $uploadInfo['fileSrc'] = $path;
 805                                 $uploadInfo['isTempSrc'] = false;
 806                         }
 807                 }
 808
 809                 if ( $this->mImportUploads ) {
 810                         return $this->processUpload( $pageInfo, $uploadInfo );
 811                 }
 812         }
 813
 814         /**
 815          * @param string $contents
 816          * @return string
 817          */
 818         private function dumpTemp( $contents ) {
 819                 $filename = tempnam( wfTempDir(), 'importupload' );
 820                 file_put_contents( $filename, $contents );
 821                 return $filename;
 822         }
 823
 824         /**
 825          * @param array $pageInfo
 826          * @param array $uploadInfo
 827          * @return mixed
 828          */
 829         private function processUpload( $pageInfo, $uploadInfo ) {
 830                 $revision = new WikiRevision;
 831                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 832
 833                 $revision->setTitle( $pageInfo['_title'] );
 834                 $revision->setID( $pageInfo['id'] );
 835                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 836                 $revision->setText( $text );
 837                 $revision->setFilename( $uploadInfo['filename'] );
 838                 if ( isset( $uploadInfo['archivename'] ) ) {
 839                         $revision->setArchiveName( $uploadInfo['archivename'] );
 840                 }
 841                 $revision->setSrc( $uploadInfo['src'] );
 842                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 843                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 844                                 !empty( $uploadInfo['isTempSrc'] ) );
 845                 }
 846                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 847                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 848                 }
 849                 $revision->setSize( intval( $uploadInfo['size'] ) );
 850                 $revision->setComment( $uploadInfo['comment'] );
 851
 852                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 853                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 854                 }
 855                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 856                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 857                 }
 858                 $revision->setNoUpdates( $this->mNoUpdates );
 859
 860                 return call_user_func( $this->mUploadCallback, $revision );
 861         }
 862
 863         /**
 864          * @return array
 865          */
 866         private function handleContributor() {
 867                 $fields = array( 'id', 'ip', 'username' );
 868                 $info = array();
 869
 870                 while ( $this->reader->read() ) {
 871                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 872                                         $this->reader->name == 'contributor' ) {
 873                                 break;
 874                         }
 875
 876                         $tag = $this->reader->name;
 877
 878                         if ( in_array( $tag, $fields ) ) {
 879                                 $info[$tag] = $this->nodeContents();
 880                         }
 881                 }
 882
 883                 return $info;
 884         }
 885
 886         /**
 887          * @param string $text
 888          * @return array|bool
 889          */
 890         private function processTitle( $text ) {
 891                 global $wgCommandLineMode;
 892
 893                 $workTitle = $text;
 894                 $origTitle = Title::newFromText( $workTitle );
 895
 896                 if ( !is_null( $this->mTargetNamespace ) && !is_null( $origTitle ) ) {
 897                         # makeTitleSafe, because $origTitle can have a interwiki (different setting of interwiki map)
 898                         # and than dbKey can begin with a lowercase char
 899                         $title = Title::makeTitleSafe( $this->mTargetNamespace,
 900                                 $origTitle->getDBkey() );
 901                 } else {
 902                         if ( !is_null( $this->mTargetRootPage ) ) {
 903                                 $workTitle = $this->mTargetRootPage . '/' . $workTitle;
 904                         }
 905                         $title = Title::newFromText( $workTitle );
 906                 }
 907
 908                 if ( is_null( $title ) ) {
 909                         # Invalid page title? Ignore the page
 910                         $this->notice( 'import-error-invalid', $workTitle );
 911                         return false;
 912                 } elseif ( $title->isExternal() ) {
 913                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
 914                         return false;
 915                 } elseif ( !$title->canExist() ) {
 916                         $this->notice( 'import-error-special', $title->getPrefixedText() );
 917                         return false;
 918                 } elseif ( !$title->userCan( 'edit' ) && !$wgCommandLineMode ) {
 919                         # Do not import if the importing wiki user cannot edit this page
 920                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
 921                         return false;
 922                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$wgCommandLineMode ) {
 923                         # Do not import if the importing wiki user cannot create this page
 924                         $this->notice( 'import-error-create', $title->getPrefixedText() );
 925                         return false;
 926                 }
 927
 928                 return array( $title, $origTitle );
 929         }
 930 }
 931
 932 /** This is a horrible hack used to keep source compatibility */
 933 class UploadSourceAdapter {
 934         /** @var array */
 935         private static $sourceRegistrations = array();
 936
 937         /** @var string */
 938         private $mSource;
 939
 940         /** @var string */
 941         private $mBuffer;
 942
 943         /** @var int */
 944         private $mPosition;
 945
 946         /**
 947          * @param ImportStreamSource $source
 948          * @return string
 949          */
 950         static function registerSource( ImportStreamSource $source ) {
 951                 $id = wfRandomString();
 952
 953                 self::$sourceRegistrations[$id] = $source;
 954
 955                 return $id;
 956         }
 957
 958         /**
 959          * @param string $path
 960          * @param string $mode
 961          * @param array $options
 962          * @param string $opened_path
 963          * @return bool
 964          */
 965         function stream_open( $path, $mode, $options, &$opened_path ) {
 966                 $url = parse_url( $path );
 967                 $id = $url['host'];
 968
 969                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
 970                         return false;
 971                 }
 972
 973                 $this->mSource = self::$sourceRegistrations[$id];
 974
 975                 return true;
 976         }
 977
 978         /**
 979          * @param int $count
 980          * @return string
 981          */
 982         function stream_read( $count ) {
 983                 $return = '';
 984                 $leave = false;
 985
 986                 while ( !$leave && !$this->mSource->atEnd() &&
 987                                 strlen( $this->mBuffer ) < $count ) {
 988                         $read = $this->mSource->readChunk();
 989
 990                         if ( !strlen( $read ) ) {
 991                                 $leave = true;
 992                         }
 993
 994                         $this->mBuffer .= $read;
 995                 }
 996
 997                 if ( strlen( $this->mBuffer ) ) {
 998                         $return = substr( $this->mBuffer, 0, $count );
 999                         $this->mBuffer = substr( $this->mBuffer, $count );
1000                 }
1001
1002                 $this->mPosition += strlen( $return );
1003
1004                 return $return;
1005         }
1006
1007         /**
1008          * @param string $data
1009          * @return bool
1010          */
1011         function stream_write( $data ) {
1012                 return false;
1013         }
1014
1015         /**
1016          * @return mixed
1017          */
1018         function stream_tell() {
1019                 return $this->mPosition;
1020         }
1021
1022         /**
1023          * @return bool
1024          */
1025         function stream_eof() {
1026                 return $this->mSource->atEnd();
1027         }
1028
1029         /**
1030          * @return array
1031          */
1032         function url_stat() {
1033                 $result = array();
1034
1035                 $result['dev'] = $result[0] = 0;
1036                 $result['ino'] = $result[1] = 0;
1037                 $result['mode'] = $result[2] = 0;
1038                 $result['nlink'] = $result[3] = 0;
1039                 $result['uid'] = $result[4] = 0;
1040                 $result['gid'] = $result[5] = 0;
1041                 $result['rdev'] = $result[6] = 0;
1042                 $result['size'] = $result[7] = 0;
1043                 $result['atime'] = $result[8] = 0;
1044                 $result['mtime'] = $result[9] = 0;
1045                 $result['ctime'] = $result[10] = 0;
1046                 $result['blksize'] = $result[11] = 0;
1047                 $result['blocks'] = $result[12] = 0;
1048
1049                 return $result;
1050         }
1051 }
1052
1053 class XMLReader2 extends XMLReader {
1054
1055         /**
1056          * @return bool|string
1057          */
1058         function nodeContents() {
1059                 if ( $this->isEmptyElement ) {
1060                         return "";
1061                 }
1062                 $buffer = "";
1063                 while ( $this->read() ) {
1064                         switch ( $this->nodeType ) {
1065                         case XmlReader::TEXT:
1066                         case XmlReader::SIGNIFICANT_WHITESPACE:
1067                                 $buffer .= $this->value;
1068                                 break;
1069                         case XmlReader::END_ELEMENT:
1070                                 return $buffer;
1071                         }
1072                 }
1073                 return $this->close();
1074         }
1075 }
1076
1077 /**
1078  * @todo document (e.g. one-sentence class description).
1079  * @ingroup SpecialPage
1080  */
1081 class WikiRevision {
1082         /** @todo Unused? */
1083         private $importer = null;
1084
1085         /** @var Title */
1086         public $title = null;
1087
1088         /** @var int */
1089         private $id = 0;
1090
1091         /** @var string */
1092         public $timestamp = "20010115000000";
1093
1094         /**
1095          * @var int
1096          * @todo Can't find any uses. Public, because that's suspicious. Get clarity. */
1097         public $user = 0;
1098
1099         /** @var string */
1100         public $user_text = "";
1101
1102         /** @var string */
1103         protected $model = null;
1104
1105         /** @var string */
1106         protected $format = null;
1107
1108         /** @var string */
1109         public $text = "";
1110
1111         /** @var int */
1112         protected $size;
1113
1114         /** @var Content */
1115         protected $content = null;
1116
1117         /** @var ContentHandler */
1118         protected $contentHandler = null;
1119
1120         /** @var string */
1121         public $comment = "";
1122
1123         /** @var bool */
1124         protected $minor = false;
1125
1126         /** @var string */
1127         protected $type = "";
1128
1129         /** @var string */
1130         protected $action = "";
1131
1132         /** @var string */
1133         protected $params = "";
1134
1135         /** @var string */
1136         protected $fileSrc = '';
1137
1138         /** @var bool|string */
1139         protected $sha1base36 = false;
1140
1141         /**
1142          * @var bool
1143          * @todo Unused?
1144          */
1145         private $isTemp = false;
1146
1147         /** @var string */
1148         protected $archiveName = '';
1149
1150         protected $filename;
1151
1152         /** @var mixed */
1153         protected $src;
1154
1155         /** @todo Unused? */
1156         private $fileIsTemp;
1157
1158         /** @var bool */
1159         private $mNoUpdates = false;
1160
1161         /**
1162          * @param Title $title
1163          * @throws MWException
1164          */
1165         function setTitle( $title ) {
1166                 if ( is_object( $title ) ) {
1167                         $this->title = $title;
1168                 } elseif ( is_null( $title ) ) {
1169                         throw new MWException( "WikiRevision given a null title in import. "
1170                                 . "You may need to adjust \$wgLegalTitleChars." );
1171                 } else {
1172                         throw new MWException( "WikiRevision given non-object title in import." );
1173                 }
1174         }
1175
1176         /**
1177          * @param int $id
1178          */
1179         function setID( $id ) {
1180                 $this->id = $id;
1181         }
1182
1183         /**
1184          * @param string $ts
1185          */
1186         function setTimestamp( $ts ) {
1187                 # 2003-08-05T18:30:02Z
1188                 $this->timestamp = wfTimestamp( TS_MW, $ts );
1189         }
1190
1191         /**
1192          * @param string $user
1193          */
1194         function setUsername( $user ) {
1195                 $this->user_text = $user;
1196         }
1197
1198         /**
1199          * @param string $ip
1200          */
1201         function setUserIP( $ip ) {
1202                 $this->user_text = $ip;
1203         }
1204
1205         /**
1206          * @param string $model
1207          */
1208         function setModel( $model ) {
1209                 $this->model = $model;
1210         }
1211
1212         /**
1213          * @param string $format
1214          */
1215         function setFormat( $format ) {
1216                 $this->format = $format;
1217         }
1218
1219         /**
1220          * @param string $text
1221          */
1222         function setText( $text ) {
1223                 $this->text = $text;
1224         }
1225
1226         /**
1227          * @param string $text
1228          */
1229         function setComment( $text ) {
1230                 $this->comment = $text;
1231         }
1232
1233         /**
1234          * @param bool $minor
1235          */
1236         function setMinor( $minor ) {
1237                 $this->minor = (bool)$minor;
1238         }
1239
1240         /**
1241          * @param mixed $src
1242          */
1243         function setSrc( $src ) {
1244                 $this->src = $src;
1245         }
1246
1247         /**
1248          * @param string $src
1249          * @param bool $isTemp
1250          */
1251         function setFileSrc( $src, $isTemp ) {
1252                 $this->fileSrc = $src;
1253                 $this->fileIsTemp = $isTemp;
1254         }
1255
1256         /**
1257          * @param string $sha1base36
1258          */
1259         function setSha1Base36( $sha1base36 ) {
1260                 $this->sha1base36 = $sha1base36;
1261         }
1262
1263         /**
1264          * @param string $filename
1265          */
1266         function setFilename( $filename ) {
1267                 $this->filename = $filename;
1268         }
1269
1270         /**
1271          * @param string $archiveName
1272          */
1273         function setArchiveName( $archiveName ) {
1274                 $this->archiveName = $archiveName;
1275         }
1276
1277         /**
1278          * @param int $size
1279          */
1280         function setSize( $size ) {
1281                 $this->size = intval( $size );
1282         }
1283
1284         /**
1285          * @param string $type
1286          */
1287         function setType( $type ) {
1288                 $this->type = $type;
1289         }
1290
1291         /**
1292          * @param string $action
1293          */
1294         function setAction( $action ) {
1295                 $this->action = $action;
1296         }
1297
1298         /**
1299          * @param array $params
1300          */
1301         function setParams( $params ) {
1302                 $this->params = $params;
1303         }
1304
1305         /**
1306          * @param bool $noupdates
1307          */
1308         public function setNoUpdates( $noupdates ) {
1309                 $this->mNoUpdates = $noupdates;
1310         }
1311
1312         /**
1313          * @return Title
1314          */
1315         function getTitle() {
1316                 return $this->title;
1317         }
1318
1319         /**
1320          * @return int
1321          */
1322         function getID() {
1323                 return $this->id;
1324         }
1325
1326         /**
1327          * @return string
1328          */
1329         function getTimestamp() {
1330                 return $this->timestamp;
1331         }
1332
1333         /**
1334          * @return string
1335          */
1336         function getUser() {
1337                 return $this->user_text;
1338         }
1339
1340         /**
1341          * @return string
1342          *
1343          * @deprecated Since 1.21, use getContent() instead.
1344          */
1345         function getText() {
1346                 ContentHandler::deprecated( __METHOD__, '1.21' );
1347
1348                 return $this->text;
1349         }
1350
1351         /**
1352          * @return ContentHandler
1353          */
1354         function getContentHandler() {
1355                 if ( is_null( $this->contentHandler ) ) {
1356                         $this->contentHandler = ContentHandler::getForModelID( $this->getModel() );
1357                 }
1358
1359                 return $this->contentHandler;
1360         }
1361
1362         /**
1363          * @return Content
1364          */
1365         function getContent() {
1366                 if ( is_null( $this->content ) ) {
1367                         $handler = $this->getContentHandler();
1368                         $this->content = $handler->unserializeContent( $this->text, $this->getFormat() );
1369                 }
1370
1371                 return $this->content;
1372         }
1373
1374         /**
1375          * @return string
1376          */
1377         function getModel() {
1378                 if ( is_null( $this->model ) ) {
1379                         $this->model = $this->getTitle()->getContentModel();
1380                 }
1381
1382                 return $this->model;
1383         }
1384
1385         /**
1386          * @return string
1387          */
1388         function getFormat() {
1389                 if ( is_null( $this->format ) ) {
1390                         $this->format = $this->getContentHandler()->getDefaultFormat();
1391                 }
1392
1393                 return $this->format;
1394         }
1395
1396         /**
1397          * @return string
1398          */
1399         function getComment() {
1400                 return $this->comment;
1401         }
1402
1403         /**
1404          * @return bool
1405          */
1406         function getMinor() {
1407                 return $this->minor;
1408         }
1409
1410         /**
1411          * @return mixed
1412          */
1413         function getSrc() {
1414                 return $this->src;
1415         }
1416
1417         /**
1418          * @return bool|string
1419          */
1420         function getSha1() {
1421                 if ( $this->sha1base36 ) {
1422                         return wfBaseConvert( $this->sha1base36, 36, 16 );
1423                 }
1424                 return false;
1425         }
1426
1427         /**
1428          * @return string
1429          */
1430         function getFileSrc() {
1431                 return $this->fileSrc;
1432         }
1433
1434         /**
1435          * @return bool
1436          */
1437         function isTempSrc() {
1438                 return $this->isTemp;
1439         }
1440
1441         /**
1442          * @return mixed
1443          */
1444         function getFilename() {
1445                 return $this->filename;
1446         }
1447
1448         /**
1449          * @return string
1450          */
1451         function getArchiveName() {
1452                 return $this->archiveName;
1453         }
1454
1455         /**
1456          * @return mixed
1457          */
1458         function getSize() {
1459                 return $this->size;
1460         }
1461
1462         /**
1463          * @return string
1464          */
1465         function getType() {
1466                 return $this->type;
1467         }
1468
1469         /**
1470          * @return string
1471          */
1472         function getAction() {
1473                 return $this->action;
1474         }
1475
1476         /**
1477          * @return string
1478          */
1479         function getParams() {
1480                 return $this->params;
1481         }
1482
1483         /**
1484          * @return bool
1485          */
1486         function importOldRevision() {
1487                 $dbw = wfGetDB( DB_MASTER );
1488
1489                 # Sneak a single revision into place
1490                 $user = User::newFromName( $this->getUser() );
1491                 if ( $user ) {
1492                         $userId = intval( $user->getId() );
1493                         $userText = $user->getName();
1494                         $userObj = $user;
1495                 } else {
1496                         $userId = 0;
1497                         $userText = $this->getUser();
1498                         $userObj = new User;
1499                 }
1500
1501                 // avoid memory leak...?
1502                 $linkCache = LinkCache::singleton();
1503                 $linkCache->clear();
1504
1505                 $page = WikiPage::factory( $this->title );
1506                 if ( !$page->exists() ) {
1507                         # must create the page...
1508                         $pageId = $page->insertOn( $dbw );
1509                         $created = true;
1510                         $oldcountable = null;
1511                 } else {
1512                         $pageId = $page->getId();
1513                         $created = false;
1514
1515                         $prior = $dbw->selectField( 'revision', '1',
1516                                 array( 'rev_page' => $pageId,
1517                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1518                                         'rev_user_text' => $userText,
1519                                         'rev_comment' => $this->getComment() ),
1520                                 __METHOD__
1521                         );
1522                         if ( $prior ) {
1523                                 // @todo FIXME: This could fail slightly for multiple matches :P
1524                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1525                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1526                                 return false;
1527                         }
1528                         $oldcountable = $page->isCountable();
1529                 }
1530
1531                 # @todo FIXME: Use original rev_id optionally (better for backups)
1532                 # Insert the row
1533                 $revision = new Revision( array(
1534                         'title' => $this->title,
1535                         'page' => $pageId,
1536                         'content_model' => $this->getModel(),
1537                         'content_format' => $this->getFormat(),
1538                         //XXX: just set 'content' => $this->getContent()?
1539                         'text' => $this->getContent()->serialize( $this->getFormat() ),
1540                         'comment' => $this->getComment(),
1541                         'user' => $userId,
1542                         'user_text' => $userText,
1543                         'timestamp' => $this->timestamp,
1544                         'minor_edit' => $this->minor,
1545                         ) );
1546                 $revision->insertOn( $dbw );
1547                 $changed = $page->updateIfNewerOn( $dbw, $revision );
1548
1549                 if ( $changed !== false && !$this->mNoUpdates ) {
1550                         wfDebug( __METHOD__ . ": running updates\n" );
1551                         $page->doEditUpdates(
1552                                 $revision,
1553                                 $userObj,
1554                                 array( 'created' => $created, 'oldcountable' => $oldcountable )
1555                         );
1556                 }
1557
1558                 return true;
1559         }
1560
1561         /**
1562          * @return mixed
1563          */
1564         function importLogItem() {
1565                 $dbw = wfGetDB( DB_MASTER );
1566                 # @todo FIXME: This will not record autoblocks
1567                 if ( !$this->getTitle() ) {
1568                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1569                                 $this->timestamp . "\n" );
1570                         return;
1571                 }
1572                 # Check if it exists already
1573                 // @todo FIXME: Use original log ID (better for backups)
1574                 $prior = $dbw->selectField( 'logging', '1',
1575                         array( 'log_type' => $this->getType(),
1576                                 'log_action' => $this->getAction(),
1577                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1578                                 'log_namespace' => $this->getTitle()->getNamespace(),
1579                                 'log_title' => $this->getTitle()->getDBkey(),
1580                                 'log_comment' => $this->getComment(),
1581                                 #'log_user_text' => $this->user_text,
1582                                 'log_params' => $this->params ),
1583                         __METHOD__
1584                 );
1585                 // @todo FIXME: This could fail slightly for multiple matches :P
1586                 if ( $prior ) {
1587                         wfDebug( __METHOD__
1588                                 . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp "
1589                                 . $this->timestamp . "\n" );
1590                         return;
1591                 }
1592                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1593                 $data = array(
1594                         'log_id' => $log_id,
1595                         'log_type' => $this->type,
1596                         'log_action' => $this->action,
1597                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1598                         'log_user' => User::idFromName( $this->user_text ),
1599                         #'log_user_text' => $this->user_text,
1600                         'log_namespace' => $this->getTitle()->getNamespace(),
1601                         'log_title' => $this->getTitle()->getDBkey(),
1602                         'log_comment' => $this->getComment(),
1603                         'log_params' => $this->params
1604                 );
1605                 $dbw->insert( 'logging', $data, __METHOD__ );
1606         }
1607
1608         /**
1609          * @return bool
1610          */
1611         function importUpload() {
1612                 # Construct a file
1613                 $archiveName = $this->getArchiveName();
1614                 if ( $archiveName ) {
1615                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1616                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1617                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1618                 } else {
1619                         $file = wfLocalFile( $this->getTitle() );
1620                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1621                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1622                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1623                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1624                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1625                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1626                         }
1627                 }
1628                 if ( !$file ) {
1629                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1630                         return false;
1631                 }
1632
1633                 # Get the file source or download if necessary
1634                 $source = $this->getFileSrc();
1635                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1636                 if ( !$source ) {
1637                         $source = $this->downloadSource();
1638                         $flags |= File::DELETE_SOURCE;
1639                 }
1640                 if ( !$source ) {
1641                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1642                         return false;
1643                 }
1644                 $sha1 = $this->getSha1();
1645                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1646                         if ( $flags & File::DELETE_SOURCE ) {
1647                                 # Broken file; delete it if it is a temporary file
1648                                 unlink( $source );
1649                         }
1650                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1651                         return false;
1652                 }
1653
1654                 $user = User::newFromName( $this->user_text );
1655
1656                 # Do the actual upload
1657                 if ( $archiveName ) {
1658                         $status = $file->uploadOld( $source, $archiveName,
1659                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1660                 } else {
1661                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1662                                 $flags, false, $this->getTimestamp(), $user );
1663                 }
1664
1665                 if ( $status->isGood() ) {
1666                         wfDebug( __METHOD__ . ": Successful\n" );
1667                         return true;
1668                 } else {
1669                         wfDebug( __METHOD__ . ': failed: ' . $status->getXml() . "\n" );
1670                         return false;
1671                 }
1672         }
1673
1674         /**
1675          * @return bool|string
1676          */
1677         function downloadSource() {
1678                 global $wgEnableUploads;
1679                 if ( !$wgEnableUploads ) {
1680                         return false;
1681                 }
1682
1683                 $tempo = tempnam( wfTempDir(), 'download' );
1684                 $f = fopen( $tempo, 'wb' );
1685                 if ( !$f ) {
1686                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1687                         return false;
1688                 }
1689
1690                 // @todo FIXME!
1691                 $src = $this->getSrc();
1692                 $data = Http::get( $src );
1693                 if ( !$data ) {
1694                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1695                         fclose( $f );
1696                         unlink( $tempo );
1697                         return false;
1698                 }
1699
1700                 fwrite( $f, $data );
1701                 fclose( $f );
1702
1703                 return $tempo;
1704         }
1705
1706 }
1707
1708 /**
1709  * @todo document (e.g. one-sentence class description).
1710  * @ingroup SpecialPage
1711  */
1712 class ImportStringSource {
1713         function __construct( $string ) {
1714                 $this->mString = $string;
1715                 $this->mRead = false;
1716         }
1717
1718         /**
1719          * @return bool
1720          */
1721         function atEnd() {
1722                 return $this->mRead;
1723         }
1724
1725         /**
1726          * @return bool|string
1727          */
1728         function readChunk() {
1729                 if ( $this->atEnd() ) {
1730                         return false;
1731                 }
1732                 $this->mRead = true;
1733                 return $this->mString;
1734         }
1735 }
1736
1737 /**
1738  * @todo document (e.g. one-sentence class description).
1739  * @ingroup SpecialPage
1740  */
1741 class ImportStreamSource {
1742         function __construct( $handle ) {
1743                 $this->mHandle = $handle;
1744         }
1745
1746         /**
1747          * @return bool
1748          */
1749         function atEnd() {
1750                 return feof( $this->mHandle );
1751         }
1752
1753         /**
1754          * @return string
1755          */
1756         function readChunk() {
1757                 return fread( $this->mHandle, 32768 );
1758         }
1759
1760         /**
1761          * @param string $filename
1762          * @return Status
1763          */
1764         static function newFromFile( $filename ) {
1765                 wfSuppressWarnings();
1766                 $file = fopen( $filename, 'rt' );
1767                 wfRestoreWarnings();
1768                 if ( !$file ) {
1769                         return Status::newFatal( "importcantopen" );
1770                 }
1771                 return Status::newGood( new ImportStreamSource( $file ) );
1772         }
1773
1774         /**
1775          * @param string $fieldname
1776          * @return Status
1777          */
1778         static function newFromUpload( $fieldname = "xmlimport" ) {
1779                 $upload =& $_FILES[$fieldname];
1780
1781                 if ( $upload === null || !$upload['name'] ) {
1782                         return Status::newFatal( 'importnofile' );
1783                 }
1784                 if ( !empty( $upload['error'] ) ) {
1785                         switch ( $upload['error'] ) {
1786                                 case 1:
1787                                         # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1788                                         return Status::newFatal( 'importuploaderrorsize' );
1789                                 case 2:
1790                                         # The uploaded file exceeds the MAX_FILE_SIZE directive that
1791                                         # was specified in the HTML form.
1792                                         return Status::newFatal( 'importuploaderrorsize' );
1793                                 case 3:
1794                                         # The uploaded file was only partially uploaded
1795                                         return Status::newFatal( 'importuploaderrorpartial' );
1796                                 case 6:
1797                                         # Missing a temporary folder.
1798                                         return Status::newFatal( 'importuploaderrortemp' );
1799                                 # case else: # Currently impossible
1800                         }
1801
1802                 }
1803                 $fname = $upload['tmp_name'];
1804                 if ( is_uploaded_file( $fname ) ) {
1805                         return ImportStreamSource::newFromFile( $fname );
1806                 } else {
1807                         return Status::newFatal( 'importnofile' );
1808                 }
1809         }
1810
1811         /**
1812          * @param string $url
1813          * @param string $method
1814          * @return Status
1815          */
1816         static function newFromURL( $url, $method = 'GET' ) {
1817                 wfDebug( __METHOD__ . ": opening $url\n" );
1818                 # Use the standard HTTP fetch function; it times out
1819                 # quicker and sorts out user-agent problems which might
1820                 # otherwise prevent importing from large sites, such
1821                 # as the Wikimedia cluster, etc.
1822                 $data = Http::request( $method, $url, array( 'followRedirects' => true ) );
1823                 if ( $data !== false ) {
1824                         $file = tmpfile();
1825                         fwrite( $file, $data );
1826                         fflush( $file );
1827                         fseek( $file, 0 );
1828                         return Status::newGood( new ImportStreamSource( $file ) );
1829                 } else {
1830                         return Status::newFatal( 'importcantopen' );
1831                 }
1832         }
1833
1834         /**
1835          * @param string $interwiki
1836          * @param string $page
1837          * @param bool $history
1838          * @param bool $templates
1839          * @param int $pageLinkDepth
1840          * @return Status
1841          */
1842         public static function newFromInterwiki( $interwiki, $page, $history = false,
1843                 $templates = false, $pageLinkDepth = 0
1844         ) {
1845                 if ( $page == '' ) {
1846                         return Status::newFatal( 'import-noarticle' );
1847                 }
1848                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1849                 if ( is_null( $link ) || !$link->isExternal() ) {
1850                         return Status::newFatal( 'importbadinterwiki' );
1851                 } else {
1852                         $params = array();
1853                         if ( $history ) {
1854                                 $params['history'] = 1;
1855                         }
1856                         if ( $templates ) {
1857                                 $params['templates'] = 1;
1858                         }
1859                         if ( $pageLinkDepth ) {
1860                                 $params['pagelink-depth'] = $pageLinkDepth;
1861                         }
1862                         $url = $link->getFullURL( $params );
1863                         # For interwikis, use POST to avoid redirects.
1864                         return ImportStreamSource::newFromURL( $url, "POST" );
1865                 }
1866         }
1867 }