includes/Import.php

   1 <?php
   2 /**
   3  * MediaWiki page data importer.
   4  *
   5  * Copyright © 2003,2005 Brion Vibber <brion@pobox.com>
   6  * https://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup SpecialPage
  25  */
  26
  27 /**
  28  * XML file reader for the page data importer
  29  *
  30  * implements Special:Import
  31  * @ingroup SpecialPage
  32  */
  33 class WikiImporter {
  34         private $reader = null;
  35         private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
  36         private $mSiteInfoCallback, $mTargetNamespace, $mTargetRootPage, $mPageOutCallback;
  37         private $mNoticeCallback, $mDebug;
  38         private $mImportUploads, $mImageBasePath;
  39         private $mNoUpdates = false;
  40
  41         /**
  42          * Creates an ImportXMLReader drawing from the source provided
  43          * @param string $source
  44          */
  45         function __construct( $source ) {
  46                 $this->reader = new XMLReader();
  47
  48                 stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  49                 $id = UploadSourceAdapter::registerSource( $source );
  50                 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
  51                         $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  52                 } else {
  53                         $this->reader->open( "uploadsource://$id" );
  54                 }
  55
  56                 // Default callbacks
  57                 $this->setRevisionCallback( array( $this, "importRevision" ) );
  58                 $this->setUploadCallback( array( $this, 'importUpload' ) );
  59                 $this->setLogItemCallback( array( $this, 'importLogItem' ) );
  60                 $this->setPageOutCallback( array( $this, 'finishImportPage' ) );
  61         }
  62
  63         private function throwXmlError( $err ) {
  64                 $this->debug( "FAILURE: $err" );
  65                 wfDebug( "WikiImporter XML error: $err\n" );
  66         }
  67
  68         private function debug( $data ) {
  69                 if ( $this->mDebug ) {
  70                         wfDebug( "IMPORT: $data\n" );
  71                 }
  72         }
  73
  74         private function warn( $data ) {
  75                 wfDebug( "IMPORT: $data\n" );
  76         }
  77
  78         private function notice( $msg /*, $param, ...*/ ) {
  79                 $params = func_get_args();
  80                 array_shift( $params );
  81
  82                 if ( is_callable( $this->mNoticeCallback ) ) {
  83                         call_user_func( $this->mNoticeCallback, $msg, $params );
  84                 } else { # No ImportReporter -> CLI
  85                         echo wfMessage( $msg, $params )->text() . "\n";
  86                 }
  87         }
  88
  89         /**
  90          * Set debug mode...
  91          * @param bool $debug
  92          */
  93         function setDebug( $debug ) {
  94                 $this->mDebug = $debug;
  95         }
  96
  97         /**
  98          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
  99          * @param bool $noupdates
 100          */
 101         function setNoUpdates( $noupdates ) {
 102                 $this->mNoUpdates = $noupdates;
 103         }
 104
 105         /**
 106          * Set a callback that displays notice messages
 107          *
 108          * @param callable $callback
 109          * @return callable
 110          */
 111         public function setNoticeCallback( $callback ) {
 112                 return wfSetVar( $this->mNoticeCallback, $callback );
 113         }
 114
 115         /**
 116          * Sets the action to perform as each new page in the stream is reached.
 117          * @param callable $callback
 118          * @return callable
 119          */
 120         public function setPageCallback( $callback ) {
 121                 $previous = $this->mPageCallback;
 122                 $this->mPageCallback = $callback;
 123                 return $previous;
 124         }
 125
 126         /**
 127          * Sets the action to perform as each page in the stream is completed.
 128          * Callback accepts the page title (as a Title object), a second object
 129          * with the original title form (in case it's been overridden into a
 130          * local namespace), and a count of revisions.
 131          *
 132          * @param callable $callback
 133          * @return callable
 134          */
 135         public function setPageOutCallback( $callback ) {
 136                 $previous = $this->mPageOutCallback;
 137                 $this->mPageOutCallback = $callback;
 138                 return $previous;
 139         }
 140
 141         /**
 142          * Sets the action to perform as each page revision is reached.
 143          * @param callable $callback
 144          * @return callable
 145          */
 146         public function setRevisionCallback( $callback ) {
 147                 $previous = $this->mRevisionCallback;
 148                 $this->mRevisionCallback = $callback;
 149                 return $previous;
 150         }
 151
 152         /**
 153          * Sets the action to perform as each file upload version is reached.
 154          * @param callable $callback
 155          * @return callable
 156          */
 157         public function setUploadCallback( $callback ) {
 158                 $previous = $this->mUploadCallback;
 159                 $this->mUploadCallback = $callback;
 160                 return $previous;
 161         }
 162
 163         /**
 164          * Sets the action to perform as each log item reached.
 165          * @param callable $callback
 166          * @return callable
 167          */
 168         public function setLogItemCallback( $callback ) {
 169                 $previous = $this->mLogItemCallback;
 170                 $this->mLogItemCallback = $callback;
 171                 return $previous;
 172         }
 173
 174         /**
 175          * Sets the action to perform when site info is encountered
 176          * @param callable $callback
 177          * @return callable
 178          */
 179         public function setSiteInfoCallback( $callback ) {
 180                 $previous = $this->mSiteInfoCallback;
 181                 $this->mSiteInfoCallback = $callback;
 182                 return $previous;
 183         }
 184
 185         /**
 186          * Set a target namespace to override the defaults
 187          * @param null|int $namespace
 188          * @return bool
 189          */
 190         public function setTargetNamespace( $namespace ) {
 191                 if ( is_null( $namespace ) ) {
 192                         // Don't override namespaces
 193                         $this->mTargetNamespace = null;
 194                 } elseif ( $namespace >= 0 ) {
 195                         // @todo FIXME: Check for validity
 196                         $this->mTargetNamespace = intval( $namespace );
 197                 } else {
 198                         return false;
 199                 }
 200         }
 201
 202         /**
 203          * Set a target root page under which all pages are imported
 204          * @param null|string $rootpage
 205          * @return Status
 206          */
 207         public function setTargetRootPage( $rootpage ) {
 208                 $status = Status::newGood();
 209                 if ( is_null( $rootpage ) ) {
 210                         // No rootpage
 211                         $this->mTargetRootPage = null;
 212                 } elseif ( $rootpage !== '' ) {
 213                         $rootpage = rtrim( $rootpage, '/' ); //avoid double slashes
 214                         $title = Title::newFromText( $rootpage, !is_null( $this->mTargetNamespace ) ? $this->mTargetNamespace : NS_MAIN );
 215                         if ( !$title || $title->isExternal() ) {
 216                                 $status->fatal( 'import-rootpage-invalid' );
 217                         } else {
 218                                 if ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 219                                         global $wgContLang;
 220
 221                                         $displayNSText = $title->getNamespace() == NS_MAIN
 222                                                 ? wfMessage( 'blanknamespace' )->text()
 223                                                 : $wgContLang->getNsText( $title->getNamespace() );
 224                                         $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 225                                 } else {
 226                                         // set namespace to 'all', so the namespace check in processTitle() can passed
 227                                         $this->setTargetNamespace( null );
 228                                         $this->mTargetRootPage = $title->getPrefixedDBkey();
 229                                 }
 230                         }
 231                 }
 232                 return $status;
 233         }
 234
 235         /**
 236          * @param string $dir
 237          */
 238         public function setImageBasePath( $dir ) {
 239                 $this->mImageBasePath = $dir;
 240         }
 241
 242         /**
 243          * @param bool $import
 244          */
 245         public function setImportUploads( $import ) {
 246                 $this->mImportUploads = $import;
 247         }
 248
 249         /**
 250          * Default per-revision callback, performs the import.
 251          * @param WikiRevision $revision
 252          * @return bool
 253          */
 254         public function importRevision( $revision ) {
 255                 if ( !$revision->getContent()->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
 256                         $this->notice( 'import-error-bad-location',
 257                                 $revision->getTitle()->getPrefixedText(),
 258                                 $revision->getID(),
 259                                 $revision->getModel(),
 260                                 $revision->getFormat() );
 261
 262                         return false;
 263                 }
 264
 265                 try {
 266                         $dbw = wfGetDB( DB_MASTER );
 267                         return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
 268                 } catch ( MWContentSerializationException $ex ) {
 269                         $this->notice( 'import-error-unserialize',
 270                                 $revision->getTitle()->getPrefixedText(),
 271                                 $revision->getID(),
 272                                 $revision->getModel(),
 273                                 $revision->getFormat() );
 274                 }
 275
 276                 return false;
 277         }
 278
 279         /**
 280          * Default per-revision callback, performs the import.
 281          * @param WikiRevision $rev
 282          * @return bool
 283          */
 284         public function importLogItem( $rev ) {
 285                 $dbw = wfGetDB( DB_MASTER );
 286                 return $dbw->deadlockLoop( array( $rev, 'importLogItem' ) );
 287         }
 288
 289         /**
 290          * Dummy for now...
 291          * @param WikiRevision $revision
 292          * @return bool
 293          */
 294         public function importUpload( $revision ) {
 295                 $dbw = wfGetDB( DB_MASTER );
 296                 return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
 297         }
 298
 299         /**
 300          * Mostly for hook use
 301          * @param Title $title
 302          * @param string $origTitle
 303          * @param int $revCount
 304          * @param int $sRevCount
 305          * @param array $pageInfo
 306          * @return
 307          */
 308         public function finishImportPage( $title, $origTitle, $revCount, $sRevCount, $pageInfo ) {
 309                 $args = func_get_args();
 310                 return wfRunHooks( 'AfterImportPage', $args );
 311         }
 312
 313         /**
 314          * Alternate per-revision callback, for debugging.
 315          * @param WikiRevision $revision
 316          */
 317         public function debugRevisionHandler( &$revision ) {
 318                 $this->debug( "Got revision:" );
 319                 if ( is_object( $revision->title ) ) {
 320                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 321                 } else {
 322                         $this->debug( "-- Title: <invalid>" );
 323                 }
 324                 $this->debug( "-- User: " . $revision->user_text );
 325                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 326                 $this->debug( "-- Comment: " . $revision->comment );
 327                 $this->debug( "-- Text: " . $revision->text );
 328         }
 329
 330         /**
 331          * Notify the callback function when a new "<page>" is reached.
 332          * @param Title $title
 333          */
 334         function pageCallback( $title ) {
 335                 if ( isset( $this->mPageCallback ) ) {
 336                         call_user_func( $this->mPageCallback, $title );
 337                 }
 338         }
 339
 340         /**
 341          * Notify the callback function when a "</page>" is closed.
 342          * @param Title $title
 343          * @param Title $origTitle
 344          * @param int $revCount
 345          * @param int $sucCount Number of revisions for which callback returned true
 346          * @param array $pageInfo Associative array of page information
 347          */
 348         private function pageOutCallback( $title, $origTitle, $revCount, $sucCount, $pageInfo ) {
 349                 if ( isset( $this->mPageOutCallback ) ) {
 350                         $args = func_get_args();
 351                         call_user_func_array( $this->mPageOutCallback, $args );
 352                 }
 353         }
 354
 355         /**
 356          * Notify the callback function of a revision
 357          * @param WikiRevision $revision
 358          * @return bool|mixed
 359          */
 360         private function revisionCallback( $revision ) {
 361                 if ( isset( $this->mRevisionCallback ) ) {
 362                         return call_user_func_array( $this->mRevisionCallback,
 363                                         array( $revision, $this ) );
 364                 } else {
 365                         return false;
 366                 }
 367         }
 368
 369         /**
 370          * Notify the callback function of a new log item
 371          * @param WikiRevision $revision
 372          * @return bool|mixed
 373          */
 374         private function logItemCallback( $revision ) {
 375                 if ( isset( $this->mLogItemCallback ) ) {
 376                         return call_user_func_array( $this->mLogItemCallback,
 377                                         array( $revision, $this ) );
 378                 } else {
 379                         return false;
 380                 }
 381         }
 382
 383         /**
 384          * Shouldn't something like this be built-in to XMLReader?
 385          * Fetches text contents of the current element, assuming
 386          * no sub-elements or such scary things.
 387          * @return string
 388          * @access private
 389          */
 390         private function nodeContents() {
 391                 if ( $this->reader->isEmptyElement ) {
 392                         return "";
 393                 }
 394                 $buffer = "";
 395                 while ( $this->reader->read() ) {
 396                         switch ( $this->reader->nodeType ) {
 397                         case XmlReader::TEXT:
 398                         case XmlReader::SIGNIFICANT_WHITESPACE:
 399                                 $buffer .= $this->reader->value;
 400                                 break;
 401                         case XmlReader::END_ELEMENT:
 402                                 return $buffer;
 403                         }
 404                 }
 405
 406                 $this->reader->close();
 407                 return '';
 408         }
 409
 410         # --------------
 411
 412         /** Left in for debugging */
 413         private function dumpElement() {
 414                 static $lookup = null;
 415                 if ( !$lookup ) {
 416                         $xmlReaderConstants = array(
 417                                 "NONE",
 418                                 "ELEMENT",
 419                                 "ATTRIBUTE",
 420                                 "TEXT",
 421                                 "CDATA",
 422                                 "ENTITY_REF",
 423                                 "ENTITY",
 424                                 "PI",
 425                                 "COMMENT",
 426                                 "DOC",
 427                                 "DOC_TYPE",
 428                                 "DOC_FRAGMENT",
 429                                 "NOTATION",
 430                                 "WHITESPACE",
 431                                 "SIGNIFICANT_WHITESPACE",
 432                                 "END_ELEMENT",
 433                                 "END_ENTITY",
 434                                 "XML_DECLARATION",
 435                         );
 436                         $lookup = array();
 437
 438                         foreach ( $xmlReaderConstants as $name ) {
 439                                 $lookup[constant( "XmlReader::$name" )] = $name;
 440                         }
 441                 }
 442
 443                 print var_dump(
 444                         $lookup[$this->reader->nodeType],
 445                         $this->reader->name,
 446                         $this->reader->value
 447                 ) . "\n\n";
 448         }
 449
 450         /**
 451          * Primary entry point
 452          * @throws MWException
 453          * @return bool
 454          */
 455         public function doImport() {
 456
 457                 // Calls to reader->read need to be wrapped in calls to
 458                 // libxml_disable_entity_loader() to avoid local file
 459                 // inclusion attacks (bug 46932).
 460                 $oldDisable = libxml_disable_entity_loader( true );
 461                 $this->reader->read();
 462
 463                 if ( $this->reader->name != 'mediawiki' ) {
 464                         libxml_disable_entity_loader( $oldDisable );
 465                         throw new MWException( "Expected <mediawiki> tag, got " .
 466                                 $this->reader->name );
 467                 }
 468                 $this->debug( "<mediawiki> tag is correct." );
 469
 470                 $this->debug( "Starting primary dump processing loop." );
 471
 472                 $keepReading = $this->reader->read();
 473                 $skip = false;
 474                 while ( $keepReading ) {
 475                         $tag = $this->reader->name;
 476                         $type = $this->reader->nodeType;
 477
 478                         if ( !wfRunHooks( 'ImportHandleToplevelXMLTag', array( $this ) ) ) {
 479                                 // Do nothing
 480                         } elseif ( $tag == 'mediawiki' && $type == XmlReader::END_ELEMENT ) {
 481                                 break;
 482                         } elseif ( $tag == 'siteinfo' ) {
 483                                 $this->handleSiteInfo();
 484                         } elseif ( $tag == 'page' ) {
 485                                 $this->handlePage();
 486                         } elseif ( $tag == 'logitem' ) {
 487                                 $this->handleLogItem();
 488                         } elseif ( $tag != '#text' ) {
 489                                 $this->warn( "Unhandled top-level XML tag $tag" );
 490
 491                                 $skip = true;
 492                         }
 493
 494                         if ( $skip ) {
 495                                 $keepReading = $this->reader->next();
 496                                 $skip = false;
 497                                 $this->debug( "Skip" );
 498                         } else {
 499                                 $keepReading = $this->reader->read();
 500                         }
 501                 }
 502
 503                 libxml_disable_entity_loader( $oldDisable );
 504                 return true;
 505         }
 506
 507         /**
 508          * @return bool
 509          * @throws MWException
 510          */
 511         private function handleSiteInfo() {
 512                 // Site info is useful, but not actually used for dump imports.
 513                 // Includes a quick short-circuit to save performance.
 514                 if ( ! $this->mSiteInfoCallback ) {
 515                         $this->reader->next();
 516                         return true;
 517                 }
 518                 throw new MWException( "SiteInfo tag is not yet handled, do not set mSiteInfoCallback" );
 519         }
 520
 521         private function handleLogItem() {
 522                 $this->debug( "Enter log item handler." );
 523                 $logInfo = array();
 524
 525                 // Fields that can just be stuffed in the pageInfo object
 526                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 527                                         'logtitle', 'params' );
 528
 529                 while ( $this->reader->read() ) {
 530                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 531                                         $this->reader->name == 'logitem' ) {
 532                                 break;
 533                         }
 534
 535                         $tag = $this->reader->name;
 536
 537                         if ( !wfRunHooks( 'ImportHandleLogItemXMLTag', array(
 538                                 $this, $logInfo
 539                         ) ) ) {
 540                                 // Do nothing
 541                         } elseif ( in_array( $tag, $normalFields ) ) {
 542                                 $logInfo[$tag] = $this->nodeContents();
 543                         } elseif ( $tag == 'contributor' ) {
 544                                 $logInfo['contributor'] = $this->handleContributor();
 545                         } elseif ( $tag != '#text' ) {
 546                                 $this->warn( "Unhandled log-item XML tag $tag" );
 547                         }
 548                 }
 549
 550                 $this->processLogItem( $logInfo );
 551         }
 552
 553         /**
 554          * @param array $logInfo
 555          * @return bool|mixed
 556          */
 557         private function processLogItem( $logInfo ) {
 558                 $revision = new WikiRevision;
 559
 560                 $revision->setID( $logInfo['id'] );
 561                 $revision->setType( $logInfo['type'] );
 562                 $revision->setAction( $logInfo['action'] );
 563                 $revision->setTimestamp( $logInfo['timestamp'] );
 564                 $revision->setParams( $logInfo['params'] );
 565                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 566                 $revision->setNoUpdates( $this->mNoUpdates );
 567
 568                 if ( isset( $logInfo['comment'] ) ) {
 569                         $revision->setComment( $logInfo['comment'] );
 570                 }
 571
 572                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 573                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 574                 }
 575                 if ( isset( $logInfo['contributor']['username'] ) ) {
 576                         $revision->setUserName( $logInfo['contributor']['username'] );
 577                 }
 578
 579                 return $this->logItemCallback( $revision );
 580         }
 581
 582         private function handlePage() {
 583                 // Handle page data.
 584                 $this->debug( "Enter page handler." );
 585                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 586
 587                 // Fields that can just be stuffed in the pageInfo object
 588                 $normalFields = array( 'title', 'id', 'redirect', 'restrictions' );
 589
 590                 $skip = false;
 591                 $badTitle = false;
 592
 593                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 594                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 595                                         $this->reader->name == 'page' ) {
 596                                 break;
 597                         }
 598
 599                         $tag = $this->reader->name;
 600
 601                         if ( $badTitle ) {
 602                                 // The title is invalid, bail out of this page
 603                                 $skip = true;
 604                         } elseif ( !wfRunHooks( 'ImportHandlePageXMLTag', array( $this,
 605                                                 &$pageInfo ) ) ) {
 606                                 // Do nothing
 607                         } elseif ( in_array( $tag, $normalFields ) ) {
 608                                 $pageInfo[$tag] = $this->nodeContents();
 609                                 if ( $tag == 'title' ) {
 610                                         $title = $this->processTitle( $pageInfo['title'] );
 611
 612                                         if ( !$title ) {
 613                                                 $badTitle = true;
 614                                                 $skip = true;
 615                                         }
 616
 617                                         $this->pageCallback( $title );
 618                                         list( $pageInfo['_title'], $origTitle ) = $title;
 619                                 }
 620                         } elseif ( $tag == 'revision' ) {
 621                                 $this->handleRevision( $pageInfo );
 622                         } elseif ( $tag == 'upload' ) {
 623                                 $this->handleUpload( $pageInfo );
 624                         } elseif ( $tag != '#text' ) {
 625                                 $this->warn( "Unhandled page XML tag $tag" );
 626                                 $skip = true;
 627                         }
 628                 }
 629
 630                 $this->pageOutCallback( $pageInfo['_title'], $origTitle,
 631                                         $pageInfo['revisionCount'],
 632                                         $pageInfo['successfulRevisionCount'],
 633                                         $pageInfo );
 634         }
 635
 636         /**
 637          * @param array $pageInfo
 638          */
 639         private function handleRevision( &$pageInfo ) {
 640                 $this->debug( "Enter revision handler" );
 641                 $revisionInfo = array();
 642
 643                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'model', 'format', 'text' );
 644
 645                 $skip = false;
 646
 647                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 648                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 649                                         $this->reader->name == 'revision' ) {
 650                                 break;
 651                         }
 652
 653                         $tag = $this->reader->name;
 654
 655                         if ( !wfRunHooks( 'ImportHandleRevisionXMLTag', array(
 656                                 $this, $pageInfo, $revisionInfo
 657                         ) ) ) {
 658                                 // Do nothing
 659                         } elseif ( in_array( $tag, $normalFields ) ) {
 660                                 $revisionInfo[$tag] = $this->nodeContents();
 661                         } elseif ( $tag == 'contributor' ) {
 662                                 $revisionInfo['contributor'] = $this->handleContributor();
 663                         } elseif ( $tag != '#text' ) {
 664                                 $this->warn( "Unhandled revision XML tag $tag" );
 665                                 $skip = true;
 666                         }
 667                 }
 668
 669                 $pageInfo['revisionCount']++;
 670                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 671                         $pageInfo['successfulRevisionCount']++;
 672                 }
 673         }
 674
 675         /**
 676          * @param array $pageInfo
 677          * @param array $revisionInfo
 678          * @return bool|mixed
 679          */
 680         private function processRevision( $pageInfo, $revisionInfo ) {
 681                 $revision = new WikiRevision;
 682
 683                 if ( isset( $revisionInfo['id'] ) ) {
 684                         $revision->setID( $revisionInfo['id'] );
 685                 }
 686                 if ( isset( $revisionInfo['text'] ) ) {
 687                         $revision->setText( $revisionInfo['text'] );
 688                 }
 689                 if ( isset( $revisionInfo['model'] ) ) {
 690                         $revision->setModel( $revisionInfo['model'] );
 691                 }
 692                 if ( isset( $revisionInfo['format'] ) ) {
 693                         $revision->setFormat( $revisionInfo['format'] );
 694                 }
 695                 $revision->setTitle( $pageInfo['_title'] );
 696
 697                 if ( isset( $revisionInfo['timestamp'] ) ) {
 698                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 699                 } else {
 700                         $revision->setTimestamp( wfTimestampNow() );
 701                 }
 702
 703                 if ( isset( $revisionInfo['comment'] ) ) {
 704                         $revision->setComment( $revisionInfo['comment'] );
 705                 }
 706
 707                 if ( isset( $revisionInfo['minor'] ) ) {
 708                         $revision->setMinor( true );
 709                 }
 710                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 711                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 712                 }
 713                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 714                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 715                 }
 716                 $revision->setNoUpdates( $this->mNoUpdates );
 717
 718                 return $this->revisionCallback( $revision );
 719         }
 720
 721         /**
 722          * @param array $pageInfo
 723          * @return mixed
 724          */
 725         private function handleUpload( &$pageInfo ) {
 726                 $this->debug( "Enter upload handler" );
 727                 $uploadInfo = array();
 728
 729                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 730                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 731
 732                 $skip = false;
 733
 734                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 735                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 736                                         $this->reader->name == 'upload' ) {
 737                                 break;
 738                         }
 739
 740                         $tag = $this->reader->name;
 741
 742                         if ( !wfRunHooks( 'ImportHandleUploadXMLTag', array(
 743                                 $this, $pageInfo
 744                         ) ) ) {
 745                                 // Do nothing
 746                         } elseif ( in_array( $tag, $normalFields ) ) {
 747                                 $uploadInfo[$tag] = $this->nodeContents();
 748                         } elseif ( $tag == 'contributor' ) {
 749                                 $uploadInfo['contributor'] = $this->handleContributor();
 750                         } elseif ( $tag == 'contents' ) {
 751                                 $contents = $this->nodeContents();
 752                                 $encoding = $this->reader->getAttribute( 'encoding' );
 753                                 if ( $encoding === 'base64' ) {
 754                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 755                                         $uploadInfo['isTempSrc'] = true;
 756                                 }
 757                         } elseif ( $tag != '#text' ) {
 758                                 $this->warn( "Unhandled upload XML tag $tag" );
 759                                 $skip = true;
 760                         }
 761                 }
 762
 763                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 764                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 765                         if ( file_exists( $path ) ) {
 766                                 $uploadInfo['fileSrc'] = $path;
 767                                 $uploadInfo['isTempSrc'] = false;
 768                         }
 769                 }
 770
 771                 if ( $this->mImportUploads ) {
 772                         return $this->processUpload( $pageInfo, $uploadInfo );
 773                 }
 774         }
 775
 776         /**
 777          * @param string $contents
 778          * @return string
 779          */
 780         private function dumpTemp( $contents ) {
 781                 $filename = tempnam( wfTempDir(), 'importupload' );
 782                 file_put_contents( $filename, $contents );
 783                 return $filename;
 784         }
 785
 786         /**
 787          * @param array $pageInfo
 788          * @param array $uploadInfo
 789          * @return mixed
 790          */
 791         private function processUpload( $pageInfo, $uploadInfo ) {
 792                 $revision = new WikiRevision;
 793                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 794
 795                 $revision->setTitle( $pageInfo['_title'] );
 796                 $revision->setID( $pageInfo['id'] );
 797                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 798                 $revision->setText( $text );
 799                 $revision->setFilename( $uploadInfo['filename'] );
 800                 if ( isset( $uploadInfo['archivename'] ) ) {
 801                         $revision->setArchiveName( $uploadInfo['archivename'] );
 802                 }
 803                 $revision->setSrc( $uploadInfo['src'] );
 804                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 805                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 806                                 !empty( $uploadInfo['isTempSrc'] ) );
 807                 }
 808                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 809                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 810                 }
 811                 $revision->setSize( intval( $uploadInfo['size'] ) );
 812                 $revision->setComment( $uploadInfo['comment'] );
 813
 814                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 815                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 816                 }
 817                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 818                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 819                 }
 820                 $revision->setNoUpdates( $this->mNoUpdates );
 821
 822                 return call_user_func( $this->mUploadCallback, $revision );
 823         }
 824
 825         /**
 826          * @return array
 827          */
 828         private function handleContributor() {
 829                 $fields = array( 'id', 'ip', 'username' );
 830                 $info = array();
 831
 832                 while ( $this->reader->read() ) {
 833                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 834                                         $this->reader->name == 'contributor' ) {
 835                                 break;
 836                         }
 837
 838                         $tag = $this->reader->name;
 839
 840                         if ( in_array( $tag, $fields ) ) {
 841                                 $info[$tag] = $this->nodeContents();
 842                         }
 843                 }
 844
 845                 return $info;
 846         }
 847
 848         /**
 849          * @param string $text
 850          * @return array|bool
 851          */
 852         private function processTitle( $text ) {
 853                 global $wgCommandLineMode;
 854
 855                 $workTitle = $text;
 856                 $origTitle = Title::newFromText( $workTitle );
 857
 858                 if ( !is_null( $this->mTargetNamespace ) && !is_null( $origTitle ) ) {
 859                         # makeTitleSafe, because $origTitle can have a interwiki (different setting of interwiki map)
 860                         # and than dbKey can begin with a lowercase char
 861                         $title = Title::makeTitleSafe( $this->mTargetNamespace,
 862                                 $origTitle->getDBkey() );
 863                 } else {
 864                         if ( !is_null( $this->mTargetRootPage ) ) {
 865                                 $workTitle = $this->mTargetRootPage . '/' . $workTitle;
 866                         }
 867                         $title = Title::newFromText( $workTitle );
 868                 }
 869
 870                 if ( is_null( $title ) ) {
 871                         # Invalid page title? Ignore the page
 872                         $this->notice( 'import-error-invalid', $workTitle );
 873                         return false;
 874                 } elseif ( $title->isExternal() ) {
 875                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
 876                         return false;
 877                 } elseif ( !$title->canExist() ) {
 878                         $this->notice( 'import-error-special', $title->getPrefixedText() );
 879                         return false;
 880                 } elseif ( !$title->userCan( 'edit' ) && !$wgCommandLineMode ) {
 881                         # Do not import if the importing wiki user cannot edit this page
 882                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
 883                         return false;
 884                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$wgCommandLineMode ) {
 885                         # Do not import if the importing wiki user cannot create this page
 886                         $this->notice( 'import-error-create', $title->getPrefixedText() );
 887                         return false;
 888                 }
 889
 890                 return array( $title, $origTitle );
 891         }
 892 }
 893
 894 /** This is a horrible hack used to keep source compatibility */
 895 class UploadSourceAdapter {
 896         static $sourceRegistrations = array();
 897
 898         private $mSource;
 899         private $mBuffer;
 900         private $mPosition;
 901
 902         /**
 903          * @param string $source
 904          * @return string
 905          */
 906         static function registerSource( $source ) {
 907                 $id = wfRandomString();
 908
 909                 self::$sourceRegistrations[$id] = $source;
 910
 911                 return $id;
 912         }
 913
 914         /**
 915          * @param string $path
 916          * @param string $mode
 917          * @param array $options
 918          * @param string $opened_path
 919          * @return bool
 920          */
 921         function stream_open( $path, $mode, $options, &$opened_path ) {
 922                 $url = parse_url( $path );
 923                 $id = $url['host'];
 924
 925                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
 926                         return false;
 927                 }
 928
 929                 $this->mSource = self::$sourceRegistrations[$id];
 930
 931                 return true;
 932         }
 933
 934         /**
 935          * @param int $count
 936          * @return string
 937          */
 938         function stream_read( $count ) {
 939                 $return = '';
 940                 $leave = false;
 941
 942                 while ( !$leave && !$this->mSource->atEnd() &&
 943                                 strlen( $this->mBuffer ) < $count ) {
 944                         $read = $this->mSource->readChunk();
 945
 946                         if ( !strlen( $read ) ) {
 947                                 $leave = true;
 948                         }
 949
 950                         $this->mBuffer .= $read;
 951                 }
 952
 953                 if ( strlen( $this->mBuffer ) ) {
 954                         $return = substr( $this->mBuffer, 0, $count );
 955                         $this->mBuffer = substr( $this->mBuffer, $count );
 956                 }
 957
 958                 $this->mPosition += strlen( $return );
 959
 960                 return $return;
 961         }
 962
 963         /**
 964          * @param string $data
 965          * @return bool
 966          */
 967         function stream_write( $data ) {
 968                 return false;
 969         }
 970
 971         /**
 972          * @return mixed
 973          */
 974         function stream_tell() {
 975                 return $this->mPosition;
 976         }
 977
 978         /**
 979          * @return bool
 980          */
 981         function stream_eof() {
 982                 return $this->mSource->atEnd();
 983         }
 984
 985         /**
 986          * @return array
 987          */
 988         function url_stat() {
 989                 $result = array();
 990
 991                 $result['dev'] = $result[0] = 0;
 992                 $result['ino'] = $result[1] = 0;
 993                 $result['mode'] = $result[2] = 0;
 994                 $result['nlink'] = $result[3] = 0;
 995                 $result['uid'] = $result[4] = 0;
 996                 $result['gid'] = $result[5] = 0;
 997                 $result['rdev'] = $result[6] = 0;
 998                 $result['size'] = $result[7] = 0;
 999                 $result['atime'] = $result[8] = 0;
1000                 $result['mtime'] = $result[9] = 0;
1001                 $result['ctime'] = $result[10] = 0;
1002                 $result['blksize'] = $result[11] = 0;
1003                 $result['blocks'] = $result[12] = 0;
1004
1005                 return $result;
1006         }
1007 }
1008
1009 class XMLReader2 extends XMLReader {
1010
1011         /**
1012          * @return bool|string
1013          */
1014         function nodeContents() {
1015                 if ( $this->isEmptyElement ) {
1016                         return "";
1017                 }
1018                 $buffer = "";
1019                 while ( $this->read() ) {
1020                         switch ( $this->nodeType ) {
1021                         case XmlReader::TEXT:
1022                         case XmlReader::SIGNIFICANT_WHITESPACE:
1023                                 $buffer .= $this->value;
1024                                 break;
1025                         case XmlReader::END_ELEMENT:
1026                                 return $buffer;
1027                         }
1028                 }
1029                 return $this->close();
1030         }
1031 }
1032
1033 /**
1034  * @todo document (e.g. one-sentence class description).
1035  * @ingroup SpecialPage
1036  */
1037 class WikiRevision {
1038         var $importer = null;
1039
1040         /**
1041          * @var Title
1042          */
1043         var $title = null;
1044         var $id = 0;
1045         var $timestamp = "20010115000000";
1046         var $user = 0;
1047         var $user_text = "";
1048         var $model = null;
1049         var $format = null;
1050         var $text = "";
1051         var $content = null;
1052         var $comment = "";
1053         var $minor = false;
1054         var $type = "";
1055         var $action = "";
1056         var $params = "";
1057         var $fileSrc = '';
1058         var $sha1base36 = false;
1059         var $isTemp = false;
1060         var $archiveName = '';
1061         var $fileIsTemp;
1062         private $mNoUpdates = false;
1063
1064         /**
1065          * @param Title $title
1066          * @throws MWException
1067          */
1068         function setTitle( $title ) {
1069                 if ( is_object( $title ) ) {
1070                         $this->title = $title;
1071                 } elseif ( is_null( $title ) ) {
1072                         throw new MWException( "WikiRevision given a null title in import. You may need to adjust \$wgLegalTitleChars." );
1073                 } else {
1074                         throw new MWException( "WikiRevision given non-object title in import." );
1075                 }
1076         }
1077
1078         /**
1079          * @param int $id
1080          */
1081         function setID( $id ) {
1082                 $this->id = $id;
1083         }
1084
1085         /**
1086          * @param string $ts
1087          */
1088         function setTimestamp( $ts ) {
1089                 # 2003-08-05T18:30:02Z
1090                 $this->timestamp = wfTimestamp( TS_MW, $ts );
1091         }
1092
1093         /**
1094          * @param string $user
1095          */
1096         function setUsername( $user ) {
1097                 $this->user_text = $user;
1098         }
1099
1100         /**
1101          * @param string $ip
1102          */
1103         function setUserIP( $ip ) {
1104                 $this->user_text = $ip;
1105         }
1106
1107         /**
1108          * @param string $model
1109          */
1110         function setModel( $model ) {
1111                 $this->model = $model;
1112         }
1113
1114         /**
1115          * @param string $format
1116          */
1117         function setFormat( $format ) {
1118                 $this->format = $format;
1119         }
1120
1121         /**
1122          * @param string $text
1123          */
1124         function setText( $text ) {
1125                 $this->text = $text;
1126         }
1127
1128         /**
1129          * @param string $text
1130          */
1131         function setComment( $text ) {
1132                 $this->comment = $text;
1133         }
1134
1135         /**
1136          * @param bool $minor
1137          */
1138         function setMinor( $minor ) {
1139                 $this->minor = (bool)$minor;
1140         }
1141
1142         /**
1143          * @param mixed $src
1144          */
1145         function setSrc( $src ) {
1146                 $this->src = $src;
1147         }
1148
1149         /**
1150          * @param string $src
1151          * @param bool $isTemp
1152          */
1153         function setFileSrc( $src, $isTemp ) {
1154                 $this->fileSrc = $src;
1155                 $this->fileIsTemp = $isTemp;
1156         }
1157
1158         /**
1159          * @param string $sha1base36
1160          */
1161         function setSha1Base36( $sha1base36 ) {
1162                 $this->sha1base36 = $sha1base36;
1163         }
1164
1165         /**
1166          * @param string $filename
1167          */
1168         function setFilename( $filename ) {
1169                 $this->filename = $filename;
1170         }
1171
1172         /**
1173          * @param string $archiveName
1174          */
1175         function setArchiveName( $archiveName ) {
1176                 $this->archiveName = $archiveName;
1177         }
1178
1179         /**
1180          * @param int $size
1181          */
1182         function setSize( $size ) {
1183                 $this->size = intval( $size );
1184         }
1185
1186         /**
1187          * @param string $type
1188          */
1189         function setType( $type ) {
1190                 $this->type = $type;
1191         }
1192
1193         /**
1194          * @param string $action
1195          */
1196         function setAction( $action ) {
1197                 $this->action = $action;
1198         }
1199
1200         /**
1201          * @param array $params
1202          */
1203         function setParams( $params ) {
1204                 $this->params = $params;
1205         }
1206
1207         /**
1208          * @param bool $noupdates
1209          */
1210         public function setNoUpdates( $noupdates ) {
1211                 $this->mNoUpdates = $noupdates;
1212         }
1213
1214         /**
1215          * @return Title
1216          */
1217         function getTitle() {
1218                 return $this->title;
1219         }
1220
1221         /**
1222          * @return int
1223          */
1224         function getID() {
1225                 return $this->id;
1226         }
1227
1228         /**
1229          * @return string
1230          */
1231         function getTimestamp() {
1232                 return $this->timestamp;
1233         }
1234
1235         /**
1236          * @return string
1237          */
1238         function getUser() {
1239                 return $this->user_text;
1240         }
1241
1242         /**
1243          * @return string
1244          *
1245          * @deprecated Since 1.21, use getContent() instead.
1246          */
1247         function getText() {
1248                 ContentHandler::deprecated( __METHOD__, '1.21' );
1249
1250                 return $this->text;
1251         }
1252
1253         /**
1254          * @return Content
1255          */
1256         function getContent() {
1257                 if ( is_null( $this->content ) ) {
1258                         $this->content =
1259                                 ContentHandler::makeContent(
1260                                         $this->text,
1261                                         $this->getTitle(),
1262                                         $this->getModel(),
1263                                         $this->getFormat()
1264                                 );
1265                 }
1266
1267                 return $this->content;
1268         }
1269
1270         /**
1271          * @return string
1272          */
1273         function getModel() {
1274                 if ( is_null( $this->model ) ) {
1275                         $this->model = $this->getTitle()->getContentModel();
1276                 }
1277
1278                 return $this->model;
1279         }
1280
1281         /**
1282          * @return string
1283          */
1284         function getFormat() {
1285                 if ( is_null( $this->model ) ) {
1286                         $this->format = ContentHandler::getForTitle( $this->getTitle() )->getDefaultFormat();
1287                 }
1288
1289                 return $this->format;
1290         }
1291
1292         /**
1293          * @return string
1294          */
1295         function getComment() {
1296                 return $this->comment;
1297         }
1298
1299         /**
1300          * @return bool
1301          */
1302         function getMinor() {
1303                 return $this->minor;
1304         }
1305
1306         /**
1307          * @return mixed
1308          */
1309         function getSrc() {
1310                 return $this->src;
1311         }
1312
1313         /**
1314          * @return bool|string
1315          */
1316         function getSha1() {
1317                 if ( $this->sha1base36 ) {
1318                         return wfBaseConvert( $this->sha1base36, 36, 16 );
1319                 }
1320                 return false;
1321         }
1322
1323         /**
1324          * @return string
1325          */
1326         function getFileSrc() {
1327                 return $this->fileSrc;
1328         }
1329
1330         /**
1331          * @return bool
1332          */
1333         function isTempSrc() {
1334                 return $this->isTemp;
1335         }
1336
1337         /**
1338          * @return mixed
1339          */
1340         function getFilename() {
1341                 return $this->filename;
1342         }
1343
1344         /**
1345          * @return string
1346          */
1347         function getArchiveName() {
1348                 return $this->archiveName;
1349         }
1350
1351         /**
1352          * @return mixed
1353          */
1354         function getSize() {
1355                 return $this->size;
1356         }
1357
1358         /**
1359          * @return string
1360          */
1361         function getType() {
1362                 return $this->type;
1363         }
1364
1365         /**
1366          * @return string
1367          */
1368         function getAction() {
1369                 return $this->action;
1370         }
1371
1372         /**
1373          * @return string
1374          */
1375         function getParams() {
1376                 return $this->params;
1377         }
1378
1379         /**
1380          * @return bool
1381          */
1382         function importOldRevision() {
1383                 $dbw = wfGetDB( DB_MASTER );
1384
1385                 # Sneak a single revision into place
1386                 $user = User::newFromName( $this->getUser() );
1387                 if ( $user ) {
1388                         $userId = intval( $user->getId() );
1389                         $userText = $user->getName();
1390                         $userObj = $user;
1391                 } else {
1392                         $userId = 0;
1393                         $userText = $this->getUser();
1394                         $userObj = new User;
1395                 }
1396
1397                 // avoid memory leak...?
1398                 $linkCache = LinkCache::singleton();
1399                 $linkCache->clear();
1400
1401                 $page = WikiPage::factory( $this->title );
1402                 if ( !$page->exists() ) {
1403                         # must create the page...
1404                         $pageId = $page->insertOn( $dbw );
1405                         $created = true;
1406                         $oldcountable = null;
1407                 } else {
1408                         $pageId = $page->getId();
1409                         $created = false;
1410
1411                         $prior = $dbw->selectField( 'revision', '1',
1412                                 array( 'rev_page' => $pageId,
1413                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1414                                         'rev_user_text' => $userText,
1415                                         'rev_comment' => $this->getComment() ),
1416                                 __METHOD__
1417                         );
1418                         if ( $prior ) {
1419                                 // @todo FIXME: This could fail slightly for multiple matches :P
1420                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1421                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1422                                 return false;
1423                         }
1424                         $oldcountable = $page->isCountable();
1425                 }
1426
1427                 # @todo FIXME: Use original rev_id optionally (better for backups)
1428                 # Insert the row
1429                 $revision = new Revision( array(
1430                         'title' => $this->title,
1431                         'page' => $pageId,
1432                         'content_model' => $this->getModel(),
1433                         'content_format' => $this->getFormat(),
1434                         'text' => $this->getContent()->serialize( $this->getFormat() ), //XXX: just set 'content' => $this->getContent()?
1435                         'comment' => $this->getComment(),
1436                         'user' => $userId,
1437                         'user_text' => $userText,
1438                         'timestamp' => $this->timestamp,
1439                         'minor_edit' => $this->minor,
1440                         ) );
1441                 $revision->insertOn( $dbw );
1442                 $changed = $page->updateIfNewerOn( $dbw, $revision );
1443
1444                 if ( $changed !== false && !$this->mNoUpdates ) {
1445                         wfDebug( __METHOD__ . ": running updates\n" );
1446                         $page->doEditUpdates( $revision, $userObj, array( 'created' => $created, 'oldcountable' => $oldcountable ) );
1447                 }
1448
1449                 return true;
1450         }
1451
1452         /**
1453          * @return mixed
1454          */
1455         function importLogItem() {
1456                 $dbw = wfGetDB( DB_MASTER );
1457                 # @todo FIXME: This will not record autoblocks
1458                 if ( !$this->getTitle() ) {
1459                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1460                                 $this->timestamp . "\n" );
1461                         return;
1462                 }
1463                 # Check if it exists already
1464                 // @todo FIXME: Use original log ID (better for backups)
1465                 $prior = $dbw->selectField( 'logging', '1',
1466                         array( 'log_type' => $this->getType(),
1467                                 'log_action' => $this->getAction(),
1468                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1469                                 'log_namespace' => $this->getTitle()->getNamespace(),
1470                                 'log_title' => $this->getTitle()->getDBkey(),
1471                                 'log_comment' => $this->getComment(),
1472                                 #'log_user_text' => $this->user_text,
1473                                 'log_params' => $this->params ),
1474                         __METHOD__
1475                 );
1476                 // @todo FIXME: This could fail slightly for multiple matches :P
1477                 if ( $prior ) {
1478                         wfDebug( __METHOD__ . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp " .
1479                                 $this->timestamp . "\n" );
1480                         return;
1481                 }
1482                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1483                 $data = array(
1484                         'log_id' => $log_id,
1485                         'log_type' => $this->type,
1486                         'log_action' => $this->action,
1487                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1488                         'log_user' => User::idFromName( $this->user_text ),
1489                         #'log_user_text' => $this->user_text,
1490                         'log_namespace' => $this->getTitle()->getNamespace(),
1491                         'log_title' => $this->getTitle()->getDBkey(),
1492                         'log_comment' => $this->getComment(),
1493                         'log_params' => $this->params
1494                 );
1495                 $dbw->insert( 'logging', $data, __METHOD__ );
1496         }
1497
1498         /**
1499          * @return bool
1500          */
1501         function importUpload() {
1502                 # Construct a file
1503                 $archiveName = $this->getArchiveName();
1504                 if ( $archiveName ) {
1505                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1506                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1507                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1508                 } else {
1509                         $file = wfLocalFile( $this->getTitle() );
1510                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1511                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1512                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1513                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1514                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1515                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1516                         }
1517                 }
1518                 if ( !$file ) {
1519                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1520                         return false;
1521                 }
1522
1523                 # Get the file source or download if necessary
1524                 $source = $this->getFileSrc();
1525                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1526                 if ( !$source ) {
1527                         $source = $this->downloadSource();
1528                         $flags |= File::DELETE_SOURCE;
1529                 }
1530                 if ( !$source ) {
1531                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1532                         return false;
1533                 }
1534                 $sha1 = $this->getSha1();
1535                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1536                         if ( $flags & File::DELETE_SOURCE ) {
1537                                 # Broken file; delete it if it is a temporary file
1538                                 unlink( $source );
1539                         }
1540                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1541                         return false;
1542                 }
1543
1544                 $user = User::newFromName( $this->user_text );
1545
1546                 # Do the actual upload
1547                 if ( $archiveName ) {
1548                         $status = $file->uploadOld( $source, $archiveName,
1549                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1550                 } else {
1551                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1552                                 $flags, false, $this->getTimestamp(), $user );
1553                 }
1554
1555                 if ( $status->isGood() ) {
1556                         wfDebug( __METHOD__ . ": Successful\n" );
1557                         return true;
1558                 } else {
1559                         wfDebug( __METHOD__ . ': failed: ' . $status->getXml() . "\n" );
1560                         return false;
1561                 }
1562         }
1563
1564         /**
1565          * @return bool|string
1566          */
1567         function downloadSource() {
1568                 global $wgEnableUploads;
1569                 if ( !$wgEnableUploads ) {
1570                         return false;
1571                 }
1572
1573                 $tempo = tempnam( wfTempDir(), 'download' );
1574                 $f = fopen( $tempo, 'wb' );
1575                 if ( !$f ) {
1576                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1577                         return false;
1578                 }
1579
1580                 // @todo FIXME!
1581                 $src = $this->getSrc();
1582                 $data = Http::get( $src );
1583                 if ( !$data ) {
1584                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1585                         fclose( $f );
1586                         unlink( $tempo );
1587                         return false;
1588                 }
1589
1590                 fwrite( $f, $data );
1591                 fclose( $f );
1592
1593                 return $tempo;
1594         }
1595
1596 }
1597
1598 /**
1599  * @todo document (e.g. one-sentence class description).
1600  * @ingroup SpecialPage
1601  */
1602 class ImportStringSource {
1603         function __construct( $string ) {
1604                 $this->mString = $string;
1605                 $this->mRead = false;
1606         }
1607
1608         /**
1609          * @return bool
1610          */
1611         function atEnd() {
1612                 return $this->mRead;
1613         }
1614
1615         /**
1616          * @return bool|string
1617          */
1618         function readChunk() {
1619                 if ( $this->atEnd() ) {
1620                         return false;
1621                 }
1622                 $this->mRead = true;
1623                 return $this->mString;
1624         }
1625 }
1626
1627 /**
1628  * @todo document (e.g. one-sentence class description).
1629  * @ingroup SpecialPage
1630  */
1631 class ImportStreamSource {
1632         function __construct( $handle ) {
1633                 $this->mHandle = $handle;
1634         }
1635
1636         /**
1637          * @return bool
1638          */
1639         function atEnd() {
1640                 return feof( $this->mHandle );
1641         }
1642
1643         /**
1644          * @return string
1645          */
1646         function readChunk() {
1647                 return fread( $this->mHandle, 32768 );
1648         }
1649
1650         /**
1651          * @param string $filename
1652          * @return Status
1653          */
1654         static function newFromFile( $filename ) {
1655                 wfSuppressWarnings();
1656                 $file = fopen( $filename, 'rt' );
1657                 wfRestoreWarnings();
1658                 if ( !$file ) {
1659                         return Status::newFatal( "importcantopen" );
1660                 }
1661                 return Status::newGood( new ImportStreamSource( $file ) );
1662         }
1663
1664         /**
1665          * @param string $fieldname
1666          * @return Status
1667          */
1668         static function newFromUpload( $fieldname = "xmlimport" ) {
1669                 $upload =& $_FILES[$fieldname];
1670
1671                 if ( $upload === null || !$upload['name'] ) {
1672                         return Status::newFatal( 'importnofile' );
1673                 }
1674                 if ( !empty( $upload['error'] ) ) {
1675                         switch ( $upload['error'] ) {
1676                                 case 1: # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1677                                         return Status::newFatal( 'importuploaderrorsize' );
1678                                 case 2: # The uploaded file exceeds the MAX_FILE_SIZE directive that was specified in the HTML form.
1679                                         return Status::newFatal( 'importuploaderrorsize' );
1680                                 case 3: # The uploaded file was only partially uploaded
1681                                         return Status::newFatal( 'importuploaderrorpartial' );
1682                                 case 6: #Missing a temporary folder.
1683                                         return Status::newFatal( 'importuploaderrortemp' );
1684                                 # case else: # Currently impossible
1685                         }
1686
1687                 }
1688                 $fname = $upload['tmp_name'];
1689                 if ( is_uploaded_file( $fname ) ) {
1690                         return ImportStreamSource::newFromFile( $fname );
1691                 } else {
1692                         return Status::newFatal( 'importnofile' );
1693                 }
1694         }
1695
1696         /**
1697          * @param string $url
1698          * @param string $method
1699          * @return Status
1700          */
1701         static function newFromURL( $url, $method = 'GET' ) {
1702                 wfDebug( __METHOD__ . ": opening $url\n" );
1703                 # Use the standard HTTP fetch function; it times out
1704                 # quicker and sorts out user-agent problems which might
1705                 # otherwise prevent importing from large sites, such
1706                 # as the Wikimedia cluster, etc.
1707                 $data = Http::request( $method, $url, array( 'followRedirects' => true ) );
1708                 if ( $data !== false ) {
1709                         $file = tmpfile();
1710                         fwrite( $file, $data );
1711                         fflush( $file );
1712                         fseek( $file, 0 );
1713                         return Status::newGood( new ImportStreamSource( $file ) );
1714                 } else {
1715                         return Status::newFatal( 'importcantopen' );
1716                 }
1717         }
1718
1719         /**
1720          * @param string $interwiki
1721          * @param string $page
1722          * @param bool $history
1723          * @param bool $templates
1724          * @param int $pageLinkDepth
1725          * @return Status
1726          */
1727         public static function newFromInterwiki( $interwiki, $page, $history = false, $templates = false, $pageLinkDepth = 0 ) {
1728                 if ( $page == '' ) {
1729                         return Status::newFatal( 'import-noarticle' );
1730                 }
1731                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1732                 if ( is_null( $link ) || !$link->isExternal() ) {
1733                         return Status::newFatal( 'importbadinterwiki' );
1734                 } else {
1735                         $params = array();
1736                         if ( $history ) {
1737                                 $params['history'] = 1;
1738                         }
1739                         if ( $templates ) {
1740                                 $params['templates'] = 1;
1741                         }
1742                         if ( $pageLinkDepth ) {
1743                                 $params['pagelink-depth'] = $pageLinkDepth;
1744                         }
1745                         $url = $link->getFullURL( $params );
1746                         # For interwikis, use POST to avoid redirects.
1747                         return ImportStreamSource::newFromURL( $url, "POST" );
1748                 }
1749         }
1750 }