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         /** @var Config */
  41         private $config;
  42
  43         /**
  44          * Creates an ImportXMLReader drawing from the source provided
  45          * @param ImportStreamSource $source
  46          * @param Config $config
  47          */
  48         function __construct( ImportStreamSource $source, Config $config = null ) {
  49                 $this->reader = new XMLReader();
  50                 if ( !$config ) {
  51                         wfDeprecated( __METHOD__ . ' without a Config instance', '1.25' );
  52                         $config = ConfigFactory::getDefaultInstance()->makeConfig( 'main' );
  53                 }
  54                 $this->config = $config;
  55
  56                 if ( !in_array( 'uploadsource', stream_get_wrappers() ) ) {
  57                         stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  58                 }
  59                 $id = UploadSourceAdapter::registerSource( $source );
  60                 if ( defined( 'LIBXML_PARSEHUGE' ) ) {
  61                         $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  62                 } else {
  63                         $this->reader->open( "uploadsource://$id" );
  64                 }
  65
  66                 // Default callbacks
  67                 $this->setRevisionCallback( array( $this, "importRevision" ) );
  68                 $this->setUploadCallback( array( $this, 'importUpload' ) );
  69                 $this->setLogItemCallback( array( $this, 'importLogItem' ) );
  70                 $this->setPageOutCallback( array( $this, 'finishImportPage' ) );
  71         }
  72
  73         /**
  74          * @return null|XMLReader
  75          */
  76         public function getReader() {
  77                 return $this->reader;
  78         }
  79
  80         public function throwXmlError( $err ) {
  81                 $this->debug( "FAILURE: $err" );
  82                 wfDebug( "WikiImporter XML error: $err\n" );
  83         }
  84
  85         public function debug( $data ) {
  86                 if ( $this->mDebug ) {
  87                         wfDebug( "IMPORT: $data\n" );
  88                 }
  89         }
  90
  91         public function warn( $data ) {
  92                 wfDebug( "IMPORT: $data\n" );
  93         }
  94
  95         public function notice( $msg /*, $param, ...*/ ) {
  96                 $params = func_get_args();
  97                 array_shift( $params );
  98
  99                 if ( is_callable( $this->mNoticeCallback ) ) {
 100                         call_user_func( $this->mNoticeCallback, $msg, $params );
 101                 } else { # No ImportReporter -> CLI
 102                         echo wfMessage( $msg, $params )->text() . "\n";
 103                 }
 104         }
 105
 106         /**
 107          * Set debug mode...
 108          * @param bool $debug
 109          */
 110         function setDebug( $debug ) {
 111                 $this->mDebug = $debug;
 112         }
 113
 114         /**
 115          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
 116          * @param bool $noupdates
 117          */
 118         function setNoUpdates( $noupdates ) {
 119                 $this->mNoUpdates = $noupdates;
 120         }
 121
 122         /**
 123          * Set a callback that displays notice messages
 124          *
 125          * @param callable $callback
 126          * @return callable
 127          */
 128         public function setNoticeCallback( $callback ) {
 129                 return wfSetVar( $this->mNoticeCallback, $callback );
 130         }
 131
 132         /**
 133          * Sets the action to perform as each new page in the stream is reached.
 134          * @param callable $callback
 135          * @return callable
 136          */
 137         public function setPageCallback( $callback ) {
 138                 $previous = $this->mPageCallback;
 139                 $this->mPageCallback = $callback;
 140                 return $previous;
 141         }
 142
 143         /**
 144          * Sets the action to perform as each page in the stream is completed.
 145          * Callback accepts the page title (as a Title object), a second object
 146          * with the original title form (in case it's been overridden into a
 147          * local namespace), and a count of revisions.
 148          *
 149          * @param callable $callback
 150          * @return callable
 151          */
 152         public function setPageOutCallback( $callback ) {
 153                 $previous = $this->mPageOutCallback;
 154                 $this->mPageOutCallback = $callback;
 155                 return $previous;
 156         }
 157
 158         /**
 159          * Sets the action to perform as each page revision is reached.
 160          * @param callable $callback
 161          * @return callable
 162          */
 163         public function setRevisionCallback( $callback ) {
 164                 $previous = $this->mRevisionCallback;
 165                 $this->mRevisionCallback = $callback;
 166                 return $previous;
 167         }
 168
 169         /**
 170          * Sets the action to perform as each file upload version is reached.
 171          * @param callable $callback
 172          * @return callable
 173          */
 174         public function setUploadCallback( $callback ) {
 175                 $previous = $this->mUploadCallback;
 176                 $this->mUploadCallback = $callback;
 177                 return $previous;
 178         }
 179
 180         /**
 181          * Sets the action to perform as each log item reached.
 182          * @param callable $callback
 183          * @return callable
 184          */
 185         public function setLogItemCallback( $callback ) {
 186                 $previous = $this->mLogItemCallback;
 187                 $this->mLogItemCallback = $callback;
 188                 return $previous;
 189         }
 190
 191         /**
 192          * Sets the action to perform when site info is encountered
 193          * @param callable $callback
 194          * @return callable
 195          */
 196         public function setSiteInfoCallback( $callback ) {
 197                 $previous = $this->mSiteInfoCallback;
 198                 $this->mSiteInfoCallback = $callback;
 199                 return $previous;
 200         }
 201
 202         /**
 203          * Set a target namespace to override the defaults
 204          * @param null|int $namespace
 205          * @return bool
 206          */
 207         public function setTargetNamespace( $namespace ) {
 208                 if ( is_null( $namespace ) ) {
 209                         // Don't override namespaces
 210                         $this->mTargetNamespace = null;
 211                 } elseif ( $namespace >= 0 ) {
 212                         // @todo FIXME: Check for validity
 213                         $this->mTargetNamespace = intval( $namespace );
 214                 } else {
 215                         return false;
 216                 }
 217         }
 218
 219         /**
 220          * Set a target root page under which all pages are imported
 221          * @param null|string $rootpage
 222          * @return Status
 223          */
 224         public function setTargetRootPage( $rootpage ) {
 225                 $status = Status::newGood();
 226                 if ( is_null( $rootpage ) ) {
 227                         // No rootpage
 228                         $this->mTargetRootPage = null;
 229                 } elseif ( $rootpage !== '' ) {
 230                         $rootpage = rtrim( $rootpage, '/' ); //avoid double slashes
 231                         $title = Title::newFromText( $rootpage, !is_null( $this->mTargetNamespace )
 232                                 ? $this->mTargetNamespace
 233                                 : NS_MAIN
 234                         );
 235
 236                         if ( !$title || $title->isExternal() ) {
 237                                 $status->fatal( 'import-rootpage-invalid' );
 238                         } else {
 239                                 if ( !MWNamespace::hasSubpages( $title->getNamespace() ) ) {
 240                                         global $wgContLang;
 241
 242                                         $displayNSText = $title->getNamespace() == NS_MAIN
 243                                                 ? wfMessage( 'blanknamespace' )->text()
 244                                                 : $wgContLang->getNsText( $title->getNamespace() );
 245                                         $status->fatal( 'import-rootpage-nosubpage', $displayNSText );
 246                                 } else {
 247                                         // set namespace to 'all', so the namespace check in processTitle() can passed
 248                                         $this->setTargetNamespace( null );
 249                                         $this->mTargetRootPage = $title->getPrefixedDBkey();
 250                                 }
 251                         }
 252                 }
 253                 return $status;
 254         }
 255
 256         /**
 257          * @param string $dir
 258          */
 259         public function setImageBasePath( $dir ) {
 260                 $this->mImageBasePath = $dir;
 261         }
 262
 263         /**
 264          * @param bool $import
 265          */
 266         public function setImportUploads( $import ) {
 267                 $this->mImportUploads = $import;
 268         }
 269
 270         /**
 271          * Default per-revision callback, performs the import.
 272          * @param WikiRevision $revision
 273          * @return bool
 274          */
 275         public function importRevision( $revision ) {
 276                 if ( !$revision->getContentHandler()->canBeUsedOn( $revision->getTitle() ) ) {
 277                         $this->notice( 'import-error-bad-location',
 278                                 $revision->getTitle()->getPrefixedText(),
 279                                 $revision->getID(),
 280                                 $revision->getModel(),
 281                                 $revision->getFormat() );
 282
 283                         return false;
 284                 }
 285
 286                 try {
 287                         $dbw = wfGetDB( DB_MASTER );
 288                         return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
 289                 } catch ( MWContentSerializationException $ex ) {
 290                         $this->notice( 'import-error-unserialize',
 291                                 $revision->getTitle()->getPrefixedText(),
 292                                 $revision->getID(),
 293                                 $revision->getModel(),
 294                                 $revision->getFormat() );
 295                 }
 296
 297                 return false;
 298         }
 299
 300         /**
 301          * Default per-revision callback, performs the import.
 302          * @param WikiRevision $revision
 303          * @return bool
 304          */
 305         public function importLogItem( $revision ) {
 306                 $dbw = wfGetDB( DB_MASTER );
 307                 return $dbw->deadlockLoop( array( $revision, 'importLogItem' ) );
 308         }
 309
 310         /**
 311          * Dummy for now...
 312          * @param WikiRevision $revision
 313          * @return bool
 314          */
 315         public function importUpload( $revision ) {
 316                 $dbw = wfGetDB( DB_MASTER );
 317                 return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
 318         }
 319
 320         /**
 321          * Mostly for hook use
 322          * @param Title $title
 323          * @param string $origTitle
 324          * @param int $revCount
 325          * @param int $sRevCount
 326          * @param array $pageInfo
 327          * @return bool
 328          */
 329         public function finishImportPage( $title, $origTitle, $revCount, $sRevCount, $pageInfo ) {
 330                 $args = func_get_args();
 331                 return wfRunHooks( 'AfterImportPage', $args );
 332         }
 333
 334         /**
 335          * Alternate per-revision callback, for debugging.
 336          * @param WikiRevision $revision
 337          */
 338         public function debugRevisionHandler( &$revision ) {
 339                 $this->debug( "Got revision:" );
 340                 if ( is_object( $revision->title ) ) {
 341                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 342                 } else {
 343                         $this->debug( "-- Title: <invalid>" );
 344                 }
 345                 $this->debug( "-- User: " . $revision->user_text );
 346                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 347                 $this->debug( "-- Comment: " . $revision->comment );
 348                 $this->debug( "-- Text: " . $revision->text );
 349         }
 350
 351         /**
 352          * Notify the callback function when a new "<page>" is reached.
 353          * @param Title $title
 354          */
 355         function pageCallback( $title ) {
 356                 if ( isset( $this->mPageCallback ) ) {
 357                         call_user_func( $this->mPageCallback, $title );
 358                 }
 359         }
 360
 361         /**
 362          * Notify the callback function when a "</page>" is closed.
 363          * @param Title $title
 364          * @param Title $origTitle
 365          * @param int $revCount
 366          * @param int $sucCount Number of revisions for which callback returned true
 367          * @param array $pageInfo Associative array of page information
 368          */
 369         private function pageOutCallback( $title, $origTitle, $revCount, $sucCount, $pageInfo ) {
 370                 if ( isset( $this->mPageOutCallback ) ) {
 371                         $args = func_get_args();
 372                         call_user_func_array( $this->mPageOutCallback, $args );
 373                 }
 374         }
 375
 376         /**
 377          * Notify the callback function of a revision
 378          * @param WikiRevision $revision
 379          * @return bool|mixed
 380          */
 381         private function revisionCallback( $revision ) {
 382                 if ( isset( $this->mRevisionCallback ) ) {
 383                         return call_user_func_array( $this->mRevisionCallback,
 384                                         array( $revision, $this ) );
 385                 } else {
 386                         return false;
 387                 }
 388         }
 389
 390         /**
 391          * Notify the callback function of a new log item
 392          * @param WikiRevision $revision
 393          * @return bool|mixed
 394          */
 395         private function logItemCallback( $revision ) {
 396                 if ( isset( $this->mLogItemCallback ) ) {
 397                         return call_user_func_array( $this->mLogItemCallback,
 398                                         array( $revision, $this ) );
 399                 } else {
 400                         return false;
 401                 }
 402         }
 403
 404         /**
 405          * Retrieves the contents of the named attribute of the current element.
 406          * @param string $attr The name of the attribute
 407          * @return string The value of the attribute or an empty string if it is not set in the current element.
 408          */
 409         public function nodeAttribute( $attr ) {
 410                 return $this->reader->getAttribute( $attr );
 411         }
 412
 413         /**
 414          * Shouldn't something like this be built-in to XMLReader?
 415          * Fetches text contents of the current element, assuming
 416          * no sub-elements or such scary things.
 417          * @return string
 418          * @access private
 419          */
 420         public function nodeContents() {
 421                 if ( $this->reader->isEmptyElement ) {
 422                         return "";
 423                 }
 424                 $buffer = "";
 425                 while ( $this->reader->read() ) {
 426                         switch ( $this->reader->nodeType ) {
 427                         case XmlReader::TEXT:
 428                         case XmlReader::SIGNIFICANT_WHITESPACE:
 429                                 $buffer .= $this->reader->value;
 430                                 break;
 431                         case XmlReader::END_ELEMENT:
 432                                 return $buffer;
 433                         }
 434                 }
 435
 436                 $this->reader->close();
 437                 return '';
 438         }
 439
 440         /**
 441          * Primary entry point
 442          * @throws MWException
 443          * @return bool
 444          */
 445         public function doImport() {
 446                 // Calls to reader->read need to be wrapped in calls to
 447                 // libxml_disable_entity_loader() to avoid local file
 448                 // inclusion attacks (bug 46932).
 449                 $oldDisable = libxml_disable_entity_loader( true );
 450                 $this->reader->read();
 451
 452                 if ( $this->reader->name != 'mediawiki' ) {
 453                         libxml_disable_entity_loader( $oldDisable );
 454                         throw new MWException( "Expected <mediawiki> tag, got " .
 455                                 $this->reader->name );
 456                 }
 457                 $this->debug( "<mediawiki> tag is correct." );
 458
 459                 $this->debug( "Starting primary dump processing loop." );
 460
 461                 $keepReading = $this->reader->read();
 462                 $skip = false;
 463                 while ( $keepReading ) {
 464                         $tag = $this->reader->name;
 465                         $type = $this->reader->nodeType;
 466
 467                         if ( !wfRunHooks( 'ImportHandleToplevelXMLTag', array( $this ) ) ) {
 468                                 // Do nothing
 469                         } elseif ( $tag == 'mediawiki' && $type == XmlReader::END_ELEMENT ) {
 470                                 break;
 471                         } elseif ( $tag == 'siteinfo' ) {
 472                                 $this->handleSiteInfo();
 473                         } elseif ( $tag == 'page' ) {
 474                                 $this->handlePage();
 475                         } elseif ( $tag == 'logitem' ) {
 476                                 $this->handleLogItem();
 477                         } elseif ( $tag != '#text' ) {
 478                                 $this->warn( "Unhandled top-level XML tag $tag" );
 479
 480                                 $skip = true;
 481                         }
 482
 483                         if ( $skip ) {
 484                                 $keepReading = $this->reader->next();
 485                                 $skip = false;
 486                                 $this->debug( "Skip" );
 487                         } else {
 488                                 $keepReading = $this->reader->read();
 489                         }
 490                 }
 491
 492                 libxml_disable_entity_loader( $oldDisable );
 493                 return true;
 494         }
 495
 496         /**
 497          * @return bool
 498          * @throws MWException
 499          */
 500         private function handleSiteInfo() {
 501                 // Site info is useful, but not actually used for dump imports.
 502                 // Includes a quick short-circuit to save performance.
 503                 if ( !$this->mSiteInfoCallback ) {
 504                         $this->reader->next();
 505                         return true;
 506                 }
 507                 throw new MWException( "SiteInfo tag is not yet handled, do not set mSiteInfoCallback" );
 508         }
 509
 510         private function handleLogItem() {
 511                 $this->debug( "Enter log item handler." );
 512                 $logInfo = array();
 513
 514                 // Fields that can just be stuffed in the pageInfo object
 515                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 516                                         'logtitle', 'params' );
 517
 518                 while ( $this->reader->read() ) {
 519                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 520                                         $this->reader->name == 'logitem' ) {
 521                                 break;
 522                         }
 523
 524                         $tag = $this->reader->name;
 525
 526                         if ( !wfRunHooks( 'ImportHandleLogItemXMLTag', array(
 527                                 $this, $logInfo
 528                         ) ) ) {
 529                                 // Do nothing
 530                         } elseif ( in_array( $tag, $normalFields ) ) {
 531                                 $logInfo[$tag] = $this->nodeContents();
 532                         } elseif ( $tag == 'contributor' ) {
 533                                 $logInfo['contributor'] = $this->handleContributor();
 534                         } elseif ( $tag != '#text' ) {
 535                                 $this->warn( "Unhandled log-item XML tag $tag" );
 536                         }
 537                 }
 538
 539                 $this->processLogItem( $logInfo );
 540         }
 541
 542         /**
 543          * @param array $logInfo
 544          * @return bool|mixed
 545          */
 546         private function processLogItem( $logInfo ) {
 547                 $revision = new WikiRevision( $this->config );
 548
 549                 $revision->setID( $logInfo['id'] );
 550                 $revision->setType( $logInfo['type'] );
 551                 $revision->setAction( $logInfo['action'] );
 552                 $revision->setTimestamp( $logInfo['timestamp'] );
 553                 $revision->setParams( $logInfo['params'] );
 554                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 555                 $revision->setNoUpdates( $this->mNoUpdates );
 556
 557                 if ( isset( $logInfo['comment'] ) ) {
 558                         $revision->setComment( $logInfo['comment'] );
 559                 }
 560
 561                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 562                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 563                 }
 564                 if ( isset( $logInfo['contributor']['username'] ) ) {
 565                         $revision->setUserName( $logInfo['contributor']['username'] );
 566                 }
 567
 568                 return $this->logItemCallback( $revision );
 569         }
 570
 571         private function handlePage() {
 572                 // Handle page data.
 573                 $this->debug( "Enter page handler." );
 574                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 575
 576                 // Fields that can just be stuffed in the pageInfo object
 577                 $normalFields = array( 'title', 'id', 'redirect', 'restrictions' );
 578
 579                 $skip = false;
 580                 $badTitle = false;
 581
 582                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 583                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 584                                         $this->reader->name == 'page' ) {
 585                                 break;
 586                         }
 587
 588                         $tag = $this->reader->name;
 589
 590                         if ( $badTitle ) {
 591                                 // The title is invalid, bail out of this page
 592                                 $skip = true;
 593                         } elseif ( !wfRunHooks( 'ImportHandlePageXMLTag', array( $this,
 594                                                 &$pageInfo ) ) ) {
 595                                 // Do nothing
 596                         } elseif ( in_array( $tag, $normalFields ) ) {
 597                                 // An XML snippet:
 598                                 // <page>
 599                                 //     <id>123</id>
 600                                 //     <title>Page</title>
 601                                 //     <redirect title="NewTitle"/>
 602                                 //     ...
 603                                 // Because the redirect tag is built differently, we need special handling for that case.
 604                                 if ( $tag == 'redirect' ) {
 605                                         $pageInfo[$tag] = $this->nodeAttribute( 'title' );
 606                                 } else {
 607                                         $pageInfo[$tag] = $this->nodeContents();
 608                                         if ( $tag == 'title' ) {
 609                                                 $title = $this->processTitle( $pageInfo['title'] );
 610
 611                                                 if ( !$title ) {
 612                                                         $badTitle = true;
 613                                                         $skip = true;
 614                                                 }
 615
 616                                                 $this->pageCallback( $title );
 617                                                 list( $pageInfo['_title'], $origTitle ) = $title;
 618                                         }
 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( $this->config );
 682
 683                 if ( isset( $revisionInfo['id'] ) ) {
 684                         $revision->setID( $revisionInfo['id'] );
 685                 }
 686                 if ( isset( $revisionInfo['model'] ) ) {
 687                         $revision->setModel( $revisionInfo['model'] );
 688                 }
 689                 if ( isset( $revisionInfo['format'] ) ) {
 690                         $revision->setFormat( $revisionInfo['format'] );
 691                 }
 692                 $revision->setTitle( $pageInfo['_title'] );
 693
 694                 if ( isset( $revisionInfo['text'] ) ) {
 695                         $handler = $revision->getContentHandler();
 696                         $text = $handler->importTransform(
 697                                 $revisionInfo['text'],
 698                                 $revision->getFormat() );
 699
 700                         $revision->setText( $text );
 701                 }
 702                 if ( isset( $revisionInfo['timestamp'] ) ) {
 703                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 704                 } else {
 705                         $revision->setTimestamp( wfTimestampNow() );
 706                 }
 707
 708                 if ( isset( $revisionInfo['comment'] ) ) {
 709                         $revision->setComment( $revisionInfo['comment'] );
 710                 }
 711
 712                 if ( isset( $revisionInfo['minor'] ) ) {
 713                         $revision->setMinor( true );
 714                 }
 715                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 716                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 717                 }
 718                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 719                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 720                 }
 721                 $revision->setNoUpdates( $this->mNoUpdates );
 722
 723                 return $this->revisionCallback( $revision );
 724         }
 725
 726         /**
 727          * @param array $pageInfo
 728          * @return mixed
 729          */
 730         private function handleUpload( &$pageInfo ) {
 731                 $this->debug( "Enter upload handler" );
 732                 $uploadInfo = array();
 733
 734                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 735                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 736
 737                 $skip = false;
 738
 739                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 740                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 741                                         $this->reader->name == 'upload' ) {
 742                                 break;
 743                         }
 744
 745                         $tag = $this->reader->name;
 746
 747                         if ( !wfRunHooks( 'ImportHandleUploadXMLTag', array(
 748                                 $this, $pageInfo
 749                         ) ) ) {
 750                                 // Do nothing
 751                         } elseif ( in_array( $tag, $normalFields ) ) {
 752                                 $uploadInfo[$tag] = $this->nodeContents();
 753                         } elseif ( $tag == 'contributor' ) {
 754                                 $uploadInfo['contributor'] = $this->handleContributor();
 755                         } elseif ( $tag == 'contents' ) {
 756                                 $contents = $this->nodeContents();
 757                                 $encoding = $this->reader->getAttribute( 'encoding' );
 758                                 if ( $encoding === 'base64' ) {
 759                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 760                                         $uploadInfo['isTempSrc'] = true;
 761                                 }
 762                         } elseif ( $tag != '#text' ) {
 763                                 $this->warn( "Unhandled upload XML tag $tag" );
 764                                 $skip = true;
 765                         }
 766                 }
 767
 768                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 769                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 770                         if ( file_exists( $path ) ) {
 771                                 $uploadInfo['fileSrc'] = $path;
 772                                 $uploadInfo['isTempSrc'] = false;
 773                         }
 774                 }
 775
 776                 if ( $this->mImportUploads ) {
 777                         return $this->processUpload( $pageInfo, $uploadInfo );
 778                 }
 779         }
 780
 781         /**
 782          * @param string $contents
 783          * @return string
 784          */
 785         private function dumpTemp( $contents ) {
 786                 $filename = tempnam( wfTempDir(), 'importupload' );
 787                 file_put_contents( $filename, $contents );
 788                 return $filename;
 789         }
 790
 791         /**
 792          * @param array $pageInfo
 793          * @param array $uploadInfo
 794          * @return mixed
 795          */
 796         private function processUpload( $pageInfo, $uploadInfo ) {
 797                 $revision = new WikiRevision( $this->config );
 798                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 799
 800                 $revision->setTitle( $pageInfo['_title'] );
 801                 $revision->setID( $pageInfo['id'] );
 802                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 803                 $revision->setText( $text );
 804                 $revision->setFilename( $uploadInfo['filename'] );
 805                 if ( isset( $uploadInfo['archivename'] ) ) {
 806                         $revision->setArchiveName( $uploadInfo['archivename'] );
 807                 }
 808                 $revision->setSrc( $uploadInfo['src'] );
 809                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 810                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 811                                 !empty( $uploadInfo['isTempSrc'] ) );
 812                 }
 813                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 814                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 815                 }
 816                 $revision->setSize( intval( $uploadInfo['size'] ) );
 817                 $revision->setComment( $uploadInfo['comment'] );
 818
 819                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 820                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 821                 }
 822                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 823                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 824                 }
 825                 $revision->setNoUpdates( $this->mNoUpdates );
 826
 827                 return call_user_func( $this->mUploadCallback, $revision );
 828         }
 829
 830         /**
 831          * @return array
 832          */
 833         private function handleContributor() {
 834                 $fields = array( 'id', 'ip', 'username' );
 835                 $info = array();
 836
 837                 while ( $this->reader->read() ) {
 838                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 839                                         $this->reader->name == 'contributor' ) {
 840                                 break;
 841                         }
 842
 843                         $tag = $this->reader->name;
 844
 845                         if ( in_array( $tag, $fields ) ) {
 846                                 $info[$tag] = $this->nodeContents();
 847                         }
 848                 }
 849
 850                 return $info;
 851         }
 852
 853         /**
 854          * @param string $text
 855          * @return array|bool
 856          */
 857         private function processTitle( $text ) {
 858                 $workTitle = $text;
 859                 $origTitle = Title::newFromText( $workTitle );
 860
 861                 if ( !is_null( $this->mTargetNamespace ) && !is_null( $origTitle ) ) {
 862                         # makeTitleSafe, because $origTitle can have a interwiki (different setting of interwiki map)
 863                         # and than dbKey can begin with a lowercase char
 864                         $title = Title::makeTitleSafe( $this->mTargetNamespace,
 865                                 $origTitle->getDBkey() );
 866                 } else {
 867                         if ( !is_null( $this->mTargetRootPage ) ) {
 868                                 $workTitle = $this->mTargetRootPage . '/' . $workTitle;
 869                         }
 870                         $title = Title::newFromText( $workTitle );
 871                 }
 872
 873                 $commandLineMode = $this->config->get( 'CommandLineMode' );
 874                 if ( is_null( $title ) ) {
 875                         # Invalid page title? Ignore the page
 876                         $this->notice( 'import-error-invalid', $workTitle );
 877                         return false;
 878                 } elseif ( $title->isExternal() ) {
 879                         $this->notice( 'import-error-interwiki', $title->getPrefixedText() );
 880                         return false;
 881                 } elseif ( !$title->canExist() ) {
 882                         $this->notice( 'import-error-special', $title->getPrefixedText() );
 883                         return false;
 884                 } elseif ( !$title->userCan( 'edit' ) && !$commandLineMode ) {
 885                         # Do not import if the importing wiki user cannot edit this page
 886                         $this->notice( 'import-error-edit', $title->getPrefixedText() );
 887                         return false;
 888                 } elseif ( !$title->exists() && !$title->userCan( 'create' ) && !$commandLineMode ) {
 889                         # Do not import if the importing wiki user cannot create this page
 890                         $this->notice( 'import-error-create', $title->getPrefixedText() );
 891                         return false;
 892                 }
 893
 894                 return array( $title, $origTitle );
 895         }
 896 }
 897
 898 /** This is a horrible hack used to keep source compatibility */
 899 class UploadSourceAdapter {
 900         /** @var array */
 901         public static $sourceRegistrations = array();
 902
 903         /** @var string */
 904         private $mSource;
 905
 906         /** @var string */
 907         private $mBuffer;
 908
 909         /** @var int */
 910         private $mPosition;
 911
 912         /**
 913          * @param ImportStreamSource $source
 914          * @return string
 915          */
 916         static function registerSource( ImportStreamSource $source ) {
 917                 $id = wfRandomString();
 918
 919                 self::$sourceRegistrations[$id] = $source;
 920
 921                 return $id;
 922         }
 923
 924         /**
 925          * @param string $path
 926          * @param string $mode
 927          * @param array $options
 928          * @param string $opened_path
 929          * @return bool
 930          */
 931         function stream_open( $path, $mode, $options, &$opened_path ) {
 932                 $url = parse_url( $path );
 933                 $id = $url['host'];
 934
 935                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
 936                         return false;
 937                 }
 938
 939                 $this->mSource = self::$sourceRegistrations[$id];
 940
 941                 return true;
 942         }
 943
 944         /**
 945          * @param int $count
 946          * @return string
 947          */
 948         function stream_read( $count ) {
 949                 $return = '';
 950                 $leave = false;
 951
 952                 while ( !$leave && !$this->mSource->atEnd() &&
 953                                 strlen( $this->mBuffer ) < $count ) {
 954                         $read = $this->mSource->readChunk();
 955
 956                         if ( !strlen( $read ) ) {
 957                                 $leave = true;
 958                         }
 959
 960                         $this->mBuffer .= $read;
 961                 }
 962
 963                 if ( strlen( $this->mBuffer ) ) {
 964                         $return = substr( $this->mBuffer, 0, $count );
 965                         $this->mBuffer = substr( $this->mBuffer, $count );
 966                 }
 967
 968                 $this->mPosition += strlen( $return );
 969
 970                 return $return;
 971         }
 972
 973         /**
 974          * @param string $data
 975          * @return bool
 976          */
 977         function stream_write( $data ) {
 978                 return false;
 979         }
 980
 981         /**
 982          * @return mixed
 983          */
 984         function stream_tell() {
 985                 return $this->mPosition;
 986         }
 987
 988         /**
 989          * @return bool
 990          */
 991         function stream_eof() {
 992                 return $this->mSource->atEnd();
 993         }
 994
 995         /**
 996          * @return array
 997          */
 998         function url_stat() {
 999                 $result = array();
1000
1001                 $result['dev'] = $result[0] = 0;
1002                 $result['ino'] = $result[1] = 0;
1003                 $result['mode'] = $result[2] = 0;
1004                 $result['nlink'] = $result[3] = 0;
1005                 $result['uid'] = $result[4] = 0;
1006                 $result['gid'] = $result[5] = 0;
1007                 $result['rdev'] = $result[6] = 0;
1008                 $result['size'] = $result[7] = 0;
1009                 $result['atime'] = $result[8] = 0;
1010                 $result['mtime'] = $result[9] = 0;
1011                 $result['ctime'] = $result[10] = 0;
1012                 $result['blksize'] = $result[11] = 0;
1013                 $result['blocks'] = $result[12] = 0;
1014
1015                 return $result;
1016         }
1017 }
1018
1019 /**
1020  * @todo document (e.g. one-sentence class description).
1021  * @ingroup SpecialPage
1022  */
1023 class WikiRevision {
1024         /** @todo Unused? */
1025         public $importer = null;
1026
1027         /** @var Title */
1028         public $title = null;
1029
1030         /** @var int */
1031         public $id = 0;
1032
1033         /** @var string */
1034         public $timestamp = "20010115000000";
1035
1036         /**
1037          * @var int
1038          * @todo Can't find any uses. Public, because that's suspicious. Get clarity. */
1039         public $user = 0;
1040
1041         /** @var string */
1042         public $user_text = "";
1043
1044         /** @var string */
1045         public $model = null;
1046
1047         /** @var string */
1048         public $format = null;
1049
1050         /** @var string */
1051         public $text = "";
1052
1053         /** @var int */
1054         protected $size;
1055
1056         /** @var Content */
1057         public $content = null;
1058
1059         /** @var ContentHandler */
1060         protected $contentHandler = null;
1061
1062         /** @var string */
1063         public $comment = "";
1064
1065         /** @var bool */
1066         public $minor = false;
1067
1068         /** @var string */
1069         public $type = "";
1070
1071         /** @var string */
1072         public $action = "";
1073
1074         /** @var string */
1075         public $params = "";
1076
1077         /** @var string */
1078         public $fileSrc = '';
1079
1080         /** @var bool|string */
1081         public $sha1base36 = false;
1082
1083         /**
1084          * @var bool
1085          * @todo Unused?
1086          */
1087         public $isTemp = false;
1088
1089         /** @var string */
1090         public $archiveName = '';
1091
1092         protected $filename;
1093
1094         /** @var mixed */
1095         protected $src;
1096
1097         /** @todo Unused? */
1098         public $fileIsTemp;
1099
1100         /** @var bool */
1101         private $mNoUpdates = false;
1102
1103         /** @var Config $config */
1104         private $config;
1105
1106         public function __construct( Config $config ) {
1107                 $this->config = $config;
1108         }
1109
1110         /**
1111          * @param Title $title
1112          * @throws MWException
1113          */
1114         function setTitle( $title ) {
1115                 if ( is_object( $title ) ) {
1116                         $this->title = $title;
1117                 } elseif ( is_null( $title ) ) {
1118                         throw new MWException( "WikiRevision given a null title in import. "
1119                                 . "You may need to adjust \$wgLegalTitleChars." );
1120                 } else {
1121                         throw new MWException( "WikiRevision given non-object title in import." );
1122                 }
1123         }
1124
1125         /**
1126          * @param int $id
1127          */
1128         function setID( $id ) {
1129                 $this->id = $id;
1130         }
1131
1132         /**
1133          * @param string $ts
1134          */
1135         function setTimestamp( $ts ) {
1136                 # 2003-08-05T18:30:02Z
1137                 $this->timestamp = wfTimestamp( TS_MW, $ts );
1138         }
1139
1140         /**
1141          * @param string $user
1142          */
1143         function setUsername( $user ) {
1144                 $this->user_text = $user;
1145         }
1146
1147         /**
1148          * @param string $ip
1149          */
1150         function setUserIP( $ip ) {
1151                 $this->user_text = $ip;
1152         }
1153
1154         /**
1155          * @param string $model
1156          */
1157         function setModel( $model ) {
1158                 $this->model = $model;
1159         }
1160
1161         /**
1162          * @param string $format
1163          */
1164         function setFormat( $format ) {
1165                 $this->format = $format;
1166         }
1167
1168         /**
1169          * @param string $text
1170          */
1171         function setText( $text ) {
1172                 $this->text = $text;
1173         }
1174
1175         /**
1176          * @param string $text
1177          */
1178         function setComment( $text ) {
1179                 $this->comment = $text;
1180         }
1181
1182         /**
1183          * @param bool $minor
1184          */
1185         function setMinor( $minor ) {
1186                 $this->minor = (bool)$minor;
1187         }
1188
1189         /**
1190          * @param mixed $src
1191          */
1192         function setSrc( $src ) {
1193                 $this->src = $src;
1194         }
1195
1196         /**
1197          * @param string $src
1198          * @param bool $isTemp
1199          */
1200         function setFileSrc( $src, $isTemp ) {
1201                 $this->fileSrc = $src;
1202                 $this->fileIsTemp = $isTemp;
1203         }
1204
1205         /**
1206          * @param string $sha1base36
1207          */
1208         function setSha1Base36( $sha1base36 ) {
1209                 $this->sha1base36 = $sha1base36;
1210         }
1211
1212         /**
1213          * @param string $filename
1214          */
1215         function setFilename( $filename ) {
1216                 $this->filename = $filename;
1217         }
1218
1219         /**
1220          * @param string $archiveName
1221          */
1222         function setArchiveName( $archiveName ) {
1223                 $this->archiveName = $archiveName;
1224         }
1225
1226         /**
1227          * @param int $size
1228          */
1229         function setSize( $size ) {
1230                 $this->size = intval( $size );
1231         }
1232
1233         /**
1234          * @param string $type
1235          */
1236         function setType( $type ) {
1237                 $this->type = $type;
1238         }
1239
1240         /**
1241          * @param string $action
1242          */
1243         function setAction( $action ) {
1244                 $this->action = $action;
1245         }
1246
1247         /**
1248          * @param array $params
1249          */
1250         function setParams( $params ) {
1251                 $this->params = $params;
1252         }
1253
1254         /**
1255          * @param bool $noupdates
1256          */
1257         public function setNoUpdates( $noupdates ) {
1258                 $this->mNoUpdates = $noupdates;
1259         }
1260
1261         /**
1262          * @return Title
1263          */
1264         function getTitle() {
1265                 return $this->title;
1266         }
1267
1268         /**
1269          * @return int
1270          */
1271         function getID() {
1272                 return $this->id;
1273         }
1274
1275         /**
1276          * @return string
1277          */
1278         function getTimestamp() {
1279                 return $this->timestamp;
1280         }
1281
1282         /**
1283          * @return string
1284          */
1285         function getUser() {
1286                 return $this->user_text;
1287         }
1288
1289         /**
1290          * @return string
1291          *
1292          * @deprecated Since 1.21, use getContent() instead.
1293          */
1294         function getText() {
1295                 ContentHandler::deprecated( __METHOD__, '1.21' );
1296
1297                 return $this->text;
1298         }
1299
1300         /**
1301          * @return ContentHandler
1302          */
1303         function getContentHandler() {
1304                 if ( is_null( $this->contentHandler ) ) {
1305                         $this->contentHandler = ContentHandler::getForModelID( $this->getModel() );
1306                 }
1307
1308                 return $this->contentHandler;
1309         }
1310
1311         /**
1312          * @return Content
1313          */
1314         function getContent() {
1315                 if ( is_null( $this->content ) ) {
1316                         $handler = $this->getContentHandler();
1317                         $this->content = $handler->unserializeContent( $this->text, $this->getFormat() );
1318                 }
1319
1320                 return $this->content;
1321         }
1322
1323         /**
1324          * @return string
1325          */
1326         function getModel() {
1327                 if ( is_null( $this->model ) ) {
1328                         $this->model = $this->getTitle()->getContentModel();
1329                 }
1330
1331                 return $this->model;
1332         }
1333
1334         /**
1335          * @return string
1336          */
1337         function getFormat() {
1338                 if ( is_null( $this->format ) ) {
1339                         $this->format = $this->getContentHandler()->getDefaultFormat();
1340                 }
1341
1342                 return $this->format;
1343         }
1344
1345         /**
1346          * @return string
1347          */
1348         function getComment() {
1349                 return $this->comment;
1350         }
1351
1352         /**
1353          * @return bool
1354          */
1355         function getMinor() {
1356                 return $this->minor;
1357         }
1358
1359         /**
1360          * @return mixed
1361          */
1362         function getSrc() {
1363                 return $this->src;
1364         }
1365
1366         /**
1367          * @return bool|string
1368          */
1369         function getSha1() {
1370                 if ( $this->sha1base36 ) {
1371                         return wfBaseConvert( $this->sha1base36, 36, 16 );
1372                 }
1373                 return false;
1374         }
1375
1376         /**
1377          * @return string
1378          */
1379         function getFileSrc() {
1380                 return $this->fileSrc;
1381         }
1382
1383         /**
1384          * @return bool
1385          */
1386         function isTempSrc() {
1387                 return $this->isTemp;
1388         }
1389
1390         /**
1391          * @return mixed
1392          */
1393         function getFilename() {
1394                 return $this->filename;
1395         }
1396
1397         /**
1398          * @return string
1399          */
1400         function getArchiveName() {
1401                 return $this->archiveName;
1402         }
1403
1404         /**
1405          * @return mixed
1406          */
1407         function getSize() {
1408                 return $this->size;
1409         }
1410
1411         /**
1412          * @return string
1413          */
1414         function getType() {
1415                 return $this->type;
1416         }
1417
1418         /**
1419          * @return string
1420          */
1421         function getAction() {
1422                 return $this->action;
1423         }
1424
1425         /**
1426          * @return string
1427          */
1428         function getParams() {
1429                 return $this->params;
1430         }
1431
1432         /**
1433          * @return bool
1434          */
1435         function importOldRevision() {
1436                 $dbw = wfGetDB( DB_MASTER );
1437
1438                 # Sneak a single revision into place
1439                 $user = User::newFromName( $this->getUser() );
1440                 if ( $user ) {
1441                         $userId = intval( $user->getId() );
1442                         $userText = $user->getName();
1443                         $userObj = $user;
1444                 } else {
1445                         $userId = 0;
1446                         $userText = $this->getUser();
1447                         $userObj = new User;
1448                 }
1449
1450                 // avoid memory leak...?
1451                 $linkCache = LinkCache::singleton();
1452                 $linkCache->clear();
1453
1454                 $page = WikiPage::factory( $this->title );
1455                 $page->loadPageData( 'fromdbmaster' );
1456                 if ( !$page->exists() ) {
1457                         # must create the page...
1458                         $pageId = $page->insertOn( $dbw );
1459                         $created = true;
1460                         $oldcountable = null;
1461                 } else {
1462                         $pageId = $page->getId();
1463                         $created = false;
1464
1465                         $prior = $dbw->selectField( 'revision', '1',
1466                                 array( 'rev_page' => $pageId,
1467                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1468                                         'rev_user_text' => $userText,
1469                                         'rev_comment' => $this->getComment() ),
1470                                 __METHOD__
1471                         );
1472                         if ( $prior ) {
1473                                 // @todo FIXME: This could fail slightly for multiple matches :P
1474                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1475                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1476                                 return false;
1477                         }
1478                         $oldcountable = $page->isCountable();
1479                 }
1480
1481                 # @todo FIXME: Use original rev_id optionally (better for backups)
1482                 # Insert the row
1483                 $revision = new Revision( array(
1484                         'title' => $this->title,
1485                         'page' => $pageId,
1486                         'content_model' => $this->getModel(),
1487                         'content_format' => $this->getFormat(),
1488                         //XXX: just set 'content' => $this->getContent()?
1489                         'text' => $this->getContent()->serialize( $this->getFormat() ),
1490                         'comment' => $this->getComment(),
1491                         'user' => $userId,
1492                         'user_text' => $userText,
1493                         'timestamp' => $this->timestamp,
1494                         'minor_edit' => $this->minor,
1495                         ) );
1496                 $revision->insertOn( $dbw );
1497                 $changed = $page->updateIfNewerOn( $dbw, $revision );
1498
1499                 if ( $changed !== false && !$this->mNoUpdates ) {
1500                         wfDebug( __METHOD__ . ": running updates\n" );
1501                         $page->doEditUpdates(
1502                                 $revision,
1503                                 $userObj,
1504                                 array( 'created' => $created, 'oldcountable' => $oldcountable )
1505                         );
1506                 }
1507
1508                 return true;
1509         }
1510
1511         function importLogItem() {
1512                 $dbw = wfGetDB( DB_MASTER );
1513                 # @todo FIXME: This will not record autoblocks
1514                 if ( !$this->getTitle() ) {
1515                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1516                                 $this->timestamp . "\n" );
1517                         return;
1518                 }
1519                 # Check if it exists already
1520                 // @todo FIXME: Use original log ID (better for backups)
1521                 $prior = $dbw->selectField( 'logging', '1',
1522                         array( 'log_type' => $this->getType(),
1523                                 'log_action' => $this->getAction(),
1524                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1525                                 'log_namespace' => $this->getTitle()->getNamespace(),
1526                                 'log_title' => $this->getTitle()->getDBkey(),
1527                                 'log_comment' => $this->getComment(),
1528                                 #'log_user_text' => $this->user_text,
1529                                 'log_params' => $this->params ),
1530                         __METHOD__
1531                 );
1532                 // @todo FIXME: This could fail slightly for multiple matches :P
1533                 if ( $prior ) {
1534                         wfDebug( __METHOD__
1535                                 . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp "
1536                                 . $this->timestamp . "\n" );
1537                         return;
1538                 }
1539                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1540                 $data = array(
1541                         'log_id' => $log_id,
1542                         'log_type' => $this->type,
1543                         'log_action' => $this->action,
1544                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1545                         'log_user' => User::idFromName( $this->user_text ),
1546                         #'log_user_text' => $this->user_text,
1547                         'log_namespace' => $this->getTitle()->getNamespace(),
1548                         'log_title' => $this->getTitle()->getDBkey(),
1549                         'log_comment' => $this->getComment(),
1550                         'log_params' => $this->params
1551                 );
1552                 $dbw->insert( 'logging', $data, __METHOD__ );
1553         }
1554
1555         /**
1556          * @return bool
1557          */
1558         function importUpload() {
1559                 # Construct a file
1560                 $archiveName = $this->getArchiveName();
1561                 if ( $archiveName ) {
1562                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1563                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1564                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1565                 } else {
1566                         $file = wfLocalFile( $this->getTitle() );
1567                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1568                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1569                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1570                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1571                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1572                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1573                         }
1574                 }
1575                 if ( !$file ) {
1576                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1577                         return false;
1578                 }
1579
1580                 # Get the file source or download if necessary
1581                 $source = $this->getFileSrc();
1582                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1583                 if ( !$source ) {
1584                         $source = $this->downloadSource();
1585                         $flags |= File::DELETE_SOURCE;
1586                 }
1587                 if ( !$source ) {
1588                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1589                         return false;
1590                 }
1591                 $sha1 = $this->getSha1();
1592                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1593                         if ( $flags & File::DELETE_SOURCE ) {
1594                                 # Broken file; delete it if it is a temporary file
1595                                 unlink( $source );
1596                         }
1597                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1598                         return false;
1599                 }
1600
1601                 $user = User::newFromName( $this->user_text );
1602
1603                 # Do the actual upload
1604                 if ( $archiveName ) {
1605                         $status = $file->uploadOld( $source, $archiveName,
1606                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1607                 } else {
1608                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1609                                 $flags, false, $this->getTimestamp(), $user );
1610                 }
1611
1612                 if ( $status->isGood() ) {
1613                         wfDebug( __METHOD__ . ": Successful\n" );
1614                         return true;
1615                 } else {
1616                         wfDebug( __METHOD__ . ': failed: ' . $status->getXml() . "\n" );
1617                         return false;
1618                 }
1619         }
1620
1621         /**
1622          * @return bool|string
1623          */
1624         function downloadSource() {
1625                 if ( !$this->config->get( 'EnableUploads' ) ) {
1626                         return false;
1627                 }
1628
1629                 $tempo = tempnam( wfTempDir(), 'download' );
1630                 $f = fopen( $tempo, 'wb' );
1631                 if ( !$f ) {
1632                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1633                         return false;
1634                 }
1635
1636                 // @todo FIXME!
1637                 $src = $this->getSrc();
1638                 $data = Http::get( $src );
1639                 if ( !$data ) {
1640                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1641                         fclose( $f );
1642                         unlink( $tempo );
1643                         return false;
1644                 }
1645
1646                 fwrite( $f, $data );
1647                 fclose( $f );
1648
1649                 return $tempo;
1650         }
1651
1652 }
1653
1654 /**
1655  * @todo document (e.g. one-sentence class description).
1656  * @ingroup SpecialPage
1657  */
1658 class ImportStringSource {
1659         function __construct( $string ) {
1660                 $this->mString = $string;
1661                 $this->mRead = false;
1662         }
1663
1664         /**
1665          * @return bool
1666          */
1667         function atEnd() {
1668                 return $this->mRead;
1669         }
1670
1671         /**
1672          * @return bool|string
1673          */
1674         function readChunk() {
1675                 if ( $this->atEnd() ) {
1676                         return false;
1677                 }
1678                 $this->mRead = true;
1679                 return $this->mString;
1680         }
1681 }
1682
1683 /**
1684  * @todo document (e.g. one-sentence class description).
1685  * @ingroup SpecialPage
1686  */
1687 class ImportStreamSource {
1688         function __construct( $handle ) {
1689                 $this->mHandle = $handle;
1690         }
1691
1692         /**
1693          * @return bool
1694          */
1695         function atEnd() {
1696                 return feof( $this->mHandle );
1697         }
1698
1699         /**
1700          * @return string
1701          */
1702         function readChunk() {
1703                 return fread( $this->mHandle, 32768 );
1704         }
1705
1706         /**
1707          * @param string $filename
1708          * @return Status
1709          */
1710         static function newFromFile( $filename ) {
1711                 wfSuppressWarnings();
1712                 $file = fopen( $filename, 'rt' );
1713                 wfRestoreWarnings();
1714                 if ( !$file ) {
1715                         return Status::newFatal( "importcantopen" );
1716                 }
1717                 return Status::newGood( new ImportStreamSource( $file ) );
1718         }
1719
1720         /**
1721          * @param string $fieldname
1722          * @return Status
1723          */
1724         static function newFromUpload( $fieldname = "xmlimport" ) {
1725                 $upload =& $_FILES[$fieldname];
1726
1727                 if ( $upload === null || !$upload['name'] ) {
1728                         return Status::newFatal( 'importnofile' );
1729                 }
1730                 if ( !empty( $upload['error'] ) ) {
1731                         switch ( $upload['error'] ) {
1732                                 case 1:
1733                                         # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1734                                         return Status::newFatal( 'importuploaderrorsize' );
1735                                 case 2:
1736                                         # The uploaded file exceeds the MAX_FILE_SIZE directive that
1737                                         # was specified in the HTML form.
1738                                         return Status::newFatal( 'importuploaderrorsize' );
1739                                 case 3:
1740                                         # The uploaded file was only partially uploaded
1741                                         return Status::newFatal( 'importuploaderrorpartial' );
1742                                 case 6:
1743                                         # Missing a temporary folder.
1744                                         return Status::newFatal( 'importuploaderrortemp' );
1745                                 # case else: # Currently impossible
1746                         }
1747
1748                 }
1749                 $fname = $upload['tmp_name'];
1750                 if ( is_uploaded_file( $fname ) ) {
1751                         return ImportStreamSource::newFromFile( $fname );
1752                 } else {
1753                         return Status::newFatal( 'importnofile' );
1754                 }
1755         }
1756
1757         /**
1758          * @param string $url
1759          * @param string $method
1760          * @return Status
1761          */
1762         static function newFromURL( $url, $method = 'GET' ) {
1763                 wfDebug( __METHOD__ . ": opening $url\n" );
1764                 # Use the standard HTTP fetch function; it times out
1765                 # quicker and sorts out user-agent problems which might
1766                 # otherwise prevent importing from large sites, such
1767                 # as the Wikimedia cluster, etc.
1768                 $data = Http::request( $method, $url, array( 'followRedirects' => true ) );
1769                 if ( $data !== false ) {
1770                         $file = tmpfile();
1771                         fwrite( $file, $data );
1772                         fflush( $file );
1773                         fseek( $file, 0 );
1774                         return Status::newGood( new ImportStreamSource( $file ) );
1775                 } else {
1776                         return Status::newFatal( 'importcantopen' );
1777                 }
1778         }
1779
1780         /**
1781          * @param string $interwiki
1782          * @param string $page
1783          * @param bool $history
1784          * @param bool $templates
1785          * @param int $pageLinkDepth
1786          * @return Status
1787          */
1788         public static function newFromInterwiki( $interwiki, $page, $history = false,
1789                 $templates = false, $pageLinkDepth = 0
1790         ) {
1791                 if ( $page == '' ) {
1792                         return Status::newFatal( 'import-noarticle' );
1793                 }
1794                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1795                 if ( is_null( $link ) || !$link->isExternal() ) {
1796                         return Status::newFatal( 'importbadinterwiki' );
1797                 } else {
1798                         $params = array();
1799                         if ( $history ) {
1800                                 $params['history'] = 1;
1801                         }
1802                         if ( $templates ) {
1803                                 $params['templates'] = 1;
1804                         }
1805                         if ( $pageLinkDepth ) {
1806                                 $params['pagelink-depth'] = $pageLinkDepth;
1807                         }
1808                         $url = $link->getFullURL( $params );
1809                         # For interwikis, use POST to avoid redirects.
1810                         return ImportStreamSource::newFromURL( $url, "POST" );
1811                 }
1812         }
1813 }