includes/Import.php

   1 <?php
   2 /**
   3  * MediaWiki page data importer
   4  *
   5  * Copyright © 2003,2005 Brion Vibber <brion@pobox.com>
   6  * http://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, $mPageOutCallback;
  37         private $mDebug;
  38         private $mImportUploads, $mImageBasePath;
  39
  40         /**
  41          * Creates an ImportXMLReader drawing from the source provided
  42          */
  43         function __construct( $source ) {
  44                 $this->reader = new XMLReader();
  45
  46                 stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  47                 $id = UploadSourceAdapter::registerSource( $source );
  48                 $this->reader->open( "uploadsource://$id" );
  49
  50                 // Default callbacks
  51                 $this->setRevisionCallback( array( $this, "importRevision" ) );
  52                 $this->setUploadCallback( array( $this, 'importUpload' ) );
  53                 $this->setLogItemCallback( array( $this, 'importLogItem' ) );
  54                 $this->setPageOutCallback( array( $this, 'finishImportPage' ) );
  55         }
  56
  57         private function throwXmlError( $err ) {
  58                 $this->debug( "FAILURE: $err" );
  59                 wfDebug( "WikiImporter XML error: $err\n" );
  60         }
  61
  62         private function debug( $data ) {
  63                 if( $this->mDebug ) {
  64                         wfDebug( "IMPORT: $data\n" );
  65                 }
  66         }
  67
  68         private function warn( $data ) {
  69                 wfDebug( "IMPORT: $data\n" );
  70         }
  71
  72         private function notice( $data ) {
  73                 global $wgCommandLineMode;
  74                 if( $wgCommandLineMode ) {
  75                         print "$data\n";
  76                 } else {
  77                         global $wgOut;
  78                         $wgOut->addHTML( "<li>" . htmlspecialchars( $data ) . "</li>\n" );
  79                 }
  80         }
  81
  82         /**
  83          * Set debug mode...
  84          */
  85         function setDebug( $debug ) {
  86                 $this->mDebug = $debug;
  87         }
  88
  89         /**
  90          * Sets the action to perform as each new page in the stream is reached.
  91          * @param $callback callback
  92          * @return callback
  93          */
  94         public function setPageCallback( $callback ) {
  95                 $previous = $this->mPageCallback;
  96                 $this->mPageCallback = $callback;
  97                 return $previous;
  98         }
  99
 100         /**
 101          * Sets the action to perform as each page in the stream is completed.
 102          * Callback accepts the page title (as a Title object), a second object
 103          * with the original title form (in case it's been overridden into a
 104          * local namespace), and a count of revisions.
 105          *
 106          * @param $callback callback
 107          * @return callback
 108          */
 109         public function setPageOutCallback( $callback ) {
 110                 $previous = $this->mPageOutCallback;
 111                 $this->mPageOutCallback = $callback;
 112                 return $previous;
 113         }
 114
 115         /**
 116          * Sets the action to perform as each page revision is reached.
 117          * @param $callback callback
 118          * @return callback
 119          */
 120         public function setRevisionCallback( $callback ) {
 121                 $previous = $this->mRevisionCallback;
 122                 $this->mRevisionCallback = $callback;
 123                 return $previous;
 124         }
 125
 126         /**
 127          * Sets the action to perform as each file upload version is reached.
 128          * @param $callback callback
 129          * @return callback
 130          */
 131         public function setUploadCallback( $callback ) {
 132                 $previous = $this->mUploadCallback;
 133                 $this->mUploadCallback = $callback;
 134                 return $previous;
 135         }
 136
 137         /**
 138          * Sets the action to perform as each log item reached.
 139          * @param $callback callback
 140          * @return callback
 141          */
 142         public function setLogItemCallback( $callback ) {
 143                 $previous = $this->mLogItemCallback;
 144                 $this->mLogItemCallback = $callback;
 145                 return $previous;
 146         }
 147
 148         /**
 149          * Sets the action to perform when site info is encountered
 150          * @param $callback callback
 151          * @return callback
 152          */
 153         public function setSiteInfoCallback( $callback ) {
 154                 $previous = $this->mSiteInfoCallback;
 155                 $this->mSiteInfoCallback = $callback;
 156                 return $previous;
 157         }
 158
 159         /**
 160          * Set a target namespace to override the defaults
 161          */
 162         public function setTargetNamespace( $namespace ) {
 163                 if( is_null( $namespace ) ) {
 164                         // Don't override namespaces
 165                         $this->mTargetNamespace = null;
 166                 } elseif( $namespace >= 0 ) {
 167                         // @todo FIXME: Check for validity
 168                         $this->mTargetNamespace = intval( $namespace );
 169                 } else {
 170                         return false;
 171                 }
 172         }
 173
 174         /**
 175          *
 176          */
 177         public function setImageBasePath( $dir ) {
 178                 $this->mImageBasePath = $dir;
 179         }
 180         public function setImportUploads( $import ) {
 181                 $this->mImportUploads = $import;
 182         }
 183
 184         /**
 185          * Default per-revision callback, performs the import.
 186          * @param $revision WikiRevision
 187          */
 188         public function importRevision( $revision ) {
 189                 $dbw = wfGetDB( DB_MASTER );
 190                 return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
 191         }
 192
 193         /**
 194          * Default per-revision callback, performs the import.
 195          * @param $rev WikiRevision
 196          */
 197         public function importLogItem( $rev ) {
 198                 $dbw = wfGetDB( DB_MASTER );
 199                 return $dbw->deadlockLoop( array( $rev, 'importLogItem' ) );
 200         }
 201
 202         /**
 203          * Dummy for now...
 204          */
 205         public function importUpload( $revision ) {
 206                 $dbw = wfGetDB( DB_MASTER );
 207                 return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
 208         }
 209
 210         /**
 211          * Mostly for hook use
 212          */
 213         public function finishImportPage( $title, $origTitle, $revCount, $sRevCount, $pageInfo ) {
 214                 $args = func_get_args();
 215                 return wfRunHooks( 'AfterImportPage', $args );
 216         }
 217
 218         /**
 219          * Alternate per-revision callback, for debugging.
 220          * @param $revision WikiRevision
 221          */
 222         public function debugRevisionHandler( &$revision ) {
 223                 $this->debug( "Got revision:" );
 224                 if( is_object( $revision->title ) ) {
 225                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 226                 } else {
 227                         $this->debug( "-- Title: <invalid>" );
 228                 }
 229                 $this->debug( "-- User: " . $revision->user_text );
 230                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 231                 $this->debug( "-- Comment: " . $revision->comment );
 232                 $this->debug( "-- Text: " . $revision->text );
 233         }
 234
 235         /**
 236          * Notify the callback function when a new <page> is reached.
 237          * @param $title Title
 238          */
 239         function pageCallback( $title ) {
 240                 if( isset( $this->mPageCallback ) ) {
 241                         call_user_func( $this->mPageCallback, $title );
 242                 }
 243         }
 244
 245         /**
 246          * Notify the callback function when a </page> is closed.
 247          * @param $title Title
 248          * @param $origTitle Title
 249          * @param $revCount Integer
 250          * @param $sucCount Int: number of revisions for which callback returned true
 251          * @param $pageInfo Array: associative array of page information
 252          */
 253         private function pageOutCallback( $title, $origTitle, $revCount, $sucCount, $pageInfo ) {
 254                 if( isset( $this->mPageOutCallback ) ) {
 255                         $args = func_get_args();
 256                         call_user_func_array( $this->mPageOutCallback, $args );
 257                 }
 258         }
 259
 260         /**
 261          * Notify the callback function of a revision
 262          * @param $revision A WikiRevision object
 263          */
 264         private function revisionCallback( $revision ) {
 265                 if ( isset( $this->mRevisionCallback ) ) {
 266                         return call_user_func_array( $this->mRevisionCallback,
 267                                         array( $revision, $this ) );
 268                 } else {
 269                         return false;
 270                 }
 271         }
 272
 273         /**
 274          * Notify the callback function of a new log item
 275          * @param $revision A WikiRevision object
 276          */
 277         private function logItemCallback( $revision ) {
 278                 if ( isset( $this->mLogItemCallback ) ) {
 279                         return call_user_func_array( $this->mLogItemCallback,
 280                                         array( $revision, $this ) );
 281                 } else {
 282                         return false;
 283                 }
 284         }
 285
 286         /**
 287          * Shouldn't something like this be built-in to XMLReader?
 288          * Fetches text contents of the current element, assuming
 289          * no sub-elements or such scary things.
 290          * @return string
 291          * @access private
 292          */
 293         private function nodeContents() {
 294                 if( $this->reader->isEmptyElement ) {
 295                         return "";
 296                 }
 297                 $buffer = "";
 298                 while( $this->reader->read() ) {
 299                         switch( $this->reader->nodeType ) {
 300                         case XmlReader::TEXT:
 301                         case XmlReader::SIGNIFICANT_WHITESPACE:
 302                                 $buffer .= $this->reader->value;
 303                                 break;
 304                         case XmlReader::END_ELEMENT:
 305                                 return $buffer;
 306                         }
 307                 }
 308
 309                 $this->reader->close();
 310                 return '';
 311         }
 312
 313         # --------------
 314
 315         /** Left in for debugging */
 316         private function dumpElement() {
 317                 static $lookup = null;
 318                 if (!$lookup) {
 319                         $xmlReaderConstants = array(
 320                                 "NONE",
 321                                 "ELEMENT",
 322                                 "ATTRIBUTE",
 323                                 "TEXT",
 324                                 "CDATA",
 325                                 "ENTITY_REF",
 326                                 "ENTITY",
 327                                 "PI",
 328                                 "COMMENT",
 329                                 "DOC",
 330                                 "DOC_TYPE",
 331                                 "DOC_FRAGMENT",
 332                                 "NOTATION",
 333                                 "WHITESPACE",
 334                                 "SIGNIFICANT_WHITESPACE",
 335                                 "END_ELEMENT",
 336                                 "END_ENTITY",
 337                                 "XML_DECLARATION",
 338                                 );
 339                         $lookup = array();
 340
 341                         foreach( $xmlReaderConstants as $name ) {
 342                                 $lookup[constant("XmlReader::$name")] = $name;
 343                         }
 344                 }
 345
 346                 print( var_dump(
 347                         $lookup[$this->reader->nodeType],
 348                         $this->reader->name,
 349                         $this->reader->value
 350                 )."\n\n" );
 351         }
 352
 353         /**
 354          * Primary entry point
 355          */
 356         public function doImport() {
 357                 $this->reader->read();
 358
 359                 if ( $this->reader->name != 'mediawiki' ) {
 360                         throw new MWException( "Expected <mediawiki> tag, got ".
 361                                 $this->reader->name );
 362                 }
 363                 $this->debug( "<mediawiki> tag is correct." );
 364
 365                 $this->debug( "Starting primary dump processing loop." );
 366
 367                 $keepReading = $this->reader->read();
 368                 $skip = false;
 369                 while ( $keepReading ) {
 370                         $tag = $this->reader->name;
 371                         $type = $this->reader->nodeType;
 372
 373                         if ( !wfRunHooks( 'ImportHandleToplevelXMLTag', $this ) ) {
 374                                 // Do nothing
 375                         } elseif ( $tag == 'mediawiki' && $type == XmlReader::END_ELEMENT ) {
 376                                 break;
 377                         } elseif ( $tag == 'siteinfo' ) {
 378                                 $this->handleSiteInfo();
 379                         } elseif ( $tag == 'page' ) {
 380                                 $this->handlePage();
 381                         } elseif ( $tag == 'logitem' ) {
 382                                 $this->handleLogItem();
 383                         } elseif ( $tag != '#text' ) {
 384                                 $this->warn( "Unhandled top-level XML tag $tag" );
 385
 386                                 $skip = true;
 387                         }
 388
 389                         if ($skip) {
 390                                 $keepReading = $this->reader->next();
 391                                 $skip = false;
 392                                 $this->debug( "Skip" );
 393                         } else {
 394                                 $keepReading = $this->reader->read();
 395                         }
 396                 }
 397
 398                 return true;
 399         }
 400
 401         private function handleSiteInfo() {
 402                 // Site info is useful, but not actually used for dump imports.
 403                 // Includes a quick short-circuit to save performance.
 404                 if ( ! $this->mSiteInfoCallback ) {
 405                         $this->reader->next();
 406                         return true;
 407                 }
 408                 throw new MWException( "SiteInfo tag is not yet handled, do not set mSiteInfoCallback" );
 409         }
 410
 411         private function handleLogItem() {
 412                 $this->debug( "Enter log item handler." );
 413                 $logInfo = array();
 414
 415                 // Fields that can just be stuffed in the pageInfo object
 416                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 417                                         'logtitle', 'params' );
 418
 419                 while ( $this->reader->read() ) {
 420                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 421                                         $this->reader->name == 'logitem') {
 422                                 break;
 423                         }
 424
 425                         $tag = $this->reader->name;
 426
 427                         if ( !wfRunHooks( 'ImportHandleLogItemXMLTag',
 428                                                 $this, $logInfo ) ) {
 429                                 // Do nothing
 430                         } elseif ( in_array( $tag, $normalFields ) ) {
 431                                 $logInfo[$tag] = $this->nodeContents();
 432                         } elseif ( $tag == 'contributor' ) {
 433                                 $logInfo['contributor'] = $this->handleContributor();
 434                         } elseif ( $tag != '#text' ) {
 435                                 $this->warn( "Unhandled log-item XML tag $tag" );
 436                         }
 437                 }
 438
 439                 $this->processLogItem( $logInfo );
 440         }
 441
 442         private function processLogItem( $logInfo ) {
 443                 $revision = new WikiRevision;
 444
 445                 $revision->setID( $logInfo['id'] );
 446                 $revision->setType( $logInfo['type'] );
 447                 $revision->setAction( $logInfo['action'] );
 448                 $revision->setTimestamp( $logInfo['timestamp'] );
 449                 $revision->setParams( $logInfo['params'] );
 450                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 451
 452                 if ( isset( $logInfo['comment'] ) ) {
 453                         $revision->setComment( $logInfo['comment'] );
 454                 }
 455
 456                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 457                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 458                 }
 459                 if ( isset( $logInfo['contributor']['username'] ) ) {
 460                         $revision->setUserName( $logInfo['contributor']['username'] );
 461                 }
 462
 463                 return $this->logItemCallback( $revision );
 464         }
 465
 466         private function handlePage() {
 467                 // Handle page data.
 468                 $this->debug( "Enter page handler." );
 469                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 470
 471                 // Fields that can just be stuffed in the pageInfo object
 472                 $normalFields = array( 'title', 'id', 'redirect', 'restrictions' );
 473
 474                 $skip = false;
 475                 $badTitle = false;
 476
 477                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 478                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 479                                         $this->reader->name == 'page') {
 480                                 break;
 481                         }
 482
 483                         $tag = $this->reader->name;
 484
 485                         if ( $badTitle ) {
 486                                 // The title is invalid, bail out of this page
 487                                 $skip = true;
 488                         } elseif ( !wfRunHooks( 'ImportHandlePageXMLTag', array( $this,
 489                                                 &$pageInfo ) ) ) {
 490                                 // Do nothing
 491                         } elseif ( in_array( $tag, $normalFields ) ) {
 492                                 $pageInfo[$tag] = $this->nodeContents();
 493                                 if ( $tag == 'title' ) {
 494                                         $title = $this->processTitle( $pageInfo['title'] );
 495
 496                                         if ( !$title ) {
 497                                                 $badTitle = true;
 498                                                 $skip = true;
 499                                         }
 500
 501                                         $this->pageCallback( $title );
 502                                         list( $pageInfo['_title'], $origTitle ) = $title;
 503                                 }
 504                         } elseif ( $tag == 'revision' ) {
 505                                 $this->handleRevision( $pageInfo );
 506                         } elseif ( $tag == 'upload' ) {
 507                                 $this->handleUpload( $pageInfo );
 508                         } elseif ( $tag != '#text' ) {
 509                                 $this->warn( "Unhandled page XML tag $tag" );
 510                                 $skip = true;
 511                         }
 512                 }
 513
 514                 $this->pageOutCallback( $pageInfo['_title'], $origTitle,
 515                                         $pageInfo['revisionCount'],
 516                                         $pageInfo['successfulRevisionCount'],
 517                                         $pageInfo );
 518         }
 519
 520         private function handleRevision( &$pageInfo ) {
 521                 $this->debug( "Enter revision handler" );
 522                 $revisionInfo = array();
 523
 524                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'text' );
 525
 526                 $skip = false;
 527
 528                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 529                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 530                                         $this->reader->name == 'revision') {
 531                                 break;
 532                         }
 533
 534                         $tag = $this->reader->name;
 535
 536                         if ( !wfRunHooks( 'ImportHandleRevisionXMLTag', $this,
 537                                                 $pageInfo, $revisionInfo ) ) {
 538                                 // Do nothing
 539                         } elseif ( in_array( $tag, $normalFields ) ) {
 540                                 $revisionInfo[$tag] = $this->nodeContents();
 541                         } elseif ( $tag == 'contributor' ) {
 542                                 $revisionInfo['contributor'] = $this->handleContributor();
 543                         } elseif ( $tag != '#text' ) {
 544                                 $this->warn( "Unhandled revision XML tag $tag" );
 545                                 $skip = true;
 546                         }
 547                 }
 548
 549                 $pageInfo['revisionCount']++;
 550                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 551                         $pageInfo['successfulRevisionCount']++;
 552                 }
 553         }
 554
 555         private function processRevision( $pageInfo, $revisionInfo ) {
 556                 $revision = new WikiRevision;
 557
 558                 if( isset( $revisionInfo['id'] ) ) {
 559                         $revision->setID( $revisionInfo['id'] );
 560                 }
 561                 if ( isset( $revisionInfo['text'] ) ) {
 562                         $revision->setText( $revisionInfo['text'] );
 563                 }
 564                 $revision->setTitle( $pageInfo['_title'] );
 565
 566                 if ( isset( $revisionInfo['timestamp'] ) ) {
 567                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 568                 } else {
 569                         $revision->setTimestamp( wfTimestampNow() );
 570                 }
 571
 572                 if ( isset( $revisionInfo['comment'] ) ) {
 573                         $revision->setComment( $revisionInfo['comment'] );
 574                 }
 575
 576                 if ( isset( $revisionInfo['minor'] ) ) {
 577                         $revision->setMinor( true );
 578                 }
 579                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 580                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 581                 }
 582                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 583                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 584                 }
 585
 586                 return $this->revisionCallback( $revision );
 587         }
 588
 589         private function handleUpload( &$pageInfo ) {
 590                 $this->debug( "Enter upload handler" );
 591                 $uploadInfo = array();
 592
 593                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 594                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 595
 596                 $skip = false;
 597
 598                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 599                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 600                                         $this->reader->name == 'upload') {
 601                                 break;
 602                         }
 603
 604                         $tag = $this->reader->name;
 605
 606                         if ( !wfRunHooks( 'ImportHandleUploadXMLTag', $this,
 607                                                 $pageInfo ) ) {
 608                                 // Do nothing
 609                         } elseif ( in_array( $tag, $normalFields ) ) {
 610                                 $uploadInfo[$tag] = $this->nodeContents();
 611                         } elseif ( $tag == 'contributor' ) {
 612                                 $uploadInfo['contributor'] = $this->handleContributor();
 613                         } elseif ( $tag == 'contents' ) {
 614                                 $contents = $this->nodeContents();
 615                                 $encoding = $this->reader->getAttribute( 'encoding' );
 616                                 if ( $encoding === 'base64' ) {
 617                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 618                                         $uploadInfo['isTempSrc'] = true;
 619                                 }
 620                         } elseif ( $tag != '#text' ) {
 621                                 $this->warn( "Unhandled upload XML tag $tag" );
 622                                 $skip = true;
 623                         }
 624                 }
 625
 626                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 627                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 628                         if ( file_exists( $path ) ) {
 629                                 $uploadInfo['fileSrc'] = $path;
 630                                 $uploadInfo['isTempSrc'] = false;
 631                         }
 632                 }
 633
 634                 if ( $this->mImportUploads ) {
 635                         return $this->processUpload( $pageInfo, $uploadInfo );
 636                 }
 637         }
 638
 639         private function dumpTemp( $contents ) {
 640                 $filename = tempnam( wfTempDir(), 'importupload' );
 641                 file_put_contents( $filename, $contents );
 642                 return $filename;
 643         }
 644
 645
 646         private function processUpload( $pageInfo, $uploadInfo ) {
 647                 $revision = new WikiRevision;
 648                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 649
 650                 $revision->setTitle( $pageInfo['_title'] );
 651                 $revision->setID( $pageInfo['id'] );
 652                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 653                 $revision->setText( $text );
 654                 $revision->setFilename( $uploadInfo['filename'] );
 655                 if ( isset( $uploadInfo['archivename'] ) ) {
 656                         $revision->setArchiveName( $uploadInfo['archivename'] );
 657                 }
 658                 $revision->setSrc( $uploadInfo['src'] );
 659                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 660                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 661                                 !empty( $uploadInfo['isTempSrc'] ) );
 662                 }
 663                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 664                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 665                 }
 666                 $revision->setSize( intval( $uploadInfo['size'] ) );
 667                 $revision->setComment( $uploadInfo['comment'] );
 668
 669                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 670                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 671                 }
 672                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 673                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 674                 }
 675
 676                 return call_user_func( $this->mUploadCallback, $revision );
 677         }
 678
 679         private function handleContributor() {
 680                 $fields = array( 'id', 'ip', 'username' );
 681                 $info = array();
 682
 683                 while ( $this->reader->read() ) {
 684                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 685                                         $this->reader->name == 'contributor') {
 686                                 break;
 687                         }
 688
 689                         $tag = $this->reader->name;
 690
 691                         if ( in_array( $tag, $fields ) ) {
 692                                 $info[$tag] = $this->nodeContents();
 693                         }
 694                 }
 695
 696                 return $info;
 697         }
 698
 699         private function processTitle( $text ) {
 700                 $workTitle = $text;
 701                 $origTitle = Title::newFromText( $workTitle );
 702
 703                 if( !is_null( $this->mTargetNamespace ) && !is_null( $origTitle ) ) {
 704                         $title = Title::makeTitle( $this->mTargetNamespace,
 705                                 $origTitle->getDBkey() );
 706                 } else {
 707                         $title = Title::newFromText( $workTitle );
 708                 }
 709
 710                 if( is_null( $title ) ) {
 711                         // Invalid page title? Ignore the page
 712                         $this->notice( "Skipping invalid page title '$workTitle'" );
 713                         return false;
 714                 } elseif( $title->getInterwiki() != '' ) {
 715                         $this->notice( "Skipping interwiki page title '$workTitle'" );
 716                         return false;
 717                 }
 718
 719                 return array( $title, $origTitle );
 720         }
 721 }
 722
 723 /** This is a horrible hack used to keep source compatibility */
 724 class UploadSourceAdapter {
 725         static $sourceRegistrations = array();
 726
 727         private $mSource;
 728         private $mBuffer;
 729         private $mPosition;
 730
 731         static function registerSource( $source ) {
 732                 $id = wfGenerateToken();
 733
 734                 self::$sourceRegistrations[$id] = $source;
 735
 736                 return $id;
 737         }
 738
 739         function stream_open( $path, $mode, $options, &$opened_path ) {
 740                 $url = parse_url($path);
 741                 $id = $url['host'];
 742
 743                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
 744                         return false;
 745                 }
 746
 747                 $this->mSource = self::$sourceRegistrations[$id];
 748
 749                 return true;
 750         }
 751
 752         function stream_read( $count ) {
 753                 $return = '';
 754                 $leave = false;
 755
 756                 while ( !$leave && !$this->mSource->atEnd() &&
 757                                 strlen($this->mBuffer) < $count ) {
 758                         $read = $this->mSource->readChunk();
 759
 760                         if ( !strlen($read) ) {
 761                                 $leave = true;
 762                         }
 763
 764                         $this->mBuffer .= $read;
 765                 }
 766
 767                 if ( strlen($this->mBuffer) ) {
 768                         $return = substr( $this->mBuffer, 0, $count );
 769                         $this->mBuffer = substr( $this->mBuffer, $count );
 770                 }
 771
 772                 $this->mPosition += strlen($return);
 773
 774                 return $return;
 775         }
 776
 777         function stream_write( $data ) {
 778                 return false;
 779         }
 780
 781         function stream_tell() {
 782                 return $this->mPosition;
 783         }
 784
 785         function stream_eof() {
 786                 return $this->mSource->atEnd();
 787         }
 788
 789         function url_stat() {
 790                 $result = array();
 791
 792                 $result['dev'] = $result[0] = 0;
 793                 $result['ino'] = $result[1] = 0;
 794                 $result['mode'] = $result[2] = 0;
 795                 $result['nlink'] = $result[3] = 0;
 796                 $result['uid'] = $result[4] = 0;
 797                 $result['gid'] = $result[5] = 0;
 798                 $result['rdev'] = $result[6] = 0;
 799                 $result['size'] = $result[7] = 0;
 800                 $result['atime'] = $result[8] = 0;
 801                 $result['mtime'] = $result[9] = 0;
 802                 $result['ctime'] = $result[10] = 0;
 803                 $result['blksize'] = $result[11] = 0;
 804                 $result['blocks'] = $result[12] = 0;
 805
 806                 return $result;
 807         }
 808 }
 809
 810 class XMLReader2 extends XMLReader {
 811         function nodeContents() {
 812                 if( $this->isEmptyElement ) {
 813                         return "";
 814                 }
 815                 $buffer = "";
 816                 while( $this->read() ) {
 817                         switch( $this->nodeType ) {
 818                         case XmlReader::TEXT:
 819                         case XmlReader::SIGNIFICANT_WHITESPACE:
 820                                 $buffer .= $this->value;
 821                                 break;
 822                         case XmlReader::END_ELEMENT:
 823                                 return $buffer;
 824                         }
 825                 }
 826                 return $this->close();
 827         }
 828 }
 829
 830 /**
 831  * @todo document (e.g. one-sentence class description).
 832  * @ingroup SpecialPage
 833  */
 834 class WikiRevision {
 835         var $importer = null;
 836         var $title = null;
 837         var $id = 0;
 838         var $timestamp = "20010115000000";
 839         var $user = 0;
 840         var $user_text = "";
 841         var $text = "";
 842         var $comment = "";
 843         var $minor = false;
 844         var $type = "";
 845         var $action = "";
 846         var $params = "";
 847         var $fileSrc = '';
 848         var $sha1base36 = false;
 849         var $isTemp = false;
 850         var $archiveName = '';
 851
 852         function setTitle( $title ) {
 853                 if( is_object( $title ) ) {
 854                         $this->title = $title;
 855                 } elseif( is_null( $title ) ) {
 856                         throw new MWException( "WikiRevision given a null title in import. You may need to adjust \$wgLegalTitleChars." );
 857                 } else {
 858                         throw new MWException( "WikiRevision given non-object title in import." );
 859                 }
 860         }
 861
 862         function setID( $id ) {
 863                 $this->id = $id;
 864         }
 865
 866         function setTimestamp( $ts ) {
 867                 # 2003-08-05T18:30:02Z
 868                 $this->timestamp = wfTimestamp( TS_MW, $ts );
 869         }
 870
 871         function setUsername( $user ) {
 872                 $this->user_text = $user;
 873         }
 874
 875         function setUserIP( $ip ) {
 876                 $this->user_text = $ip;
 877         }
 878
 879         function setText( $text ) {
 880                 $this->text = $text;
 881         }
 882
 883         function setComment( $text ) {
 884                 $this->comment = $text;
 885         }
 886
 887         function setMinor( $minor ) {
 888                 $this->minor = (bool)$minor;
 889         }
 890
 891         function setSrc( $src ) {
 892                 $this->src = $src;
 893         }
 894         function setFileSrc( $src, $isTemp ) {
 895                 $this->fileSrc = $src;
 896                 $this->fileIsTemp = $isTemp;
 897         }
 898         function setSha1Base36( $sha1base36 ) {
 899                 $this->sha1base36 = $sha1base36;
 900         }
 901
 902         function setFilename( $filename ) {
 903                 $this->filename = $filename;
 904         }
 905         function setArchiveName( $archiveName ) {
 906                 $this->archiveName = $archiveName;
 907         }
 908
 909         function setSize( $size ) {
 910                 $this->size = intval( $size );
 911         }
 912
 913         function setType( $type ) {
 914                 $this->type = $type;
 915         }
 916
 917         function setAction( $action ) {
 918                 $this->action = $action;
 919         }
 920
 921         function setParams( $params ) {
 922                 $this->params = $params;
 923         }
 924
 925         /**
 926          * @return Title
 927          */
 928         function getTitle() {
 929                 return $this->title;
 930         }
 931
 932         function getID() {
 933                 return $this->id;
 934         }
 935
 936         function getTimestamp() {
 937                 return $this->timestamp;
 938         }
 939
 940         function getUser() {
 941                 return $this->user_text;
 942         }
 943
 944         function getText() {
 945                 return $this->text;
 946         }
 947
 948         function getComment() {
 949                 return $this->comment;
 950         }
 951
 952         function getMinor() {
 953                 return $this->minor;
 954         }
 955
 956         function getSrc() {
 957                 return $this->src;
 958         }
 959         function getSha1() {
 960                 if ( $this->sha1base36 ) {
 961                         return wfBaseConvert( $this->sha1base36, 36, 16 );
 962                 }
 963                 return false;
 964         }
 965         function getFileSrc() {
 966                 return $this->fileSrc;
 967         }
 968         function isTempSrc() {
 969                 return $this->isTemp;
 970         }
 971
 972         function getFilename() {
 973                 return $this->filename;
 974         }
 975         function getArchiveName() {
 976                 return $this->archiveName;
 977         }
 978
 979         function getSize() {
 980                 return $this->size;
 981         }
 982
 983         function getType() {
 984                 return $this->type;
 985         }
 986
 987         function getAction() {
 988                 return $this->action;
 989         }
 990
 991         function getParams() {
 992                 return $this->params;
 993         }
 994
 995         function importOldRevision() {
 996                 $dbw = wfGetDB( DB_MASTER );
 997
 998                 # Sneak a single revision into place
 999                 $user = User::newFromName( $this->getUser() );
1000                 if( $user ) {
1001                         $userId = intval( $user->getId() );
1002                         $userText = $user->getName();
1003                         $userObj = $user;
1004                 } else {
1005                         $userId = 0;
1006                         $userText = $this->getUser();
1007                         $userObj = new User;
1008                 }
1009
1010                 // avoid memory leak...?
1011                 $linkCache = LinkCache::singleton();
1012                 $linkCache->clear();
1013
1014                 $article = new Article( $this->title );
1015                 $pageId = $article->getId();
1016                 if( $pageId == 0 ) {
1017                         # must create the page...
1018                         $pageId = $article->insertOn( $dbw );
1019                         $created = true;
1020                         $oldcountable = null;
1021                 } else {
1022                         $created = false;
1023
1024                         $prior = $dbw->selectField( 'revision', '1',
1025                                 array( 'rev_page' => $pageId,
1026                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1027                                         'rev_user_text' => $userText,
1028                                         'rev_comment'   => $this->getComment() ),
1029                                 __METHOD__
1030                         );
1031                         if( $prior ) {
1032                                 // @todo FIXME: This could fail slightly for multiple matches :P
1033                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1034                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1035                                 return false;
1036                         }
1037                         $oldcountable = $article->isCountable();
1038                 }
1039
1040                 # @todo FIXME: Use original rev_id optionally (better for backups)
1041                 # Insert the row
1042                 $revision = new Revision( array(
1043                         'page'       => $pageId,
1044                         'text'       => $this->getText(),
1045                         'comment'    => $this->getComment(),
1046                         'user'       => $userId,
1047                         'user_text'  => $userText,
1048                         'timestamp'  => $this->timestamp,
1049                         'minor_edit' => $this->minor,
1050                         ) );
1051                 $revision->insertOn( $dbw );
1052                 $changed = $article->updateIfNewerOn( $dbw, $revision );
1053
1054                 if ( $changed !== false ) {
1055                         wfDebug( __METHOD__ . ": running updates\n" );
1056                         $article->doEditUpdates( $revision, $userObj, array( 'created' => $created, 'oldcountable' => $oldcountable ) );
1057                 }
1058
1059                 return true;
1060         }
1061
1062         function importLogItem() {
1063                 $dbw = wfGetDB( DB_MASTER );
1064                 # @todo FIXME: This will not record autoblocks
1065                 if( !$this->getTitle() ) {
1066                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1067                                 $this->timestamp . "\n" );
1068                         return;
1069                 }
1070                 # Check if it exists already
1071                 // @todo FIXME: Use original log ID (better for backups)
1072                 $prior = $dbw->selectField( 'logging', '1',
1073                         array( 'log_type' => $this->getType(),
1074                                 'log_action'    => $this->getAction(),
1075                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1076                                 'log_namespace' => $this->getTitle()->getNamespace(),
1077                                 'log_title'     => $this->getTitle()->getDBkey(),
1078                                 'log_comment'   => $this->getComment(),
1079                                 #'log_user_text' => $this->user_text,
1080                                 'log_params'    => $this->params ),
1081                         __METHOD__
1082                 );
1083                 // @todo FIXME: This could fail slightly for multiple matches :P
1084                 if( $prior ) {
1085                         wfDebug( __METHOD__ . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp " .
1086                                 $this->timestamp . "\n" );
1087                         return false;
1088                 }
1089                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1090                 $data = array(
1091                         'log_id' => $log_id,
1092                         'log_type' => $this->type,
1093                         'log_action' => $this->action,
1094                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1095                         'log_user' => User::idFromName( $this->user_text ),
1096                         #'log_user_text' => $this->user_text,
1097                         'log_namespace' => $this->getTitle()->getNamespace(),
1098                         'log_title' => $this->getTitle()->getDBkey(),
1099                         'log_comment' => $this->getComment(),
1100                         'log_params' => $this->params
1101                 );
1102                 $dbw->insert( 'logging', $data, __METHOD__ );
1103         }
1104
1105         function importUpload() {
1106                 # Construct a file
1107                 $archiveName = $this->getArchiveName();
1108                 if ( $archiveName ) {
1109                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1110                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1111                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1112                 } else {
1113                         $file = wfLocalFile( $this->getTitle() );
1114                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1115                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1116                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1117                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1118                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1119                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1120                         }
1121                 }
1122                 if( !$file ) {
1123                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1124                         return false;
1125                 }
1126
1127                 # Get the file source or download if necessary
1128                 $source = $this->getFileSrc();
1129                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1130                 if ( !$source ) {
1131                         $source = $this->downloadSource();
1132                         $flags |= File::DELETE_SOURCE;
1133                 }
1134                 if( !$source ) {
1135                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1136                         return false;
1137                 }
1138                 $sha1 = $this->getSha1();
1139                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1140                         if ( $flags & File::DELETE_SOURCE ) {
1141                                 # Broken file; delete it if it is a temporary file
1142                                 unlink( $source );
1143                         }
1144                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1145                         return false;
1146                 }
1147
1148                 $user = User::newFromName( $this->user_text );
1149
1150                 # Do the actual upload
1151                 if ( $archiveName ) {
1152                         $status = $file->uploadOld( $source, $archiveName,
1153                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1154                 } else {
1155                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1156                                 $flags, false, $this->getTimestamp(), $user );
1157                 }
1158
1159                 if ( $status->isGood() ) {
1160                         wfDebug( __METHOD__ . ": Succesful\n" );
1161                         return true;
1162                 } else {
1163                         wfDebug( __METHOD__ . ': failed: ' . $status->getXml() . "\n" );
1164                         return false;
1165                 }
1166         }
1167
1168         function downloadSource() {
1169                 global $wgEnableUploads;
1170                 if( !$wgEnableUploads ) {
1171                         return false;
1172                 }
1173
1174                 $tempo = tempnam( wfTempDir(), 'download' );
1175                 $f = fopen( $tempo, 'wb' );
1176                 if( !$f ) {
1177                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1178                         return false;
1179                 }
1180
1181                 // @todo FIXME!
1182                 $src = $this->getSrc();
1183                 $data = Http::get( $src );
1184                 if( !$data ) {
1185                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1186                         fclose( $f );
1187                         unlink( $tempo );
1188                         return false;
1189                 }
1190
1191                 fwrite( $f, $data );
1192                 fclose( $f );
1193
1194                 return $tempo;
1195         }
1196
1197 }
1198
1199 /**
1200  * @todo document (e.g. one-sentence class description).
1201  * @ingroup SpecialPage
1202  */
1203 class ImportStringSource {
1204         function __construct( $string ) {
1205                 $this->mString = $string;
1206                 $this->mRead = false;
1207         }
1208
1209         function atEnd() {
1210                 return $this->mRead;
1211         }
1212
1213         function readChunk() {
1214                 if( $this->atEnd() ) {
1215                         return false;
1216                 } else {
1217                         $this->mRead = true;
1218                         return $this->mString;
1219                 }
1220         }
1221 }
1222
1223 /**
1224  * @todo document (e.g. one-sentence class description).
1225  * @ingroup SpecialPage
1226  */
1227 class ImportStreamSource {
1228         function __construct( $handle ) {
1229                 $this->mHandle = $handle;
1230         }
1231
1232         function atEnd() {
1233                 return feof( $this->mHandle );
1234         }
1235
1236         function readChunk() {
1237                 return fread( $this->mHandle, 32768 );
1238         }
1239
1240         static function newFromFile( $filename ) {
1241                 wfSuppressWarnings();
1242                 $file = fopen( $filename, 'rt' );
1243                 wfRestoreWarnings();
1244                 if( !$file ) {
1245                         return Status::newFatal( "importcantopen" );
1246                 }
1247                 return Status::newGood( new ImportStreamSource( $file ) );
1248         }
1249
1250         static function newFromUpload( $fieldname = "xmlimport" ) {
1251                 $upload =& $_FILES[$fieldname];
1252
1253                 if( !isset( $upload ) || !$upload['name'] ) {
1254                         return Status::newFatal( 'importnofile' );
1255                 }
1256                 if( !empty( $upload['error'] ) ) {
1257                         switch($upload['error']){
1258                                 case 1: # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1259                                         return Status::newFatal( 'importuploaderrorsize' );
1260                                 case 2: # The uploaded file exceeds the MAX_FILE_SIZE directive that was specified in the HTML form.
1261                                         return Status::newFatal( 'importuploaderrorsize' );
1262                                 case 3: # The uploaded file was only partially uploaded
1263                                         return Status::newFatal( 'importuploaderrorpartial' );
1264                                 case 6: #Missing a temporary folder.
1265                                         return Status::newFatal( 'importuploaderrortemp' );
1266                                 # case else: # Currently impossible
1267                         }
1268
1269                 }
1270                 $fname = $upload['tmp_name'];
1271                 if( is_uploaded_file( $fname ) ) {
1272                         return ImportStreamSource::newFromFile( $fname );
1273                 } else {
1274                         return Status::newFatal( 'importnofile' );
1275                 }
1276         }
1277
1278         static function newFromURL( $url, $method = 'GET' ) {
1279                 wfDebug( __METHOD__ . ": opening $url\n" );
1280                 # Use the standard HTTP fetch function; it times out
1281                 # quicker and sorts out user-agent problems which might
1282                 # otherwise prevent importing from large sites, such
1283                 # as the Wikimedia cluster, etc.
1284                 $data = Http::request( $method, $url );
1285                 if( $data !== false ) {
1286                         $file = tmpfile();
1287                         fwrite( $file, $data );
1288                         fflush( $file );
1289                         fseek( $file, 0 );
1290                         return Status::newGood( new ImportStreamSource( $file ) );
1291                 } else {
1292                         return Status::newFatal( 'importcantopen' );
1293                 }
1294         }
1295
1296         public static function newFromInterwiki( $interwiki, $page, $history = false, $templates = false, $pageLinkDepth = 0 ) {
1297                 if( $page == '' ) {
1298                         return Status::newFatal( 'import-noarticle' );
1299                 }
1300                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1301                 if( is_null( $link ) || $link->getInterwiki() == '' ) {
1302                         return Status::newFatal( 'importbadinterwiki' );
1303                 } else {
1304                         $params = array();
1305                         if ( $history ) $params['history'] = 1;
1306                         if ( $templates ) $params['templates'] = 1;
1307                         if ( $pageLinkDepth ) $params['pagelink-depth'] = $pageLinkDepth;
1308                         $url = $link->getFullUrl( $params );
1309                         # For interwikis, use POST to avoid redirects.
1310                         return ImportStreamSource::newFromURL( $url, "POST" );
1311                 }
1312         }
1313 }