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