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