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         private $mNoUpdates = false;
  40
  41         /**
  42          * Creates an ImportXMLReader drawing from the source provided
  43          */
  44         function __construct( $source ) {
  45                 $this->reader = new XMLReader();
  46
  47                 stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  48                 $id = UploadSourceAdapter::registerSource( $source );
  49                 if (defined( 'LIBXML_PARSEHUGE' ) ) {
  50                         $this->reader->open( "uploadsource://$id", null, LIBXML_PARSEHUGE );
  51                 }
  52                 else {
  53                         $this->reader->open( "uploadsource://$id" );
  54                 }
  55
  56                 // Default callbacks
  57                 $this->setRevisionCallback( array( $this, "importRevision" ) );
  58                 $this->setUploadCallback( array( $this, 'importUpload' ) );
  59                 $this->setLogItemCallback( array( $this, 'importLogItem' ) );
  60                 $this->setPageOutCallback( array( $this, 'finishImportPage' ) );
  61         }
  62
  63         private function throwXmlError( $err ) {
  64                 $this->debug( "FAILURE: $err" );
  65                 wfDebug( "WikiImporter XML error: $err\n" );
  66         }
  67
  68         private function debug( $data ) {
  69                 if( $this->mDebug ) {
  70                         wfDebug( "IMPORT: $data\n" );
  71                 }
  72         }
  73
  74         private function warn( $data ) {
  75                 wfDebug( "IMPORT: $data\n" );
  76         }
  77
  78         private function notice( $data ) {
  79                 global $wgCommandLineMode;
  80                 if( $wgCommandLineMode ) {
  81                         print "$data\n";
  82                 } else {
  83                         global $wgOut;
  84                         $wgOut->addHTML( "<li>" . htmlspecialchars( $data ) . "</li>\n" );
  85                 }
  86         }
  87
  88         /**
  89          * Set debug mode...
  90          */
  91         function setDebug( $debug ) {
  92                 $this->mDebug = $debug;
  93         }
  94
  95         /**
  96          * Set 'no updates' mode. In this mode, the link tables will not be updated by the importer
  97          */
  98         function setNoUpdates( $noupdates ) {
  99                 $this->mNoUpdates = $noupdates;
 100         }
 101
 102         /**
 103          * Sets the action to perform as each new page in the stream is reached.
 104          * @param $callback callback
 105          * @return callback
 106          */
 107         public function setPageCallback( $callback ) {
 108                 $previous = $this->mPageCallback;
 109                 $this->mPageCallback = $callback;
 110                 return $previous;
 111         }
 112
 113         /**
 114          * Sets the action to perform as each page in the stream is completed.
 115          * Callback accepts the page title (as a Title object), a second object
 116          * with the original title form (in case it's been overridden into a
 117          * local namespace), and a count of revisions.
 118          *
 119          * @param $callback callback
 120          * @return callback
 121          */
 122         public function setPageOutCallback( $callback ) {
 123                 $previous = $this->mPageOutCallback;
 124                 $this->mPageOutCallback = $callback;
 125                 return $previous;
 126         }
 127
 128         /**
 129          * Sets the action to perform as each page revision is reached.
 130          * @param $callback callback
 131          * @return callback
 132          */
 133         public function setRevisionCallback( $callback ) {
 134                 $previous = $this->mRevisionCallback;
 135                 $this->mRevisionCallback = $callback;
 136                 return $previous;
 137         }
 138
 139         /**
 140          * Sets the action to perform as each file upload version is reached.
 141          * @param $callback callback
 142          * @return callback
 143          */
 144         public function setUploadCallback( $callback ) {
 145                 $previous = $this->mUploadCallback;
 146                 $this->mUploadCallback = $callback;
 147                 return $previous;
 148         }
 149
 150         /**
 151          * Sets the action to perform as each log item reached.
 152          * @param $callback callback
 153          * @return callback
 154          */
 155         public function setLogItemCallback( $callback ) {
 156                 $previous = $this->mLogItemCallback;
 157                 $this->mLogItemCallback = $callback;
 158                 return $previous;
 159         }
 160
 161         /**
 162          * Sets the action to perform when site info is encountered
 163          * @param $callback callback
 164          * @return callback
 165          */
 166         public function setSiteInfoCallback( $callback ) {
 167                 $previous = $this->mSiteInfoCallback;
 168                 $this->mSiteInfoCallback = $callback;
 169                 return $previous;
 170         }
 171
 172         /**
 173          * Set a target namespace to override the defaults
 174          */
 175         public function setTargetNamespace( $namespace ) {
 176                 if( is_null( $namespace ) ) {
 177                         // Don't override namespaces
 178                         $this->mTargetNamespace = null;
 179                 } elseif( $namespace >= 0 ) {
 180                         // @todo FIXME: Check for validity
 181                         $this->mTargetNamespace = intval( $namespace );
 182                 } else {
 183                         return false;
 184                 }
 185         }
 186
 187         /**
 188          *
 189          */
 190         public function setImageBasePath( $dir ) {
 191                 $this->mImageBasePath = $dir;
 192         }
 193         public function setImportUploads( $import ) {
 194                 $this->mImportUploads = $import;
 195         }
 196
 197         /**
 198          * Default per-revision callback, performs the import.
 199          * @param $revision WikiRevision
 200          */
 201         public function importRevision( $revision ) {
 202                 $dbw = wfGetDB( DB_MASTER );
 203                 return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
 204         }
 205
 206         /**
 207          * Default per-revision callback, performs the import.
 208          * @param $rev WikiRevision
 209          */
 210         public function importLogItem( $rev ) {
 211                 $dbw = wfGetDB( DB_MASTER );
 212                 return $dbw->deadlockLoop( array( $rev, 'importLogItem' ) );
 213         }
 214
 215         /**
 216          * Dummy for now...
 217          */
 218         public function importUpload( $revision ) {
 219                 $dbw = wfGetDB( DB_MASTER );
 220                 return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
 221         }
 222
 223         /**
 224          * Mostly for hook use
 225          */
 226         public function finishImportPage( $title, $origTitle, $revCount, $sRevCount, $pageInfo ) {
 227                 $args = func_get_args();
 228                 return wfRunHooks( 'AfterImportPage', $args );
 229         }
 230
 231         /**
 232          * Alternate per-revision callback, for debugging.
 233          * @param $revision WikiRevision
 234          */
 235         public function debugRevisionHandler( &$revision ) {
 236                 $this->debug( "Got revision:" );
 237                 if( is_object( $revision->title ) ) {
 238                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 239                 } else {
 240                         $this->debug( "-- Title: <invalid>" );
 241                 }
 242                 $this->debug( "-- User: " . $revision->user_text );
 243                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 244                 $this->debug( "-- Comment: " . $revision->comment );
 245                 $this->debug( "-- Text: " . $revision->text );
 246         }
 247
 248         /**
 249          * Notify the callback function when a new <page> is reached.
 250          * @param $title Title
 251          */
 252         function pageCallback( $title ) {
 253                 if( isset( $this->mPageCallback ) ) {
 254                         call_user_func( $this->mPageCallback, $title );
 255                 }
 256         }
 257
 258         /**
 259          * Notify the callback function when a </page> is closed.
 260          * @param $title Title
 261          * @param $origTitle Title
 262          * @param $revCount Integer
 263          * @param $sucCount Int: number of revisions for which callback returned true
 264          * @param $pageInfo Array: associative array of page information
 265          */
 266         private function pageOutCallback( $title, $origTitle, $revCount, $sucCount, $pageInfo ) {
 267                 if( isset( $this->mPageOutCallback ) ) {
 268                         $args = func_get_args();
 269                         call_user_func_array( $this->mPageOutCallback, $args );
 270                 }
 271         }
 272
 273         /**
 274          * Notify the callback function of a revision
 275          * @param $revision A WikiRevision object
 276          */
 277         private function revisionCallback( $revision ) {
 278                 if ( isset( $this->mRevisionCallback ) ) {
 279                         return call_user_func_array( $this->mRevisionCallback,
 280                                         array( $revision, $this ) );
 281                 } else {
 282                         return false;
 283                 }
 284         }
 285
 286         /**
 287          * Notify the callback function of a new log item
 288          * @param $revision A WikiRevision object
 289          */
 290         private function logItemCallback( $revision ) {
 291                 if ( isset( $this->mLogItemCallback ) ) {
 292                         return call_user_func_array( $this->mLogItemCallback,
 293                                         array( $revision, $this ) );
 294                 } else {
 295                         return false;
 296                 }
 297         }
 298
 299         /**
 300          * Shouldn't something like this be built-in to XMLReader?
 301          * Fetches text contents of the current element, assuming
 302          * no sub-elements or such scary things.
 303          * @return string
 304          * @access private
 305          */
 306         private function nodeContents() {
 307                 if( $this->reader->isEmptyElement ) {
 308                         return "";
 309                 }
 310                 $buffer = "";
 311                 while( $this->reader->read() ) {
 312                         switch( $this->reader->nodeType ) {
 313                         case XmlReader::TEXT:
 314                         case XmlReader::SIGNIFICANT_WHITESPACE:
 315                                 $buffer .= $this->reader->value;
 316                                 break;
 317                         case XmlReader::END_ELEMENT:
 318                                 return $buffer;
 319                         }
 320                 }
 321
 322                 $this->reader->close();
 323                 return '';
 324         }
 325
 326         # --------------
 327
 328         /** Left in for debugging */
 329         private function dumpElement() {
 330                 static $lookup = null;
 331                 if (!$lookup) {
 332                         $xmlReaderConstants = array(
 333                                 "NONE",
 334                                 "ELEMENT",
 335                                 "ATTRIBUTE",
 336                                 "TEXT",
 337                                 "CDATA",
 338                                 "ENTITY_REF",
 339                                 "ENTITY",
 340                                 "PI",
 341                                 "COMMENT",
 342                                 "DOC",
 343                                 "DOC_TYPE",
 344                                 "DOC_FRAGMENT",
 345                                 "NOTATION",
 346                                 "WHITESPACE",
 347                                 "SIGNIFICANT_WHITESPACE",
 348                                 "END_ELEMENT",
 349                                 "END_ENTITY",
 350                                 "XML_DECLARATION",
 351                                 );
 352                         $lookup = array();
 353
 354                         foreach( $xmlReaderConstants as $name ) {
 355                                 $lookup[constant("XmlReader::$name")] = $name;
 356                         }
 357                 }
 358
 359                 print( var_dump(
 360                         $lookup[$this->reader->nodeType],
 361                         $this->reader->name,
 362                         $this->reader->value
 363                 )."\n\n" );
 364         }
 365
 366         /**
 367          * Primary entry point
 368          */
 369         public function doImport() {
 370                 $this->reader->read();
 371
 372                 if ( $this->reader->name != 'mediawiki' ) {
 373                         throw new MWException( "Expected <mediawiki> tag, got ".
 374                                 $this->reader->name );
 375                 }
 376                 $this->debug( "<mediawiki> tag is correct." );
 377
 378                 $this->debug( "Starting primary dump processing loop." );
 379
 380                 $keepReading = $this->reader->read();
 381                 $skip = false;
 382                 while ( $keepReading ) {
 383                         $tag = $this->reader->name;
 384                         $type = $this->reader->nodeType;
 385
 386                         if ( !wfRunHooks( 'ImportHandleToplevelXMLTag', $this ) ) {
 387                                 // Do nothing
 388                         } elseif ( $tag == 'mediawiki' && $type == XmlReader::END_ELEMENT ) {
 389                                 break;
 390                         } elseif ( $tag == 'siteinfo' ) {
 391                                 $this->handleSiteInfo();
 392                         } elseif ( $tag == 'page' ) {
 393                                 $this->handlePage();
 394                         } elseif ( $tag == 'logitem' ) {
 395                                 $this->handleLogItem();
 396                         } elseif ( $tag != '#text' ) {
 397                                 $this->warn( "Unhandled top-level XML tag $tag" );
 398
 399                                 $skip = true;
 400                         }
 401
 402                         if ($skip) {
 403                                 $keepReading = $this->reader->next();
 404                                 $skip = false;
 405                                 $this->debug( "Skip" );
 406                         } else {
 407                                 $keepReading = $this->reader->read();
 408                         }
 409                 }
 410
 411                 return true;
 412         }
 413
 414         private function handleSiteInfo() {
 415                 // Site info is useful, but not actually used for dump imports.
 416                 // Includes a quick short-circuit to save performance.
 417                 if ( ! $this->mSiteInfoCallback ) {
 418                         $this->reader->next();
 419                         return true;
 420                 }
 421                 throw new MWException( "SiteInfo tag is not yet handled, do not set mSiteInfoCallback" );
 422         }
 423
 424         private function handleLogItem() {
 425                 $this->debug( "Enter log item handler." );
 426                 $logInfo = array();
 427
 428                 // Fields that can just be stuffed in the pageInfo object
 429                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 430                                         'logtitle', 'params' );
 431
 432                 while ( $this->reader->read() ) {
 433                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 434                                         $this->reader->name == 'logitem') {
 435                                 break;
 436                         }
 437
 438                         $tag = $this->reader->name;
 439
 440                         if ( !wfRunHooks( 'ImportHandleLogItemXMLTag',
 441                                                 $this, $logInfo ) ) {
 442                                 // Do nothing
 443                         } elseif ( in_array( $tag, $normalFields ) ) {
 444                                 $logInfo[$tag] = $this->nodeContents();
 445                         } elseif ( $tag == 'contributor' ) {
 446                                 $logInfo['contributor'] = $this->handleContributor();
 447                         } elseif ( $tag != '#text' ) {
 448                                 $this->warn( "Unhandled log-item XML tag $tag" );
 449                         }
 450                 }
 451
 452                 $this->processLogItem( $logInfo );
 453         }
 454
 455         private function processLogItem( $logInfo ) {
 456                 $revision = new WikiRevision;
 457
 458                 $revision->setID( $logInfo['id'] );
 459                 $revision->setType( $logInfo['type'] );
 460                 $revision->setAction( $logInfo['action'] );
 461                 $revision->setTimestamp( $logInfo['timestamp'] );
 462                 $revision->setParams( $logInfo['params'] );
 463                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 464                 $revision->setNoUpdates( $this->mNoUpdates );
 465
 466                 if ( isset( $logInfo['comment'] ) ) {
 467                         $revision->setComment( $logInfo['comment'] );
 468                 }
 469
 470                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 471                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 472                 }
 473                 if ( isset( $logInfo['contributor']['username'] ) ) {
 474                         $revision->setUserName( $logInfo['contributor']['username'] );
 475                 }
 476
 477                 return $this->logItemCallback( $revision );
 478         }
 479
 480         private function handlePage() {
 481                 // Handle page data.
 482                 $this->debug( "Enter page handler." );
 483                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 484
 485                 // Fields that can just be stuffed in the pageInfo object
 486                 $normalFields = array( 'title', 'id', 'redirect', 'restrictions' );
 487
 488                 $skip = false;
 489                 $badTitle = false;
 490
 491                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 492                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 493                                         $this->reader->name == 'page') {
 494                                 break;
 495                         }
 496
 497                         $tag = $this->reader->name;
 498
 499                         if ( $badTitle ) {
 500                                 // The title is invalid, bail out of this page
 501                                 $skip = true;
 502                         } elseif ( !wfRunHooks( 'ImportHandlePageXMLTag', array( $this,
 503                                                 &$pageInfo ) ) ) {
 504                                 // Do nothing
 505                         } elseif ( in_array( $tag, $normalFields ) ) {
 506                                 $pageInfo[$tag] = $this->nodeContents();
 507                                 if ( $tag == 'title' ) {
 508                                         $title = $this->processTitle( $pageInfo['title'] );
 509
 510                                         if ( !$title ) {
 511                                                 $badTitle = true;
 512                                                 $skip = true;
 513                                         }
 514
 515                                         $this->pageCallback( $title );
 516                                         list( $pageInfo['_title'], $origTitle ) = $title;
 517                                 }
 518                         } elseif ( $tag == 'revision' ) {
 519                                 $this->handleRevision( $pageInfo );
 520                         } elseif ( $tag == 'upload' ) {
 521                                 $this->handleUpload( $pageInfo );
 522                         } elseif ( $tag != '#text' ) {
 523                                 $this->warn( "Unhandled page XML tag $tag" );
 524                                 $skip = true;
 525                         }
 526                 }
 527
 528                 $this->pageOutCallback( $pageInfo['_title'], $origTitle,
 529                                         $pageInfo['revisionCount'],
 530                                         $pageInfo['successfulRevisionCount'],
 531                                         $pageInfo );
 532         }
 533
 534         private function handleRevision( &$pageInfo ) {
 535                 $this->debug( "Enter revision handler" );
 536                 $revisionInfo = array();
 537
 538                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'text' );
 539
 540                 $skip = false;
 541
 542                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 543                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 544                                         $this->reader->name == 'revision') {
 545                                 break;
 546                         }
 547
 548                         $tag = $this->reader->name;
 549
 550                         if ( !wfRunHooks( 'ImportHandleRevisionXMLTag', $this,
 551                                                 $pageInfo, $revisionInfo ) ) {
 552                                 // Do nothing
 553                         } elseif ( in_array( $tag, $normalFields ) ) {
 554                                 $revisionInfo[$tag] = $this->nodeContents();
 555                         } elseif ( $tag == 'contributor' ) {
 556                                 $revisionInfo['contributor'] = $this->handleContributor();
 557                         } elseif ( $tag != '#text' ) {
 558                                 $this->warn( "Unhandled revision XML tag $tag" );
 559                                 $skip = true;
 560                         }
 561                 }
 562
 563                 $pageInfo['revisionCount']++;
 564                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 565                         $pageInfo['successfulRevisionCount']++;
 566                 }
 567         }
 568
 569         private function processRevision( $pageInfo, $revisionInfo ) {
 570                 $revision = new WikiRevision;
 571
 572                 if( isset( $revisionInfo['id'] ) ) {
 573                         $revision->setID( $revisionInfo['id'] );
 574                 }
 575                 if ( isset( $revisionInfo['text'] ) ) {
 576                         $revision->setText( $revisionInfo['text'] );
 577                 }
 578                 $revision->setTitle( $pageInfo['_title'] );
 579
 580                 if ( isset( $revisionInfo['timestamp'] ) ) {
 581                         $revision->setTimestamp( $revisionInfo['timestamp'] );
 582                 } else {
 583                         $revision->setTimestamp( wfTimestampNow() );
 584                 }
 585
 586                 if ( isset( $revisionInfo['comment'] ) ) {
 587                         $revision->setComment( $revisionInfo['comment'] );
 588                 }
 589
 590                 if ( isset( $revisionInfo['minor'] ) ) {
 591                         $revision->setMinor( true );
 592                 }
 593                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 594                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 595                 }
 596                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 597                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 598                 }
 599                 $revision->setNoUpdates( $this->mNoUpdates );
 600
 601                 return $this->revisionCallback( $revision );
 602         }
 603
 604         private function handleUpload( &$pageInfo ) {
 605                 $this->debug( "Enter upload handler" );
 606                 $uploadInfo = array();
 607
 608                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 609                                         'src', 'size', 'sha1base36', 'archivename', 'rel' );
 610
 611                 $skip = false;
 612
 613                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 614                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 615                                         $this->reader->name == 'upload') {
 616                                 break;
 617                         }
 618
 619                         $tag = $this->reader->name;
 620
 621                         if ( !wfRunHooks( 'ImportHandleUploadXMLTag', $this,
 622                                                 $pageInfo ) ) {
 623                                 // Do nothing
 624                         } elseif ( in_array( $tag, $normalFields ) ) {
 625                                 $uploadInfo[$tag] = $this->nodeContents();
 626                         } elseif ( $tag == 'contributor' ) {
 627                                 $uploadInfo['contributor'] = $this->handleContributor();
 628                         } elseif ( $tag == 'contents' ) {
 629                                 $contents = $this->nodeContents();
 630                                 $encoding = $this->reader->getAttribute( 'encoding' );
 631                                 if ( $encoding === 'base64' ) {
 632                                         $uploadInfo['fileSrc'] = $this->dumpTemp( base64_decode( $contents ) );
 633                                         $uploadInfo['isTempSrc'] = true;
 634                                 }
 635                         } elseif ( $tag != '#text' ) {
 636                                 $this->warn( "Unhandled upload XML tag $tag" );
 637                                 $skip = true;
 638                         }
 639                 }
 640
 641                 if ( $this->mImageBasePath && isset( $uploadInfo['rel'] ) ) {
 642                         $path = "{$this->mImageBasePath}/{$uploadInfo['rel']}";
 643                         if ( file_exists( $path ) ) {
 644                                 $uploadInfo['fileSrc'] = $path;
 645                                 $uploadInfo['isTempSrc'] = false;
 646                         }
 647                 }
 648
 649                 if ( $this->mImportUploads ) {
 650                         return $this->processUpload( $pageInfo, $uploadInfo );
 651                 }
 652         }
 653
 654         private function dumpTemp( $contents ) {
 655                 $filename = tempnam( wfTempDir(), 'importupload' );
 656                 file_put_contents( $filename, $contents );
 657                 return $filename;
 658         }
 659
 660
 661         private function processUpload( $pageInfo, $uploadInfo ) {
 662                 $revision = new WikiRevision;
 663                 $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 664
 665                 $revision->setTitle( $pageInfo['_title'] );
 666                 $revision->setID( $pageInfo['id'] );
 667                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 668                 $revision->setText( $text );
 669                 $revision->setFilename( $uploadInfo['filename'] );
 670                 if ( isset( $uploadInfo['archivename'] ) ) {
 671                         $revision->setArchiveName( $uploadInfo['archivename'] );
 672                 }
 673                 $revision->setSrc( $uploadInfo['src'] );
 674                 if ( isset( $uploadInfo['fileSrc'] ) ) {
 675                         $revision->setFileSrc( $uploadInfo['fileSrc'],
 676                                 !empty( $uploadInfo['isTempSrc'] ) );
 677                 }
 678                 if ( isset( $uploadInfo['sha1base36'] ) ) {
 679                         $revision->setSha1Base36( $uploadInfo['sha1base36'] );
 680                 }
 681                 $revision->setSize( intval( $uploadInfo['size'] ) );
 682                 $revision->setComment( $uploadInfo['comment'] );
 683
 684                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 685                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 686                 }
 687                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 688                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 689                 }
 690                 $revision->setNoUpdates( $this->mNoUpdates );
 691
 692                 return call_user_func( $this->mUploadCallback, $revision );
 693         }
 694
 695         private function handleContributor() {
 696                 $fields = array( 'id', 'ip', 'username' );
 697                 $info = array();
 698
 699                 while ( $this->reader->read() ) {
 700                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 701                                         $this->reader->name == 'contributor') {
 702                                 break;
 703                         }
 704
 705                         $tag = $this->reader->name;
 706
 707                         if ( in_array( $tag, $fields ) ) {
 708                                 $info[$tag] = $this->nodeContents();
 709                         }
 710                 }
 711
 712                 return $info;
 713         }
 714
 715         /**
 716          * @return Array or false
 717          */
 718         private function processTitle( $text ) {
 719                 global $wgCommandLineMode;
 720
 721                 $workTitle = $text;
 722                 $origTitle = Title::newFromText( $workTitle );
 723
 724                 if( !is_null( $this->mTargetNamespace ) && !is_null( $origTitle ) ) {
 725                         $title = Title::makeTitle( $this->mTargetNamespace,
 726                                 $origTitle->getDBkey() );
 727                 } else {
 728                         $title = Title::newFromText( $workTitle );
 729                 }
 730
 731                 if( is_null( $title ) ) {
 732                         # Invalid page title? Ignore the page
 733                         $this->notice( "Skipping invalid page title '$workTitle'" );
 734                         return false;
 735                 } elseif( $title->getInterwiki() != '' ) {
 736                         $this->notice( "Skipping interwiki page title '$workTitle'" );
 737                         return false;
 738                 } elseif( !$title->userCan( 'edit' ) && !$wgCommandLineMode ) {
 739                         # Do not import if the importing wiki user cannot edit this page
 740                         $this->notice( wfMessage( 'import-error-edit', $title->getText() )->text() );
 741                         return false;
 742                 } elseif( !$title->exists() && !$title->userCan( 'create' ) && !$wgCommandLineMode ) {
 743                         # Do not import if the importing wiki user cannot create this page
 744                         $this->notice( wfMessage( 'import-error-create', $title->getText() )->text() );
 745                         return false;
 746                 }
 747
 748                 return array( $title, $origTitle );
 749         }
 750 }
 751
 752 /** This is a horrible hack used to keep source compatibility */
 753 class UploadSourceAdapter {
 754         static $sourceRegistrations = array();
 755
 756         private $mSource;
 757         private $mBuffer;
 758         private $mPosition;
 759
 760         static function registerSource( $source ) {
 761                 $id = wfGenerateToken();
 762
 763                 self::$sourceRegistrations[$id] = $source;
 764
 765                 return $id;
 766         }
 767
 768         function stream_open( $path, $mode, $options, &$opened_path ) {
 769                 $url = parse_url($path);
 770                 $id = $url['host'];
 771
 772                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
 773                         return false;
 774                 }
 775
 776                 $this->mSource = self::$sourceRegistrations[$id];
 777
 778                 return true;
 779         }
 780
 781         function stream_read( $count ) {
 782                 $return = '';
 783                 $leave = false;
 784
 785                 while ( !$leave && !$this->mSource->atEnd() &&
 786                                 strlen($this->mBuffer) < $count ) {
 787                         $read = $this->mSource->readChunk();
 788
 789                         if ( !strlen($read) ) {
 790                                 $leave = true;
 791                         }
 792
 793                         $this->mBuffer .= $read;
 794                 }
 795
 796                 if ( strlen($this->mBuffer) ) {
 797                         $return = substr( $this->mBuffer, 0, $count );
 798                         $this->mBuffer = substr( $this->mBuffer, $count );
 799                 }
 800
 801                 $this->mPosition += strlen($return);
 802
 803                 return $return;
 804         }
 805
 806         function stream_write( $data ) {
 807                 return false;
 808         }
 809
 810         function stream_tell() {
 811                 return $this->mPosition;
 812         }
 813
 814         function stream_eof() {
 815                 return $this->mSource->atEnd();
 816         }
 817
 818         function url_stat() {
 819                 $result = array();
 820
 821                 $result['dev'] = $result[0] = 0;
 822                 $result['ino'] = $result[1] = 0;
 823                 $result['mode'] = $result[2] = 0;
 824                 $result['nlink'] = $result[3] = 0;
 825                 $result['uid'] = $result[4] = 0;
 826                 $result['gid'] = $result[5] = 0;
 827                 $result['rdev'] = $result[6] = 0;
 828                 $result['size'] = $result[7] = 0;
 829                 $result['atime'] = $result[8] = 0;
 830                 $result['mtime'] = $result[9] = 0;
 831                 $result['ctime'] = $result[10] = 0;
 832                 $result['blksize'] = $result[11] = 0;
 833                 $result['blocks'] = $result[12] = 0;
 834
 835                 return $result;
 836         }
 837 }
 838
 839 class XMLReader2 extends XMLReader {
 840         function nodeContents() {
 841                 if( $this->isEmptyElement ) {
 842                         return "";
 843                 }
 844                 $buffer = "";
 845                 while( $this->read() ) {
 846                         switch( $this->nodeType ) {
 847                         case XmlReader::TEXT:
 848                         case XmlReader::SIGNIFICANT_WHITESPACE:
 849                                 $buffer .= $this->value;
 850                                 break;
 851                         case XmlReader::END_ELEMENT:
 852                                 return $buffer;
 853                         }
 854                 }
 855                 return $this->close();
 856         }
 857 }
 858
 859 /**
 860  * @todo document (e.g. one-sentence class description).
 861  * @ingroup SpecialPage
 862  */
 863 class WikiRevision {
 864         var $importer = null;
 865         var $title = null;
 866         var $id = 0;
 867         var $timestamp = "20010115000000";
 868         var $user = 0;
 869         var $user_text = "";
 870         var $text = "";
 871         var $comment = "";
 872         var $minor = false;
 873         var $type = "";
 874         var $action = "";
 875         var $params = "";
 876         var $fileSrc = '';
 877         var $sha1base36 = false;
 878         var $isTemp = false;
 879         var $archiveName = '';
 880         private $mNoUpdates = false;
 881
 882         function setTitle( $title ) {
 883                 if( is_object( $title ) ) {
 884                         $this->title = $title;
 885                 } elseif( is_null( $title ) ) {
 886                         throw new MWException( "WikiRevision given a null title in import. You may need to adjust \$wgLegalTitleChars." );
 887                 } else {
 888                         throw new MWException( "WikiRevision given non-object title in import." );
 889                 }
 890         }
 891
 892         function setID( $id ) {
 893                 $this->id = $id;
 894         }
 895
 896         function setTimestamp( $ts ) {
 897                 # 2003-08-05T18:30:02Z
 898                 $this->timestamp = wfTimestamp( TS_MW, $ts );
 899         }
 900
 901         function setUsername( $user ) {
 902                 $this->user_text = $user;
 903         }
 904
 905         function setUserIP( $ip ) {
 906                 $this->user_text = $ip;
 907         }
 908
 909         function setText( $text ) {
 910                 $this->text = $text;
 911         }
 912
 913         function setComment( $text ) {
 914                 $this->comment = $text;
 915         }
 916
 917         function setMinor( $minor ) {
 918                 $this->minor = (bool)$minor;
 919         }
 920
 921         function setSrc( $src ) {
 922                 $this->src = $src;
 923         }
 924         function setFileSrc( $src, $isTemp ) {
 925                 $this->fileSrc = $src;
 926                 $this->fileIsTemp = $isTemp;
 927         }
 928         function setSha1Base36( $sha1base36 ) {
 929                 $this->sha1base36 = $sha1base36;
 930         }
 931
 932         function setFilename( $filename ) {
 933                 $this->filename = $filename;
 934         }
 935         function setArchiveName( $archiveName ) {
 936                 $this->archiveName = $archiveName;
 937         }
 938
 939         function setSize( $size ) {
 940                 $this->size = intval( $size );
 941         }
 942
 943         function setType( $type ) {
 944                 $this->type = $type;
 945         }
 946
 947         function setAction( $action ) {
 948                 $this->action = $action;
 949         }
 950
 951         function setParams( $params ) {
 952                 $this->params = $params;
 953         }
 954
 955         public function setNoUpdates( $noupdates ) {
 956                 $this->mNoUpdates = $noupdates;
 957         }
 958
 959         /**
 960          * @return Title
 961          */
 962         function getTitle() {
 963                 return $this->title;
 964         }
 965
 966         function getID() {
 967                 return $this->id;
 968         }
 969
 970         function getTimestamp() {
 971                 return $this->timestamp;
 972         }
 973
 974         function getUser() {
 975                 return $this->user_text;
 976         }
 977
 978         function getText() {
 979                 return $this->text;
 980         }
 981
 982         function getComment() {
 983                 return $this->comment;
 984         }
 985
 986         function getMinor() {
 987                 return $this->minor;
 988         }
 989
 990         function getSrc() {
 991                 return $this->src;
 992         }
 993         function getSha1() {
 994                 if ( $this->sha1base36 ) {
 995                         return wfBaseConvert( $this->sha1base36, 36, 16 );
 996                 }
 997                 return false;
 998         }
 999         function getFileSrc() {
1000                 return $this->fileSrc;
1001         }
1002         function isTempSrc() {
1003                 return $this->isTemp;
1004         }
1005
1006         function getFilename() {
1007                 return $this->filename;
1008         }
1009         function getArchiveName() {
1010                 return $this->archiveName;
1011         }
1012
1013         function getSize() {
1014                 return $this->size;
1015         }
1016
1017         function getType() {
1018                 return $this->type;
1019         }
1020
1021         function getAction() {
1022                 return $this->action;
1023         }
1024
1025         function getParams() {
1026                 return $this->params;
1027         }
1028
1029         function importOldRevision() {
1030                 $dbw = wfGetDB( DB_MASTER );
1031
1032                 # Sneak a single revision into place
1033                 $user = User::newFromName( $this->getUser() );
1034                 if( $user ) {
1035                         $userId = intval( $user->getId() );
1036                         $userText = $user->getName();
1037                         $userObj = $user;
1038                 } else {
1039                         $userId = 0;
1040                         $userText = $this->getUser();
1041                         $userObj = new User;
1042                 }
1043
1044                 // avoid memory leak...?
1045                 $linkCache = LinkCache::singleton();
1046                 $linkCache->clear();
1047
1048                 $article = new Article( $this->title );
1049                 $pageId = $article->getId();
1050                 if( $pageId == 0 ) {
1051                         # must create the page...
1052                         $pageId = $article->insertOn( $dbw );
1053                         $created = true;
1054                         $oldcountable = null;
1055                 } else {
1056                         $created = false;
1057
1058                         $prior = $dbw->selectField( 'revision', '1',
1059                                 array( 'rev_page' => $pageId,
1060                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
1061                                         'rev_user_text' => $userText,
1062                                         'rev_comment'   => $this->getComment() ),
1063                                 __METHOD__
1064                         );
1065                         if( $prior ) {
1066                                 // @todo FIXME: This could fail slightly for multiple matches :P
1067                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
1068                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
1069                                 return false;
1070                         }
1071                         $oldcountable = $article->isCountable();
1072                 }
1073
1074                 # @todo FIXME: Use original rev_id optionally (better for backups)
1075                 # Insert the row
1076                 $revision = new Revision( array(
1077                         'page'       => $pageId,
1078                         'text'       => $this->getText(),
1079                         'comment'    => $this->getComment(),
1080                         'user'       => $userId,
1081                         'user_text'  => $userText,
1082                         'timestamp'  => $this->timestamp,
1083                         'minor_edit' => $this->minor,
1084                         ) );
1085                 $revision->insertOn( $dbw );
1086                 $changed = $article->updateIfNewerOn( $dbw, $revision );
1087
1088                 if ( $changed !== false && !$this->mNoUpdates ) {
1089                         wfDebug( __METHOD__ . ": running updates\n" );
1090                         $article->doEditUpdates( $revision, $userObj, array( 'created' => $created, 'oldcountable' => $oldcountable ) );
1091                 }
1092
1093                 return true;
1094         }
1095
1096         function importLogItem() {
1097                 $dbw = wfGetDB( DB_MASTER );
1098                 # @todo FIXME: This will not record autoblocks
1099                 if( !$this->getTitle() ) {
1100                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
1101                                 $this->timestamp . "\n" );
1102                         return;
1103                 }
1104                 # Check if it exists already
1105                 // @todo FIXME: Use original log ID (better for backups)
1106                 $prior = $dbw->selectField( 'logging', '1',
1107                         array( 'log_type' => $this->getType(),
1108                                 'log_action'    => $this->getAction(),
1109                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1110                                 'log_namespace' => $this->getTitle()->getNamespace(),
1111                                 'log_title'     => $this->getTitle()->getDBkey(),
1112                                 'log_comment'   => $this->getComment(),
1113                                 #'log_user_text' => $this->user_text,
1114                                 'log_params'    => $this->params ),
1115                         __METHOD__
1116                 );
1117                 // @todo FIXME: This could fail slightly for multiple matches :P
1118                 if( $prior ) {
1119                         wfDebug( __METHOD__ . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp " .
1120                                 $this->timestamp . "\n" );
1121                         return false;
1122                 }
1123                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1124                 $data = array(
1125                         'log_id' => $log_id,
1126                         'log_type' => $this->type,
1127                         'log_action' => $this->action,
1128                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1129                         'log_user' => User::idFromName( $this->user_text ),
1130                         #'log_user_text' => $this->user_text,
1131                         'log_namespace' => $this->getTitle()->getNamespace(),
1132                         'log_title' => $this->getTitle()->getDBkey(),
1133                         'log_comment' => $this->getComment(),
1134                         'log_params' => $this->params
1135                 );
1136                 $dbw->insert( 'logging', $data, __METHOD__ );
1137         }
1138
1139         function importUpload() {
1140                 # Construct a file
1141                 $archiveName = $this->getArchiveName();
1142                 if ( $archiveName ) {
1143                         wfDebug( __METHOD__ . "Importing archived file as $archiveName\n" );
1144                         $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1145                                 RepoGroup::singleton()->getLocalRepo(), $archiveName );
1146                 } else {
1147                         $file = wfLocalFile( $this->getTitle() );
1148                         wfDebug( __METHOD__ . 'Importing new file as ' . $file->getName() . "\n" );
1149                         if ( $file->exists() && $file->getTimestamp() > $this->getTimestamp() ) {
1150                                 $archiveName = $file->getTimestamp() . '!' . $file->getName();
1151                                 $file = OldLocalFile::newFromArchiveName( $this->getTitle(),
1152                                         RepoGroup::singleton()->getLocalRepo(), $archiveName );
1153                                 wfDebug( __METHOD__ . "File already exists; importing as $archiveName\n" );
1154                         }
1155                 }
1156                 if( !$file ) {
1157                         wfDebug( __METHOD__ . ': Bad file for ' . $this->getTitle() . "\n" );
1158                         return false;
1159                 }
1160
1161                 # Get the file source or download if necessary
1162                 $source = $this->getFileSrc();
1163                 $flags = $this->isTempSrc() ? File::DELETE_SOURCE : 0;
1164                 if ( !$source ) {
1165                         $source = $this->downloadSource();
1166                         $flags |= File::DELETE_SOURCE;
1167                 }
1168                 if( !$source ) {
1169                         wfDebug( __METHOD__ . ": Could not fetch remote file.\n" );
1170                         return false;
1171                 }
1172                 $sha1 = $this->getSha1();
1173                 if ( $sha1 && ( $sha1 !== sha1_file( $source ) ) ) {
1174                         if ( $flags & File::DELETE_SOURCE ) {
1175                                 # Broken file; delete it if it is a temporary file
1176                                 unlink( $source );
1177                         }
1178                         wfDebug( __METHOD__ . ": Corrupt file $source.\n" );
1179                         return false;
1180                 }
1181
1182                 $user = User::newFromName( $this->user_text );
1183
1184                 # Do the actual upload
1185                 if ( $archiveName ) {
1186                         $status = $file->uploadOld( $source, $archiveName,
1187                                 $this->getTimestamp(), $this->getComment(), $user, $flags );
1188                 } else {
1189                         $status = $file->upload( $source, $this->getComment(), $this->getComment(),
1190                                 $flags, false, $this->getTimestamp(), $user );
1191                 }
1192
1193                 if ( $status->isGood() ) {
1194                         wfDebug( __METHOD__ . ": Succesful\n" );
1195                         return true;
1196                 } else {
1197                         wfDebug( __METHOD__ . ': failed: ' . $status->getXml() . "\n" );
1198                         return false;
1199                 }
1200         }
1201
1202         function downloadSource() {
1203                 global $wgEnableUploads;
1204                 if( !$wgEnableUploads ) {
1205                         return false;
1206                 }
1207
1208                 $tempo = tempnam( wfTempDir(), 'download' );
1209                 $f = fopen( $tempo, 'wb' );
1210                 if( !$f ) {
1211                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1212                         return false;
1213                 }
1214
1215                 // @todo FIXME!
1216                 $src = $this->getSrc();
1217                 $data = Http::get( $src );
1218                 if( !$data ) {
1219                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1220                         fclose( $f );
1221                         unlink( $tempo );
1222                         return false;
1223                 }
1224
1225                 fwrite( $f, $data );
1226                 fclose( $f );
1227
1228                 return $tempo;
1229         }
1230
1231 }
1232
1233 /**
1234  * @todo document (e.g. one-sentence class description).
1235  * @ingroup SpecialPage
1236  */
1237 class ImportStringSource {
1238         function __construct( $string ) {
1239                 $this->mString = $string;
1240                 $this->mRead = false;
1241         }
1242
1243         function atEnd() {
1244                 return $this->mRead;
1245         }
1246
1247         function readChunk() {
1248                 if( $this->atEnd() ) {
1249                         return false;
1250                 } else {
1251                         $this->mRead = true;
1252                         return $this->mString;
1253                 }
1254         }
1255 }
1256
1257 /**
1258  * @todo document (e.g. one-sentence class description).
1259  * @ingroup SpecialPage
1260  */
1261 class ImportStreamSource {
1262         function __construct( $handle ) {
1263                 $this->mHandle = $handle;
1264         }
1265
1266         function atEnd() {
1267                 return feof( $this->mHandle );
1268         }
1269
1270         function readChunk() {
1271                 return fread( $this->mHandle, 32768 );
1272         }
1273
1274         static function newFromFile( $filename ) {
1275                 wfSuppressWarnings();
1276                 $file = fopen( $filename, 'rt' );
1277                 wfRestoreWarnings();
1278                 if( !$file ) {
1279                         return Status::newFatal( "importcantopen" );
1280                 }
1281                 return Status::newGood( new ImportStreamSource( $file ) );
1282         }
1283
1284         static function newFromUpload( $fieldname = "xmlimport" ) {
1285                 $upload =& $_FILES[$fieldname];
1286
1287                 if( !isset( $upload ) || !$upload['name'] ) {
1288                         return Status::newFatal( 'importnofile' );
1289                 }
1290                 if( !empty( $upload['error'] ) ) {
1291                         switch($upload['error']){
1292                                 case 1: # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1293                                         return Status::newFatal( 'importuploaderrorsize' );
1294                                 case 2: # The uploaded file exceeds the MAX_FILE_SIZE directive that was specified in the HTML form.
1295                                         return Status::newFatal( 'importuploaderrorsize' );
1296                                 case 3: # The uploaded file was only partially uploaded
1297                                         return Status::newFatal( 'importuploaderrorpartial' );
1298                                 case 6: #Missing a temporary folder.
1299                                         return Status::newFatal( 'importuploaderrortemp' );
1300                                 # case else: # Currently impossible
1301                         }
1302
1303                 }
1304                 $fname = $upload['tmp_name'];
1305                 if( is_uploaded_file( $fname ) ) {
1306                         return ImportStreamSource::newFromFile( $fname );
1307                 } else {
1308                         return Status::newFatal( 'importnofile' );
1309                 }
1310         }
1311
1312         static function newFromURL( $url, $method = 'GET' ) {
1313                 wfDebug( __METHOD__ . ": opening $url\n" );
1314                 # Use the standard HTTP fetch function; it times out
1315                 # quicker and sorts out user-agent problems which might
1316                 # otherwise prevent importing from large sites, such
1317                 # as the Wikimedia cluster, etc.
1318                 $data = Http::request( $method, $url );
1319                 if( $data !== false ) {
1320                         $file = tmpfile();
1321                         fwrite( $file, $data );
1322                         fflush( $file );
1323                         fseek( $file, 0 );
1324                         return Status::newGood( new ImportStreamSource( $file ) );
1325                 } else {
1326                         return Status::newFatal( 'importcantopen' );
1327                 }
1328         }
1329
1330         public static function newFromInterwiki( $interwiki, $page, $history = false, $templates = false, $pageLinkDepth = 0 ) {
1331                 if( $page == '' ) {
1332                         return Status::newFatal( 'import-noarticle' );
1333                 }
1334                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1335                 if( is_null( $link ) || $link->getInterwiki() == '' ) {
1336                         return Status::newFatal( 'importbadinterwiki' );
1337                 } else {
1338                         $params = array();
1339                         if ( $history ) $params['history'] = 1;
1340                         if ( $templates ) $params['templates'] = 1;
1341                         if ( $pageLinkDepth ) $params['pagelink-depth'] = $pageLinkDepth;
1342                         $url = $link->getFullUrl( $params );
1343                         # For interwikis, use POST to avoid redirects.
1344                         return ImportStreamSource::newFromURL( $url, "POST" );
1345                 }
1346         }
1347 }