includes/Import.php

   1 <?php
   2 /**
   3  * MediaWiki page data importer
   4  *
   5  * Copyright © 2003,2005 Brion Vibber <brion@pobox.com>
   6  * http://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup SpecialPage
  25  */
  26
  27 /**
  28  * XML file reader for the page data importer
  29  *
  30  * implements Special:Import
  31  * @ingroup SpecialPage
  32  */
  33 class WikiImporter {
  34         private $reader = null;
  35         private $mLogItemCallback, $mUploadCallback, $mRevisionCallback, $mPageCallback;
  36         private $mSiteInfoCallback, $mTargetNamespace, $mPageOutCallback;
  37         private $mDebug;
  38
  39         /**
  40          * Creates an ImportXMLReader drawing from the source provided
  41         */
  42         function __construct( $source ) {
  43                 $this->reader = new XMLReader();
  44
  45                 stream_wrapper_register( 'uploadsource', 'UploadSourceAdapter' );
  46                 $id = UploadSourceAdapter::registerSource( $source );
  47                 $this->reader->open( "uploadsource://$id" );
  48
  49                 // Default callbacks
  50                 $this->setRevisionCallback( array( $this, "importRevision" ) );
  51                 $this->setUploadCallback( array( $this, 'importUpload' ) );
  52                 $this->setLogItemCallback( array( $this, 'importLogItem' ) );
  53                 $this->setPageOutCallback( array( $this, 'finishImportPage' ) );
  54         }
  55
  56         private function throwXmlError( $err ) {
  57                 $this->debug( "FAILURE: $err" );
  58                 wfDebug( "WikiImporter XML error: $err\n" );
  59         }
  60
  61         private function debug( $data ) {
  62                 if( $this->mDebug ) {
  63                         wfDebug( "IMPORT: $data\n" );
  64                 }
  65         }
  66
  67         private function warn( $data ) {
  68                 wfDebug( "IMPORT: $data\n" );
  69         }
  70
  71         private function notice( $data ) {
  72                 global $wgCommandLineMode;
  73                 if( $wgCommandLineMode ) {
  74                         print "$data\n";
  75                 } else {
  76                         global $wgOut;
  77                         $wgOut->addHTML( "<li>" . htmlspecialchars( $data ) . "</li>\n" );
  78                 }
  79         }
  80
  81         /**
  82          * Set debug mode...
  83          */
  84         function setDebug( $debug ) {
  85                 $this->mDebug = $debug;
  86         }
  87
  88         /**
  89          * Sets the action to perform as each new page in the stream is reached.
  90          * @param $callback callback
  91          * @return callback
  92          */
  93         public function setPageCallback( $callback ) {
  94                 $previous = $this->mPageCallback;
  95                 $this->mPageCallback = $callback;
  96                 return $previous;
  97         }
  98
  99         /**
 100          * Sets the action to perform as each page in the stream is completed.
 101          * Callback accepts the page title (as a Title object), a second object
 102          * with the original title form (in case it's been overridden into a
 103          * local namespace), and a count of revisions.
 104          *
 105          * @param $callback callback
 106          * @return callback
 107          */
 108         public function setPageOutCallback( $callback ) {
 109                 $previous = $this->mPageOutCallback;
 110                 $this->mPageOutCallback = $callback;
 111                 return $previous;
 112         }
 113
 114         /**
 115          * Sets the action to perform as each page revision is reached.
 116          * @param $callback callback
 117          * @return callback
 118          */
 119         public function setRevisionCallback( $callback ) {
 120                 $previous = $this->mRevisionCallback;
 121                 $this->mRevisionCallback = $callback;
 122                 return $previous;
 123         }
 124
 125         /**
 126          * Sets the action to perform as each file upload version is reached.
 127          * @param $callback callback
 128          * @return callback
 129          */
 130         public function setUploadCallback( $callback ) {
 131                 $previous = $this->mUploadCallback;
 132                 $this->mUploadCallback = $callback;
 133                 return $previous;
 134         }
 135
 136         /**
 137          * Sets the action to perform as each log item reached.
 138          * @param $callback callback
 139          * @return callback
 140          */
 141         public function setLogItemCallback( $callback ) {
 142                 $previous = $this->mLogItemCallback;
 143                 $this->mLogItemCallback = $callback;
 144                 return $previous;
 145         }
 146
 147         /**
 148          * Sets the action to perform when site info is encountered
 149          * @param $callback callback
 150          * @return callback
 151          */
 152         public function setSiteInfoCallback( $callback ) {
 153                 $previous = $this->mSiteInfoCallback;
 154                 $this->mSiteInfoCallback = $callback;
 155                 return $previous;
 156         }
 157
 158         /**
 159          * Set a target namespace to override the defaults
 160          */
 161         public function setTargetNamespace( $namespace ) {
 162                 if( is_null( $namespace ) ) {
 163                         // Don't override namespaces
 164                         $this->mTargetNamespace = null;
 165                 } elseif( $namespace >= 0 ) {
 166                         // FIXME: Check for validity
 167                         $this->mTargetNamespace = intval( $namespace );
 168                 } else {
 169                         return false;
 170                 }
 171         }
 172
 173         /**
 174          * Default per-revision callback, performs the import.
 175          * @param $revision WikiRevision
 176          */
 177         public function importRevision( $revision ) {
 178                 $dbw = wfGetDB( DB_MASTER );
 179                 return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
 180         }
 181
 182         /**
 183          * Default per-revision callback, performs the import.
 184          * @param $rev WikiRevision
 185          */
 186         public function importLogItem( $rev ) {
 187                 $dbw = wfGetDB( DB_MASTER );
 188                 return $dbw->deadlockLoop( array( $rev, 'importLogItem' ) );
 189         }
 190
 191         /**
 192          * Dummy for now...
 193          */
 194         public function importUpload( $revision ) {
 195                            $revision->importUpload();
 196                 //$dbw = wfGetDB( DB_MASTER );
 197                 //return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
 198                 return false;
 199         }
 200
 201         /**
 202          * Mostly for hook use
 203          */
 204         public function finishImportPage( $title, $origTitle, $revCount, $sRevCount, $pageInfo ) {
 205                 $args = func_get_args();
 206                 return wfRunHooks( 'AfterImportPage', $args );
 207         }
 208
 209         /**
 210          * Alternate per-revision callback, for debugging.
 211          * @param $revision WikiRevision
 212          */
 213         public function debugRevisionHandler( &$revision ) {
 214                 $this->debug( "Got revision:" );
 215                 if( is_object( $revision->title ) ) {
 216                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 217                 } else {
 218                         $this->debug( "-- Title: <invalid>" );
 219                 }
 220                 $this->debug( "-- User: " . $revision->user_text );
 221                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 222                 $this->debug( "-- Comment: " . $revision->comment );
 223                 $this->debug( "-- Text: " . $revision->text );
 224         }
 225
 226         /**
 227          * Notify the callback function when a new <page> is reached.
 228          * @param $title Title
 229          */
 230         function pageCallback( $title ) {
 231                 if( isset( $this->mPageCallback ) ) {
 232                         call_user_func( $this->mPageCallback, $title );
 233                 }
 234         }
 235
 236         /**
 237          * Notify the callback function when a </page> is closed.
 238          * @param $title Title
 239          * @param $origTitle Title
 240          * @param $revCount Integer
 241          * @param $sucCount Int: number of revisions for which callback returned true
 242          * @param $pageInfo Array: associative array of page information
 243          */
 244         private function pageOutCallback( $title, $origTitle, $revCount, $sucCount, $pageInfo ) {
 245                 if( isset( $this->mPageOutCallback ) ) {
 246                         $args = func_get_args();
 247                         call_user_func_array( $this->mPageOutCallback, $args );
 248                 }
 249         }
 250
 251         /**
 252          * Notify the callback function of a revision
 253          * @param $revision A WikiRevision object
 254          */
 255         private function revisionCallback( $revision ) {
 256                 if ( isset( $this->mRevisionCallback ) ) {
 257                         return call_user_func_array( $this->mRevisionCallback,
 258                                         array( $revision, $this ) );
 259                 } else {
 260                         return false;
 261                 }
 262         }
 263
 264         /**
 265          * Notify the callback function of a new log item
 266          * @param $revision A WikiRevision object
 267          */
 268         private function logItemCallback( $revision ) {
 269                 if ( isset( $this->mLogItemCallback ) ) {
 270                         return call_user_func_array( $this->mLogItemCallback,
 271                                         array( $revision, $this ) );
 272                 } else {
 273                         return false;
 274                 }
 275         }
 276
 277         /**
 278          * Shouldn't something like this be built-in to XMLReader?
 279          * Fetches text contents of the current element, assuming
 280          * no sub-elements or such scary things.
 281          * @return string
 282          * @access private
 283          */
 284         private function nodeContents() {
 285                 if( $this->reader->isEmptyElement ) {
 286                         return "";
 287                 }
 288                 $buffer = "";
 289                 while( $this->reader->read() ) {
 290                         switch( $this->reader->nodeType ) {
 291                         case XmlReader::TEXT:
 292                         case XmlReader::SIGNIFICANT_WHITESPACE:
 293                                 $buffer .= $this->reader->value;
 294                                 break;
 295                         case XmlReader::END_ELEMENT:
 296                                 return $buffer;
 297                         }
 298                 }
 299
 300                 $this->reader->close();
 301                 return '';
 302         }
 303
 304         # --------------
 305
 306         /** Left in for debugging */
 307         private function dumpElement() {
 308                 static $lookup = null;
 309                 if (!$lookup) {
 310                         $xmlReaderConstants = array(
 311                                 "NONE",
 312                                 "ELEMENT",
 313                                 "ATTRIBUTE",
 314                                 "TEXT",
 315                                 "CDATA",
 316                                 "ENTITY_REF",
 317                                 "ENTITY",
 318                                 "PI",
 319                                 "COMMENT",
 320                                 "DOC",
 321                                 "DOC_TYPE",
 322                                 "DOC_FRAGMENT",
 323                                 "NOTATION",
 324                                 "WHITESPACE",
 325                                 "SIGNIFICANT_WHITESPACE",
 326                                 "END_ELEMENT",
 327                                 "END_ENTITY",
 328                                 "XML_DECLARATION",
 329                                 );
 330                         $lookup = array();
 331
 332                         foreach( $xmlReaderConstants as $name ) {
 333                                 $lookup[constant("XmlReader::$name")] = $name;
 334                         }
 335                 }
 336
 337                 print( var_dump(
 338                         $lookup[$this->reader->nodeType],
 339                         $this->reader->name,
 340                         $this->reader->value
 341                 )."\n\n" );
 342         }
 343
 344         /**
 345          * Primary entry point
 346          */
 347         public function doImport() {
 348                 $this->reader->read();
 349
 350                 if ( $this->reader->name != 'mediawiki' ) {
 351                         throw new MWException( "Expected <mediawiki> tag, got ".
 352                                 $this->reader->name );
 353                 }
 354                 $this->debug( "<mediawiki> tag is correct." );
 355
 356                 $this->debug( "Starting primary dump processing loop." );
 357
 358                 $keepReading = $this->reader->read();
 359                 $skip = false;
 360                 while ( $keepReading ) {
 361                         $tag = $this->reader->name;
 362                         $type = $this->reader->nodeType;
 363
 364                         if ( !wfRunHooks( 'ImportHandleToplevelXMLTag', $this ) ) {
 365                                 // Do nothing
 366                         } elseif ( $tag == 'mediawiki' && $type == XmlReader::END_ELEMENT ) {
 367                                 break;
 368                         } elseif ( $tag == 'siteinfo' ) {
 369                                 $this->handleSiteInfo();
 370                         } elseif ( $tag == 'page' ) {
 371                                 $this->handlePage();
 372                         } elseif ( $tag == 'logitem' ) {
 373                                 $this->handleLogItem();
 374                         } elseif ( $tag != '#text' ) {
 375                                 $this->warn( "Unhandled top-level XML tag $tag" );
 376
 377                                 $skip = true;
 378                         }
 379
 380                         if ($skip) {
 381                                 $keepReading = $this->reader->next();
 382                                 $skip = false;
 383                                 $this->debug( "Skip" );
 384                         } else {
 385                                 $keepReading = $this->reader->read();
 386                         }
 387                 }
 388
 389                 return true;
 390         }
 391
 392         private function handleSiteInfo() {
 393                 // Site info is useful, but not actually used for dump imports.
 394                 // Includes a quick short-circuit to save performance.
 395                 if ( ! $this->mSiteInfoCallback ) {
 396                         $this->reader->next();
 397                         return true;
 398                 }
 399                 throw new MWException( "SiteInfo tag is not yet handled, do not set mSiteInfoCallback" );
 400         }
 401
 402         private function handleLogItem() {
 403                 $this->debug( "Enter log item handler." );
 404                 $logInfo = array();
 405
 406                 // Fields that can just be stuffed in the pageInfo object
 407                 $normalFields = array( 'id', 'comment', 'type', 'action', 'timestamp',
 408                                         'logtitle', 'params' );
 409
 410                 while ( $this->reader->read() ) {
 411                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 412                                         $this->reader->name == 'logitem') {
 413                                 break;
 414                         }
 415
 416                         $tag = $this->reader->name;
 417
 418                         if ( !wfRunHooks( 'ImportHandleLogItemXMLTag',
 419                                                 $this, $logInfo ) ) {
 420                                 // Do nothing
 421                         } elseif ( in_array( $tag, $normalFields ) ) {
 422                                 $logInfo[$tag] = $this->nodeContents();
 423                         } elseif ( $tag == 'contributor' ) {
 424                                 $logInfo['contributor'] = $this->handleContributor();
 425                         } elseif ( $tag != '#text' ) {
 426                                 $this->warn( "Unhandled log-item XML tag $tag" );
 427                         }
 428                 }
 429
 430                 $this->processLogItem( $logInfo );
 431         }
 432
 433         private function processLogItem( $logInfo ) {
 434                 $revision = new WikiRevision;
 435
 436                 $revision->setID( $logInfo['id'] );
 437                 $revision->setType( $logInfo['type'] );
 438                 $revision->setAction( $logInfo['action'] );
 439                 $revision->setTimestamp( $logInfo['timestamp'] );
 440                 $revision->setParams( $logInfo['params'] );
 441                 $revision->setTitle( Title::newFromText( $logInfo['logtitle'] ) );
 442
 443                 if ( isset( $logInfo['comment'] ) ) {
 444                         $revision->setComment( $logInfo['comment'] );
 445                 }
 446
 447                 if ( isset( $logInfo['contributor']['ip'] ) ) {
 448                         $revision->setUserIP( $logInfo['contributor']['ip'] );
 449                 }
 450                 if ( isset( $logInfo['contributor']['username'] ) ) {
 451                         $revision->setUserName( $logInfo['contributor']['username'] );
 452                 }
 453
 454                 return $this->logItemCallback( $revision );
 455         }
 456
 457         private function handlePage() {
 458                 // Handle page data.
 459                 $this->debug( "Enter page handler." );
 460                 $pageInfo = array( 'revisionCount' => 0, 'successfulRevisionCount' => 0 );
 461
 462                 // Fields that can just be stuffed in the pageInfo object
 463                 $normalFields = array( 'title', 'id', 'redirect', 'restrictions' );
 464
 465                 $skip = false;
 466                 $badTitle = false;
 467
 468                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 469                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 470                                         $this->reader->name == 'page') {
 471                                 break;
 472                         }
 473
 474                         $tag = $this->reader->name;
 475
 476                         if ( $badTitle ) {
 477                                 // The title is invalid, bail out of this page
 478                                 $skip = true;
 479                         } elseif ( !wfRunHooks( 'ImportHandlePageXMLTag', array( $this,
 480                                                 &$pageInfo ) ) ) {
 481                                 // Do nothing
 482                         } elseif ( in_array( $tag, $normalFields ) ) {
 483                                 $pageInfo[$tag] = $this->nodeContents();
 484                                 if ( $tag == 'title' ) {
 485                                         $title = $this->processTitle( $pageInfo['title'] );
 486
 487                                         if ( !$title ) {
 488                                                 $badTitle = true;
 489                                                 $skip = true;
 490                                         }
 491
 492                                         $this->pageCallback( $title );
 493                                         list( $pageInfo['_title'], $origTitle ) = $title;
 494                                 }
 495                         } elseif ( $tag == 'revision' ) {
 496                                 $this->handleRevision( $pageInfo );
 497                         } elseif ( $tag == 'upload' ) {
 498                                 $this->handleUpload( $pageInfo );
 499                         } elseif ( $tag != '#text' ) {
 500                                 $this->warn( "Unhandled page XML tag $tag" );
 501                                 $skip = true;
 502                         }
 503                 }
 504
 505                 $this->pageOutCallback( $pageInfo['_title'], $origTitle,
 506                                         $pageInfo['revisionCount'],
 507                                         $pageInfo['successfulRevisionCount'],
 508                                         $pageInfo );
 509         }
 510
 511         private function handleRevision( &$pageInfo ) {
 512                 $this->debug( "Enter revision handler" );
 513                 $revisionInfo = array();
 514
 515                 $normalFields = array( 'id', 'timestamp', 'comment', 'minor', 'text' );
 516
 517                 $skip = false;
 518
 519                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 520                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 521                                         $this->reader->name == 'revision') {
 522                                 break;
 523                         }
 524
 525                         $tag = $this->reader->name;
 526
 527                         if ( !wfRunHooks( 'ImportHandleRevisionXMLTag', $this,
 528                                                 $pageInfo, $revisionInfo ) ) {
 529                                 // Do nothing
 530                         } elseif ( in_array( $tag, $normalFields ) ) {
 531                                 $revisionInfo[$tag] = $this->nodeContents();
 532                         } elseif ( $tag == 'contributor' ) {
 533                                 $revisionInfo['contributor'] = $this->handleContributor();
 534                         } elseif ( $tag != '#text' ) {
 535                                 $this->warn( "Unhandled revision XML tag $tag" );
 536                                 $skip = true;
 537                         }
 538                 }
 539
 540                 $pageInfo['revisionCount']++;
 541                 if ( $this->processRevision( $pageInfo, $revisionInfo ) ) {
 542                         $pageInfo['successfulRevisionCount']++;
 543                 }
 544         }
 545
 546         private function processRevision( $pageInfo, $revisionInfo ) {
 547                 $revision = new WikiRevision;
 548
 549                 $revision->setID( $revisionInfo['id'] );
 550                 $revision->setText( $revisionInfo['text'] );
 551                 $revision->setTitle( $pageInfo['_title'] );
 552                 $revision->setTimestamp( $revisionInfo['timestamp'] );
 553
 554                 if ( isset( $revisionInfo['comment'] ) ) {
 555                         $revision->setComment( $revisionInfo['comment'] );
 556                 }
 557
 558                 if ( isset( $revisionInfo['minor'] ) )
 559                         $revision->setMinor( true );
 560
 561                 if ( isset( $revisionInfo['contributor']['ip'] ) ) {
 562                         $revision->setUserIP( $revisionInfo['contributor']['ip'] );
 563                 }
 564                 if ( isset( $revisionInfo['contributor']['username'] ) ) {
 565                         $revision->setUserName( $revisionInfo['contributor']['username'] );
 566                 }
 567
 568                 return $this->revisionCallback( $revision );
 569         }
 570
 571         private function handleUpload( &$pageInfo ) {
 572                 $this->debug( "Enter upload handler" );
 573                 $uploadInfo = array();
 574
 575                 $normalFields = array( 'timestamp', 'comment', 'filename', 'text',
 576                                         'src', 'size' );
 577
 578                 $skip = false;
 579
 580                 while ( $skip ? $this->reader->next() : $this->reader->read() ) {
 581                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 582                                         $this->reader->name == 'upload') {
 583                                 break;
 584                         }
 585
 586                         $tag = $this->reader->name;
 587
 588                         if ( !wfRunHooks( 'ImportHandleUploadXMLTag', $this,
 589                                                 $pageInfo ) ) {
 590                                 // Do nothing
 591                         } elseif ( in_array( $tag, $normalFields ) ) {
 592                                 $uploadInfo[$tag] = $this->nodeContents();
 593                         } elseif ( $tag == 'contributor' ) {
 594                                 $uploadInfo['contributor'] = $this->handleContributor();
 595                         } elseif ( $tag != '#text' ) {
 596                                 $this->warn( "Unhandled upload XML tag $tag" );
 597                                 $skip = true;
 598                         }
 599                 }
 600
 601                 return $this->processUpload( $pageInfo, $uploadInfo );
 602         }
 603
 604
 605         private function processUpload( $pageInfo, $uploadInfo ) {
 606                 $revision = new WikiRevision;
 607                            $text = isset( $uploadInfo['text'] ) ? $uploadInfo['text'] : '';
 608
 609                 $revision->setTitle( $pageInfo['_title'] );
 610                            $revision->setID( $pageInfo['id'] );
 611                 $revision->setTimestamp( $uploadInfo['timestamp'] );
 612                            $revision->setText( $text );
 613                 $revision->setFilename( $uploadInfo['filename'] );
 614                 $revision->setSrc( $uploadInfo['src'] );
 615                 $revision->setSize( intval( $uploadInfo['size'] ) );
 616                 $revision->setComment( $uploadInfo['comment'] );
 617
 618                 if ( isset( $uploadInfo['contributor']['ip'] ) ) {
 619                         $revision->setUserIP( $uploadInfo['contributor']['ip'] );
 620                 }
 621                 if ( isset( $uploadInfo['contributor']['username'] ) ) {
 622                         $revision->setUserName( $uploadInfo['contributor']['username'] );
 623                 }
 624
 625                            return call_user_func( $this->mUploadCallback, $revision );
 626         }
 627
 628         private function handleContributor() {
 629                 $fields = array( 'id', 'ip', 'username' );
 630                 $info = array();
 631
 632                 while ( $this->reader->read() ) {
 633                         if ( $this->reader->nodeType == XmlReader::END_ELEMENT &&
 634                                         $this->reader->name == 'contributor') {
 635                                 break;
 636                         }
 637
 638                         $tag = $this->reader->name;
 639
 640                         if ( in_array( $tag, $fields ) ) {
 641                                 $info[$tag] = $this->nodeContents();
 642                         }
 643                 }
 644
 645                 return $info;
 646         }
 647
 648         private function processTitle( $text ) {
 649                 $workTitle = $text;
 650                 $origTitle = Title::newFromText( $workTitle );
 651
 652                 if( !is_null( $this->mTargetNamespace ) && !is_null( $origTitle ) ) {
 653                         $title = Title::makeTitle( $this->mTargetNamespace,
 654                                 $origTitle->getDBkey() );
 655                 } else {
 656                         $title = Title::newFromText( $workTitle );
 657                 }
 658
 659                 if( is_null( $title ) ) {
 660                         // Invalid page title? Ignore the page
 661                         $this->notice( "Skipping invalid page title '$workTitle'" );
 662                         return false;
 663                 } elseif( $title->getInterwiki() != '' ) {
 664                         $this->notice( "Skipping interwiki page title '$workTitle'" );
 665                         return false;
 666                 }
 667
 668                 return array( $title, $origTitle );
 669         }
 670 }
 671
 672 /** This is a horrible hack used to keep source compatibility */
 673 class UploadSourceAdapter {
 674         static $sourceRegistrations = array();
 675
 676         private $mSource;
 677         private $mBuffer;
 678         private $mPosition;
 679
 680         static function registerSource( $source ) {
 681                 $id = wfGenerateToken();
 682
 683                 self::$sourceRegistrations[$id] = $source;
 684
 685                 return $id;
 686         }
 687
 688         function stream_open( $path, $mode, $options, &$opened_path ) {
 689                 $url = parse_url($path);
 690                 $id = $url['host'];
 691
 692                 if ( !isset( self::$sourceRegistrations[$id] ) ) {
 693                         return false;
 694                 }
 695
 696                 $this->mSource = self::$sourceRegistrations[$id];
 697
 698                 return true;
 699         }
 700
 701         function stream_read( $count ) {
 702                 $return = '';
 703                 $leave = false;
 704
 705                 while ( !$leave && !$this->mSource->atEnd() &&
 706                                 strlen($this->mBuffer) < $count ) {
 707                         $read = $this->mSource->readChunk();
 708
 709                         if ( !strlen($read) ) {
 710                                 $leave = true;
 711                         }
 712
 713                         $this->mBuffer .= $read;
 714                 }
 715
 716                 if ( strlen($this->mBuffer) ) {
 717                         $return = substr( $this->mBuffer, 0, $count );
 718                         $this->mBuffer = substr( $this->mBuffer, $count );
 719                 }
 720
 721                 $this->mPosition += strlen($return);
 722
 723                 return $return;
 724         }
 725
 726         function stream_write( $data ) {
 727                 return false;
 728         }
 729
 730         function stream_tell() {
 731                 return $this->mPosition;
 732         }
 733
 734         function stream_eof() {
 735                 return $this->mSource->atEnd();
 736         }
 737
 738         function url_stat() {
 739                 $result = array();
 740
 741                 $result['dev'] = $result[0] = 0;
 742                 $result['ino'] = $result[1] = 0;
 743                 $result['mode'] = $result[2] = 0;
 744                 $result['nlink'] = $result[3] = 0;
 745                 $result['uid'] = $result[4] = 0;
 746                 $result['gid'] = $result[5] = 0;
 747                 $result['rdev'] = $result[6] = 0;
 748                 $result['size'] = $result[7] = 0;
 749                 $result['atime'] = $result[8] = 0;
 750                 $result['mtime'] = $result[9] = 0;
 751                 $result['ctime'] = $result[10] = 0;
 752                 $result['blksize'] = $result[11] = 0;
 753                 $result['blocks'] = $result[12] = 0;
 754
 755                 return $result;
 756         }
 757 }
 758
 759 class XMLReader2 extends XMLReader {
 760         function nodeContents() {
 761                 if( $this->isEmptyElement ) {
 762                         return "";
 763                 }
 764                 $buffer = "";
 765                 while( $this->read() ) {
 766                         switch( $this->nodeType ) {
 767                         case XmlReader::TEXT:
 768                         case XmlReader::SIGNIFICANT_WHITESPACE:
 769                                 $buffer .= $this->value;
 770                                 break;
 771                         case XmlReader::END_ELEMENT:
 772                                 return $buffer;
 773                         }
 774                 }
 775                 return $this->close();
 776         }
 777 }
 778
 779 /**
 780  * @todo document (e.g. one-sentence class description).
 781  * @ingroup SpecialPage
 782  */
 783 class WikiRevision {
 784         var $title = null;
 785         var $id = 0;
 786         var $timestamp = "20010115000000";
 787         var $user = 0;
 788         var $user_text = "";
 789         var $text = "";
 790         var $comment = "";
 791         var $minor = false;
 792         var $type = "";
 793         var $action = "";
 794         var $params = "";
 795
 796         function setTitle( $title ) {
 797                 if( is_object( $title ) ) {
 798                         $this->title = $title;
 799                 } elseif( is_null( $title ) ) {
 800                         throw new MWException( "WikiRevision given a null title in import. You may need to adjust \$wgLegalTitleChars." );
 801                 } else {
 802                         throw new MWException( "WikiRevision given non-object title in import." );
 803                 }
 804         }
 805
 806         function setID( $id ) {
 807                 $this->id = $id;
 808         }
 809
 810         function setTimestamp( $ts ) {
 811                 # 2003-08-05T18:30:02Z
 812                 $this->timestamp = wfTimestamp( TS_MW, $ts );
 813         }
 814
 815         function setUsername( $user ) {
 816                 $this->user_text = $user;
 817         }
 818
 819         function setUserIP( $ip ) {
 820                 $this->user_text = $ip;
 821         }
 822
 823         function setText( $text ) {
 824                 $this->text = $text;
 825         }
 826
 827         function setComment( $text ) {
 828                 $this->comment = $text;
 829         }
 830
 831         function setMinor( $minor ) {
 832                 $this->minor = (bool)$minor;
 833         }
 834
 835         function setSrc( $src ) {
 836                 $this->src = $src;
 837         }
 838
 839         function setFilename( $filename ) {
 840                 $this->filename = $filename;
 841         }
 842
 843         function setSize( $size ) {
 844                 $this->size = intval( $size );
 845         }
 846
 847         function setType( $type ) {
 848                 $this->type = $type;
 849         }
 850
 851         function setAction( $action ) {
 852                 $this->action = $action;
 853         }
 854
 855         function setParams( $params ) {
 856                 $this->params = $params;
 857         }
 858
 859         function getTitle() {
 860                 return $this->title;
 861         }
 862
 863         function getID() {
 864                 return $this->id;
 865         }
 866
 867         function getTimestamp() {
 868                 return $this->timestamp;
 869         }
 870
 871         function getUser() {
 872                 return $this->user_text;
 873         }
 874
 875         function getText() {
 876                 return $this->text;
 877         }
 878
 879         function getComment() {
 880                 return $this->comment;
 881         }
 882
 883         function getMinor() {
 884                 return $this->minor;
 885         }
 886
 887         function getSrc() {
 888                 return $this->src;
 889         }
 890
 891         function getFilename() {
 892                 return $this->filename;
 893         }
 894
 895         function getSize() {
 896                 return $this->size;
 897         }
 898
 899         function getType() {
 900                 return $this->type;
 901         }
 902
 903         function getAction() {
 904                 return $this->action;
 905         }
 906
 907         function getParams() {
 908                 return $this->params;
 909         }
 910
 911         function importOldRevision() {
 912                 $dbw = wfGetDB( DB_MASTER );
 913
 914                 # Sneak a single revision into place
 915                 $user = User::newFromName( $this->getUser() );
 916                 if( $user ) {
 917                         $userId = intval( $user->getId() );
 918                         $userText = $user->getName();
 919                 } else {
 920                         $userId = 0;
 921                         $userText = $this->getUser();
 922                 }
 923
 924                 // avoid memory leak...?
 925                 $linkCache = LinkCache::singleton();
 926                 $linkCache->clear();
 927
 928                 $article = new Article( $this->title );
 929                 $pageId = $article->getId();
 930                 if( $pageId == 0 ) {
 931                         # must create the page...
 932                         $pageId = $article->insertOn( $dbw );
 933                         $created = true;
 934                 } else {
 935                         $created = false;
 936
 937                         $prior = $dbw->selectField( 'revision', '1',
 938                                 array( 'rev_page' => $pageId,
 939                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
 940                                         'rev_user_text' => $userText,
 941                                         'rev_comment'   => $this->getComment() ),
 942                                 __METHOD__
 943                         );
 944                         if( $prior ) {
 945                                 // FIXME: this could fail slightly for multiple matches :P
 946                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
 947                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
 948                                 return false;
 949                         }
 950                 }
 951
 952                 # FIXME: Use original rev_id optionally (better for backups)
 953                 # Insert the row
 954                 $revision = new Revision( array(
 955                         'page'       => $pageId,
 956                         'text'       => $this->getText(),
 957                         'comment'    => $this->getComment(),
 958                         'user'       => $userId,
 959                         'user_text'  => $userText,
 960                         'timestamp'  => $this->timestamp,
 961                         'minor_edit' => $this->minor,
 962                         ) );
 963                 $revId = $revision->insertOn( $dbw );
 964                 $changed = $article->updateIfNewerOn( $dbw, $revision );
 965
 966                 # To be on the safe side...
 967                 $tempTitle = $GLOBALS['wgTitle'];
 968                 $GLOBALS['wgTitle'] = $this->title;
 969
 970                 if( $created ) {
 971                         wfDebug( __METHOD__ . ": running onArticleCreate\n" );
 972                         Article::onArticleCreate( $this->title );
 973
 974                         wfDebug( __METHOD__ . ": running create updates\n" );
 975                         $article->createUpdates( $revision );
 976
 977                 } elseif( $changed ) {
 978                         wfDebug( __METHOD__ . ": running onArticleEdit\n" );
 979                         Article::onArticleEdit( $this->title );
 980
 981                         wfDebug( __METHOD__ . ": running edit updates\n" );
 982                         $article->editUpdates(
 983                                 $this->getText(),
 984                                 $this->getComment(),
 985                                 $this->minor,
 986                                 $this->timestamp,
 987                                 $revId );
 988                 }
 989                 $GLOBALS['wgTitle'] = $tempTitle;
 990
 991                 return true;
 992         }
 993
 994         function importLogItem() {
 995                 $dbw = wfGetDB( DB_MASTER );
 996                 # FIXME: this will not record autoblocks
 997                 if( !$this->getTitle() ) {
 998                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
 999                                 $this->timestamp . "\n" );
1000                         return;
1001                 }
1002                 # Check if it exists already
1003                 // FIXME: use original log ID (better for backups)
1004                 $prior = $dbw->selectField( 'logging', '1',
1005                         array( 'log_type' => $this->getType(),
1006                                 'log_action'    => $this->getAction(),
1007                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1008                                 'log_namespace' => $this->getTitle()->getNamespace(),
1009                                 'log_title'     => $this->getTitle()->getDBkey(),
1010                                 'log_comment'   => $this->getComment(),
1011                                 #'log_user_text' => $this->user_text,
1012                                 'log_params'    => $this->params ),
1013                         __METHOD__
1014                 );
1015                 // FIXME: this could fail slightly for multiple matches :P
1016                 if( $prior ) {
1017                         wfDebug( __METHOD__ . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp " .
1018                                 $this->timestamp . "\n" );
1019                         return false;
1020                 }
1021                 $log_id = $dbw->nextSequenceValue( 'logging_log_id_seq' );
1022                 $data = array(
1023                         'log_id' => $log_id,
1024                         'log_type' => $this->type,
1025                         'log_action' => $this->action,
1026                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
1027                         'log_user' => User::idFromName( $this->user_text ),
1028                         #'log_user_text' => $this->user_text,
1029                         'log_namespace' => $this->getTitle()->getNamespace(),
1030                         'log_title' => $this->getTitle()->getDBkey(),
1031                         'log_comment' => $this->getComment(),
1032                         'log_params' => $this->params
1033                 );
1034                 $dbw->insert( 'logging', $data, __METHOD__ );
1035         }
1036
1037         function importUpload() {
1038                 wfDebug( __METHOD__ . ": STUB\n" );
1039
1040                 /**
1041                         // from file revert...
1042                         $source = $this->file->getArchiveVirtualUrl( $this->oldimage );
1043                         $comment = $wgRequest->getText( 'wpComment' );
1044                         // TODO: Preserve file properties from database instead of reloading from file
1045                         $status = $this->file->upload( $source, $comment, $comment );
1046                         if( $status->isGood() ) {
1047                 */
1048
1049                 /**
1050                         // from file upload...
1051                 $this->mLocalFile = wfLocalFile( $nt );
1052                 $this->mDestName = $this->mLocalFile->getName();
1053                 //....
1054                         $status = $this->mLocalFile->upload( $this->mTempPath, $this->mComment, $pageText,
1055                         File::DELETE_SOURCE, $this->mFileProps );
1056                         if ( !$status->isGood() ) {
1057                                 $resultDetails = array( 'internal' => $status->getWikiText() );
1058                 */
1059
1060                            // @todo Fixme: it may create a page without our desire, also wrong potentially.
1061                 // and, it will record a *current* upload, but we might want an archive version here
1062
1063                 $file = wfLocalFile( $this->getTitle() );
1064                 if( !$file ) {
1065                         wfDebug( "IMPORT: Bad file. :(\n" );
1066                         return false;
1067                 }
1068
1069                 $source = $this->downloadSource();
1070                 if( !$source ) {
1071                         wfDebug( "IMPORT: Could not fetch remote file. :(\n" );
1072                         return false;
1073                 }
1074
1075                 $user = User::newFromName( $this->user_text );
1076
1077                 $status = $file->upload( $source,
1078                         $this->getComment(),
1079                         $this->getComment(), // Initial page, if none present...
1080                         File::DELETE_SOURCE,
1081                         false, // props...
1082                                            $this->getTimestamp(),
1083                                            is_object( $user ) ? ( $user->isLoggedIn() ? $user : null ) : null );
1084
1085                 if( $status->isGood() ) {
1086                         // yay?
1087                         wfDebug( "IMPORT: is ok?\n" );
1088                         return true;
1089                 }
1090
1091                 wfDebug( "IMPORT: is bad? " . $status->getXml() . "\n" );
1092                 return false;
1093
1094         }
1095
1096         function downloadSource() {
1097                 global $wgEnableUploads;
1098                 if( !$wgEnableUploads ) {
1099                         return false;
1100                 }
1101
1102                 $tempo = tempnam( wfTempDir(), 'download' );
1103                 $f = fopen( $tempo, 'wb' );
1104                 if( !$f ) {
1105                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
1106                         return false;
1107                 }
1108
1109                 // @todo Fixme!
1110                 $src = $this->getSrc();
1111                 $data = Http::get( $src );
1112                 if( !$data ) {
1113                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
1114                         fclose( $f );
1115                         unlink( $tempo );
1116                         return false;
1117                 }
1118
1119                 fwrite( $f, $data );
1120                 fclose( $f );
1121
1122                 return $tempo;
1123         }
1124
1125 }
1126
1127 /**
1128  * @todo document (e.g. one-sentence class description).
1129  * @ingroup SpecialPage
1130  */
1131 class ImportStringSource {
1132         function __construct( $string ) {
1133                 $this->mString = $string;
1134                 $this->mRead = false;
1135         }
1136
1137         function atEnd() {
1138                 return $this->mRead;
1139         }
1140
1141         function readChunk() {
1142                 if( $this->atEnd() ) {
1143                         return false;
1144                 } else {
1145                         $this->mRead = true;
1146                         return $this->mString;
1147                 }
1148         }
1149 }
1150
1151 /**
1152  * @todo document (e.g. one-sentence class description).
1153  * @ingroup SpecialPage
1154  */
1155 class ImportStreamSource {
1156         function __construct( $handle ) {
1157                 $this->mHandle = $handle;
1158         }
1159
1160         function atEnd() {
1161                 return feof( $this->mHandle );
1162         }
1163
1164         function readChunk() {
1165                 return fread( $this->mHandle, 32768 );
1166         }
1167
1168         static function newFromFile( $filename ) {
1169                 $file = @fopen( $filename, 'rt' );
1170                 if( !$file ) {
1171                         return Status::newFatal( "importcantopen" );
1172                 }
1173                 return Status::newGood( new ImportStreamSource( $file ) );
1174         }
1175
1176         static function newFromUpload( $fieldname = "xmlimport" ) {
1177                 $upload =& $_FILES[$fieldname];
1178
1179                 if( !isset( $upload ) || !$upload['name'] ) {
1180                         return Status::newFatal( 'importnofile' );
1181                 }
1182                 if( !empty( $upload['error'] ) ) {
1183                         switch($upload['error']){
1184                                 case 1: # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1185                                         return Status::newFatal( 'importuploaderrorsize' );
1186                                 case 2: # The uploaded file exceeds the MAX_FILE_SIZE directive that was specified in the HTML form.
1187                                         return Status::newFatal( 'importuploaderrorsize' );
1188                                 case 3: # The uploaded file was only partially uploaded
1189                                         return Status::newFatal( 'importuploaderrorpartial' );
1190                                 case 6: #Missing a temporary folder.
1191                                         return Status::newFatal( 'importuploaderrortemp' );
1192                                 # case else: # Currently impossible
1193                         }
1194
1195                 }
1196                 $fname = $upload['tmp_name'];
1197                 if( is_uploaded_file( $fname ) ) {
1198                         return ImportStreamSource::newFromFile( $fname );
1199                 } else {
1200                         return Status::newFatal( 'importnofile' );
1201                 }
1202         }
1203
1204         static function newFromURL( $url, $method = 'GET' ) {
1205                 wfDebug( __METHOD__ . ": opening $url\n" );
1206                 # Use the standard HTTP fetch function; it times out
1207                 # quicker and sorts out user-agent problems which might
1208                 # otherwise prevent importing from large sites, such
1209                 # as the Wikimedia cluster, etc.
1210                 $data = Http::request( $method, $url );
1211                 if( $data !== false ) {
1212                         $file = tmpfile();
1213                         fwrite( $file, $data );
1214                         fflush( $file );
1215                         fseek( $file, 0 );
1216                         return Status::newGood( new ImportStreamSource( $file ) );
1217                 } else {
1218                         return Status::newFatal( 'importcantopen' );
1219                 }
1220         }
1221
1222         public static function newFromInterwiki( $interwiki, $page, $history = false, $templates = false, $pageLinkDepth = 0 ) {
1223                 if( $page == '' ) {
1224                         return Status::newFatal( 'import-noarticle' );
1225                 }
1226                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1227                 if( is_null( $link ) || $link->getInterwiki() == '' ) {
1228                         return Status::newFatal( 'importbadinterwiki' );
1229                 } else {
1230                         $params = array();
1231                         if ( $history ) $params['history'] = 1;
1232                         if ( $templates ) $params['templates'] = 1;
1233                         if ( $pageLinkDepth ) $params['pagelink-depth'] = $pageLinkDepth;
1234                         $url = $link->getFullUrl( $params );
1235                         # For interwikis, use POST to avoid redirects.
1236                         return ImportStreamSource::newFromURL( $url, "POST" );
1237                 }
1238         }
1239 }