includes/Import.php

   1 <?php
   2 /**
   3  * MediaWiki page data importer
   4  * Copyright (C) 2003,2005 Brion Vibber <brion@pobox.com>
   5  * http://www.mediawiki.org/
   6  *
   7  * This program is free software; you can redistribute it and/or modify
   8  * it under the terms of the GNU General Public License as published by
   9  * the Free Software Foundation; either version 2 of the License, or
  10  * (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  15  * GNU General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU General Public License along
  18  * with this program; if not, write to the Free Software Foundation, Inc.,
  19  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  20  * http://www.gnu.org/copyleft/gpl.html
  21  *
  22  * @file
  23  * @ingroup SpecialPage
  24  */
  25
  26 /**
  27  *
  28  * @ingroup SpecialPage
  29  */
  30 class WikiRevision {
  31         var $title = null;
  32         var $id = 0;
  33         var $timestamp = "20010115000000";
  34         var $user = 0;
  35         var $user_text = "";
  36         var $text = "";
  37         var $comment = "";
  38         var $minor = false;
  39         var $type = "";
  40         var $action = "";
  41         var $params = "";
  42
  43         function setTitle( $title ) {
  44                 if( is_object( $title ) ) {
  45                         $this->title = $title;
  46                 } elseif( is_null( $title ) ) {
  47                         throw new MWException( "WikiRevision given a null title in import. You may need to adjust \$wgLegalTitleChars." );
  48                 } else {
  49                         throw new MWException( "WikiRevision given non-object title in import." );
  50                 }
  51         }
  52
  53         function setID( $id ) {
  54                 $this->id = $id;
  55         }
  56
  57         function setTimestamp( $ts ) {
  58                 # 2003-08-05T18:30:02Z
  59                 $this->timestamp = wfTimestamp( TS_MW, $ts );
  60         }
  61
  62         function setUsername( $user ) {
  63                 $this->user_text = $user;
  64         }
  65
  66         function setUserIP( $ip ) {
  67                 $this->user_text = $ip;
  68         }
  69
  70         function setText( $text ) {
  71                 $this->text = $text;
  72         }
  73
  74         function setComment( $text ) {
  75                 $this->comment = $text;
  76         }
  77
  78         function setMinor( $minor ) {
  79                 $this->minor = (bool)$minor;
  80         }
  81
  82         function setSrc( $src ) {
  83                 $this->src = $src;
  84         }
  85
  86         function setFilename( $filename ) {
  87                 $this->filename = $filename;
  88         }
  89
  90         function setSize( $size ) {
  91                 $this->size = intval( $size );
  92         }
  93
  94         function setType( $type ) {
  95                 $this->type = $type;
  96         }
  97
  98         function setAction( $action ) {
  99                 $this->action = $action;
 100         }
 101
 102         function setParams( $params ) {
 103                 $this->params = $params;
 104         }
 105
 106         function getTitle() {
 107                 return $this->title;
 108         }
 109
 110         function getID() {
 111                 return $this->id;
 112         }
 113
 114         function getTimestamp() {
 115                 return $this->timestamp;
 116         }
 117
 118         function getUser() {
 119                 return $this->user_text;
 120         }
 121
 122         function getText() {
 123                 return $this->text;
 124         }
 125
 126         function getComment() {
 127                 return $this->comment;
 128         }
 129
 130         function getMinor() {
 131                 return $this->minor;
 132         }
 133
 134         function getSrc() {
 135                 return $this->src;
 136         }
 137
 138         function getFilename() {
 139                 return $this->filename;
 140         }
 141
 142         function getSize() {
 143                 return $this->size;
 144         }
 145
 146         function getType() {
 147                 return $this->type;
 148         }
 149
 150         function getAction() {
 151                 return $this->action;
 152         }
 153
 154         function getParams() {
 155                 return $this->params;
 156         }
 157
 158         function importOldRevision() {
 159                 $dbw = wfGetDB( DB_MASTER );
 160
 161                 # Sneak a single revision into place
 162                 $user = User::newFromName( $this->getUser() );
 163                 if( $user ) {
 164                         $userId = intval( $user->getId() );
 165                         $userText = $user->getName();
 166                 } else {
 167                         $userId = 0;
 168                         $userText = $this->getUser();
 169                 }
 170
 171                 // avoid memory leak...?
 172                 $linkCache = LinkCache::singleton();
 173                 $linkCache->clear();
 174
 175                 $article = new Article( $this->title );
 176                 $pageId = $article->getId();
 177                 if( $pageId == 0 ) {
 178                         # must create the page...
 179                         $pageId = $article->insertOn( $dbw );
 180                         $created = true;
 181                 } else {
 182                         $created = false;
 183
 184                         $prior = $dbw->selectField( 'revision', '1',
 185                                 array( 'rev_page' => $pageId,
 186                                         'rev_timestamp' => $dbw->timestamp( $this->timestamp ),
 187                                         'rev_user_text' => $userText,
 188                                         'rev_comment'   => $this->getComment() ),
 189                                 __METHOD__
 190                         );
 191                         if( $prior ) {
 192                                 // FIXME: this could fail slightly for multiple matches :P
 193                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
 194                                         $this->title->getPrefixedText() . "]], timestamp " . $this->timestamp . "\n" );
 195                                 return false;
 196                         }
 197                 }
 198
 199                 # FIXME: Use original rev_id optionally (better for backups)
 200                 # Insert the row
 201                 $revision = new Revision( array(
 202                         'page'       => $pageId,
 203                         'text'       => $this->getText(),
 204                         'comment'    => $this->getComment(),
 205                         'user'       => $userId,
 206                         'user_text'  => $userText,
 207                         'timestamp'  => $this->timestamp,
 208                         'minor_edit' => $this->minor,
 209                         ) );
 210                 $revId = $revision->insertOn( $dbw );
 211                 $changed = $article->updateIfNewerOn( $dbw, $revision );
 212
 213                 # To be on the safe side...
 214                 $tempTitle = $GLOBALS['wgTitle'];
 215                 $GLOBALS['wgTitle'] = $this->title;
 216
 217                 if( $created ) {
 218                         wfDebug( __METHOD__ . ": running onArticleCreate\n" );
 219                         Article::onArticleCreate( $this->title );
 220
 221                         wfDebug( __METHOD__ . ": running create updates\n" );
 222                         $article->createUpdates( $revision );
 223
 224                 } elseif( $changed ) {
 225                         wfDebug( __METHOD__ . ": running onArticleEdit\n" );
 226                         Article::onArticleEdit( $this->title, 'skiptransclusions' ); // leave templatelinks for editUpdates()
 227
 228                         wfDebug( __METHOD__ . ": running edit updates\n" );
 229                         $article->editUpdates(
 230                                 $this->getText(),
 231                                 $this->getComment(),
 232                                 $this->minor,
 233                                 $this->timestamp,
 234                                 $revId );
 235                 }
 236                 $GLOBALS['wgTitle'] = $tempTitle;
 237
 238                 return true;
 239         }
 240
 241         function importLogItem() {
 242                 $dbw = wfGetDB( DB_MASTER );
 243                 # FIXME: this will not record autoblocks
 244                 if( !$this->getTitle() ) {
 245                         wfDebug( __METHOD__ . ": skipping invalid {$this->type}/{$this->action} log time, timestamp " .
 246                                 $this->timestamp . "\n" );
 247                         return;
 248                 }
 249                 # Check if it exists already
 250                 // FIXME: use original log ID (better for backups)
 251                 $prior = $dbw->selectField( 'logging', '1',
 252                         array( 'log_type' => $this->getType(),
 253                                 'log_action'    => $this->getAction(),
 254                                 'log_timestamp' => $dbw->timestamp( $this->timestamp ),
 255                                 'log_namespace' => $this->getTitle()->getNamespace(),
 256                                 'log_title'     => $this->getTitle()->getDBkey(),
 257                                 'log_comment'   => $this->getComment(),
 258                                 #'log_user_text' => $this->user_text,
 259                                 'log_params'    => $this->params ),
 260                         __METHOD__
 261                 );
 262                 // FIXME: this could fail slightly for multiple matches :P
 263                 if( $prior ) {
 264                         wfDebug( __METHOD__ . ": skipping existing item for Log:{$this->type}/{$this->action}, timestamp " .
 265                                 $this->timestamp . "\n" );
 266                         return false;
 267                 }
 268                 $log_id = $dbw->nextSequenceValue( 'log_log_id_seq' );
 269                 $data = array(
 270                         'log_id' => $log_id,
 271                         'log_type' => $this->type,
 272                         'log_action' => $this->action,
 273                         'log_timestamp' => $dbw->timestamp( $this->timestamp ),
 274                         'log_user' => User::idFromName( $this->user_text ),
 275                         #'log_user_text' => $this->user_text,
 276                         'log_namespace' => $this->getTitle()->getNamespace(),
 277                         'log_title' => $this->getTitle()->getDBkey(),
 278                         'log_comment' => $this->getComment(),
 279                         'log_params' => $this->params
 280                 );
 281                 $dbw->insert( 'logging', $data, __METHOD__ );
 282         }
 283
 284         function importUpload() {
 285                 wfDebug( __METHOD__ . ": STUB\n" );
 286
 287                 /**
 288                         // from file revert...
 289                         $source = $this->file->getArchiveVirtualUrl( $this->oldimage );
 290                         $comment = $wgRequest->getText( 'wpComment' );
 291                         // TODO: Preserve file properties from database instead of reloading from file
 292                         $status = $this->file->upload( $source, $comment, $comment );
 293                         if( $status->isGood() ) {
 294                 */
 295
 296                 /**
 297                         // from file upload...
 298                 $this->mLocalFile = wfLocalFile( $nt );
 299                 $this->mDestName = $this->mLocalFile->getName();
 300                 //....
 301                         $status = $this->mLocalFile->upload( $this->mTempPath, $this->mComment, $pageText,
 302                         File::DELETE_SOURCE, $this->mFileProps );
 303                         if ( !$status->isGood() ) {
 304                                 $resultDetails = array( 'internal' => $status->getWikiText() );
 305                 */
 306
 307                 // @fixme upload() uses $wgUser, which is wrong here
 308                 // it may also create a page without our desire, also wrong potentially.
 309                 // and, it will record a *current* upload, but we might want an archive version here
 310
 311                 $file = wfLocalFile( $this->getTitle() );
 312                 if( !$file ) {
 313                         var_dump( $file );
 314                         wfDebug( "IMPORT: Bad file. :(\n" );
 315                         return false;
 316                 }
 317
 318                 $source = $this->downloadSource();
 319                 if( !$source ) {
 320                         wfDebug( "IMPORT: Could not fetch remote file. :(\n" );
 321                         return false;
 322                 }
 323
 324                 $status = $file->upload( $source,
 325                         $this->getComment(),
 326                         $this->getComment(), // Initial page, if none present...
 327                         File::DELETE_SOURCE,
 328                         false, // props...
 329                         $this->getTimestamp() );
 330
 331                 if( $status->isGood() ) {
 332                         // yay?
 333                         wfDebug( "IMPORT: is ok?\n" );
 334                         return true;
 335                 }
 336
 337                 wfDebug( "IMPORT: is bad? " . $status->getXml() . "\n" );
 338                 return false;
 339
 340         }
 341
 342         function downloadSource() {
 343                 global $wgEnableUploads;
 344                 if( !$wgEnableUploads ) {
 345                         return false;
 346                 }
 347
 348                 $tempo = tempnam( wfTempDir(), 'download' );
 349                 $f = fopen( $tempo, 'wb' );
 350                 if( !$f ) {
 351                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
 352                         return false;
 353                 }
 354
 355                 // @fixme!
 356                 $src = $this->getSrc();
 357                 $data = Http::get( $src );
 358                 if( !$data ) {
 359                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
 360                         fclose( $f );
 361                         unlink( $tempo );
 362                         return false;
 363                 }
 364
 365                 fwrite( $f, $data );
 366                 fclose( $f );
 367
 368                 return $tempo;
 369         }
 370
 371 }
 372
 373 /**
 374  * implements Special:Import
 375  * @ingroup SpecialPage
 376  */
 377 class WikiImporter {
 378         var $mDebug = false;
 379         var $mSource = null;
 380         var $mPageCallback = null;
 381         var $mPageOutCallback = null;
 382         var $mRevisionCallback = null;
 383         var $mLogItemCallback = null;
 384         var $mUploadCallback = null;
 385         var $mTargetNamespace = null;
 386         var $mXmlNamespace = false;
 387         var $lastfield;
 388         var $tagStack = array();
 389
 390         function __construct( $source ) {
 391                 $this->setRevisionCallback( array( $this, "importRevision" ) );
 392                 $this->setUploadCallback( array( $this, "importUpload" ) );
 393                 $this->setLogItemCallback( array( $this, "importLogItem" ) );
 394                 $this->mSource = $source;
 395         }
 396
 397         function throwXmlError( $err ) {
 398                 $this->debug( "FAILURE: $err" );
 399                 wfDebug( "WikiImporter XML error: $err\n" );
 400         }
 401
 402         function handleXmlNamespace ( $parser, $data, $prefix=false, $uri=false ) {
 403                  if( preg_match( '/www.mediawiki.org/',$prefix ) ) {
 404                         $prefix = str_replace( '/','\/',$prefix );
 405                         $this->mXmlNamespace='/^'.$prefix.':/';
 406                  }
 407         }
 408
 409         function stripXmlNamespace($name) {
 410                 if( $this->mXmlNamespace ) {
 411                        return(preg_replace($this->mXmlNamespace,'',$name,1));
 412                 }
 413                 else {
 414                        return($name);
 415                 }
 416         }
 417
 418         # --------------
 419
 420         function doImport() {
 421                 if( empty( $this->mSource ) ) {
 422                         return new WikiErrorMsg( "importnotext" );
 423                 }
 424
 425                 $parser = xml_parser_create_ns( "UTF-8" );
 426
 427                 # case folding violates XML standard, turn it off
 428                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 429
 430                 xml_set_object( $parser, $this );
 431                 xml_set_element_handler( $parser, "in_start", "" );
 432                 xml_set_start_namespace_decl_handler( $parser, "handleXmlNamespace" );
 433
 434                 $offset = 0; // for context extraction on error reporting
 435                 do {
 436                         $chunk = $this->mSource->readChunk();
 437                         if( !xml_parse( $parser, $chunk, $this->mSource->atEnd() ) ) {
 438                                 wfDebug( "WikiImporter::doImport encountered XML parsing error\n" );
 439                                 return new WikiXmlError( $parser, wfMsgHtml( 'import-parse-failure' ), $chunk, $offset );
 440                         }
 441                         $offset += strlen( $chunk );
 442                 } while( $chunk !== false && !$this->mSource->atEnd() );
 443                 xml_parser_free( $parser );
 444
 445                 return true;
 446         }
 447
 448         function debug( $data ) {
 449                 if( $this->mDebug ) {
 450                         wfDebug( "IMPORT: $data\n" );
 451                 }
 452         }
 453
 454         function notice( $data ) {
 455                 global $wgCommandLineMode;
 456                 if( $wgCommandLineMode ) {
 457                         print "$data\n";
 458                 } else {
 459                         global $wgOut;
 460                         $wgOut->addHTML( "<li>" . htmlspecialchars( $data ) . "</li>\n" );
 461                 }
 462         }
 463
 464         /**
 465          * Set debug mode...
 466          */
 467         function setDebug( $debug ) {
 468                 $this->mDebug = $debug;
 469         }
 470
 471         /**
 472          * Sets the action to perform as each new page in the stream is reached.
 473          * @param $callback callback
 474          * @return callback
 475          */
 476         function setPageCallback( $callback ) {
 477                 $previous = $this->mPageCallback;
 478                 $this->mPageCallback = $callback;
 479                 return $previous;
 480         }
 481
 482         /**
 483          * Sets the action to perform as each page in the stream is completed.
 484          * Callback accepts the page title (as a Title object), a second object
 485          * with the original title form (in case it's been overridden into a
 486          * local namespace), and a count of revisions.
 487          *
 488          * @param $callback callback
 489          * @return callback
 490          */
 491         function setPageOutCallback( $callback ) {
 492                 $previous = $this->mPageOutCallback;
 493                 $this->mPageOutCallback = $callback;
 494                 return $previous;
 495         }
 496
 497         /**
 498          * Sets the action to perform as each page revision is reached.
 499          * @param $callback callback
 500          * @return callback
 501          */
 502         function setRevisionCallback( $callback ) {
 503                 $previous = $this->mRevisionCallback;
 504                 $this->mRevisionCallback = $callback;
 505                 return $previous;
 506         }
 507
 508         /**
 509          * Sets the action to perform as each file upload version is reached.
 510          * @param $callback callback
 511          * @return callback
 512          */
 513         function setUploadCallback( $callback ) {
 514                 $previous = $this->mUploadCallback;
 515                 $this->mUploadCallback = $callback;
 516                 return $previous;
 517         }
 518
 519         /**
 520          * Sets the action to perform as each log item reached.
 521          * @param $callback callback
 522          * @return callback
 523          */
 524         function setLogItemCallback( $callback ) {
 525                 $previous = $this->mLogItemCallback;
 526                 $this->mLogItemCallback = $callback;
 527                 return $previous;
 528         }
 529
 530         /**
 531          * Set a target namespace to override the defaults
 532          */
 533         function setTargetNamespace( $namespace ) {
 534                 if( is_null( $namespace ) ) {
 535                         // Don't override namespaces
 536                         $this->mTargetNamespace = null;
 537                 } elseif( $namespace >= 0 ) {
 538                         // FIXME: Check for validity
 539                         $this->mTargetNamespace = intval( $namespace );
 540                 } else {
 541                         return false;
 542                 }
 543         }
 544
 545         /**
 546          * Default per-revision callback, performs the import.
 547          * @param $revision WikiRevision
 548          * @private
 549          */
 550         function importRevision( $revision ) {
 551                 $dbw = wfGetDB( DB_MASTER );
 552                 return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
 553         }
 554
 555         /**
 556          * Default per-revision callback, performs the import.
 557          * @param $revision WikiRevision
 558          * @private
 559          */
 560         function importLogItem( $rev ) {
 561                 $dbw = wfGetDB( DB_MASTER );
 562                 return $dbw->deadlockLoop( array( $rev, 'importLogItem' ) );
 563         }
 564
 565         /**
 566          * Dummy for now...
 567          */
 568         function importUpload( $revision ) {
 569                 //$dbw = wfGetDB( DB_MASTER );
 570                 //return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
 571                 return false;
 572         }
 573
 574         /**
 575          * Alternate per-revision callback, for debugging.
 576          * @param $revision WikiRevision
 577          * @private
 578          */
 579         function debugRevisionHandler( &$revision ) {
 580                 $this->debug( "Got revision:" );
 581                 if( is_object( $revision->title ) ) {
 582                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 583                 } else {
 584                         $this->debug( "-- Title: <invalid>" );
 585                 }
 586                 $this->debug( "-- User: " . $revision->user_text );
 587                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 588                 $this->debug( "-- Comment: " . $revision->comment );
 589                 $this->debug( "-- Text: " . $revision->text );
 590         }
 591
 592         /**
 593          * Notify the callback function when a new <page> is reached.
 594          * @param $title Title
 595          * @private
 596          */
 597         function pageCallback( $title ) {
 598                 if( is_callable( $this->mPageCallback ) ) {
 599                         call_user_func( $this->mPageCallback, $title );
 600                 }
 601         }
 602
 603         /**
 604          * Notify the callback function when a </page> is closed.
 605          * @param $title Title
 606          * @param $origTitle Title
 607          * @param $revisionCount int
 608          * @param $successCount Int: number of revisions for which callback returned true
 609          * @private
 610          */
 611         function pageOutCallback( $title, $origTitle, $revisionCount, $successCount ) {
 612                 if( is_callable( $this->mPageOutCallback ) ) {
 613                         call_user_func( $this->mPageOutCallback, $title, $origTitle,
 614                                 $revisionCount, $successCount );
 615                 }
 616         }
 617
 618         # XML parser callbacks from here out -- beware!
 619         function donothing( $parser, $x, $y="" ) {
 620                 #$this->debug( "donothing" );
 621         }
 622
 623         function in_start( $parser, $name, $attribs ) {
 624                 $name = $this->stripXmlNamespace($name);
 625                 $this->debug( "in_start $name" );
 626                 if( $name != "mediawiki" ) {
 627                         return $this->throwXMLerror( "Expected <mediawiki>, got <$name>" );
 628                 }
 629                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 630         }
 631
 632         function in_mediawiki( $parser, $name, $attribs ) {
 633                 $name = $this->stripXmlNamespace($name);
 634                 $this->debug( "in_mediawiki $name" );
 635                 if( $name == 'siteinfo' ) {
 636                         xml_set_element_handler( $parser, "in_siteinfo", "out_siteinfo" );
 637                 } elseif( $name == 'page' ) {
 638                         $this->push( $name );
 639                         $this->workRevisionCount = 0;
 640                         $this->workSuccessCount = 0;
 641                         $this->uploadCount = 0;
 642                         $this->uploadSuccessCount = 0;
 643                         xml_set_element_handler( $parser, "in_page", "out_page" );
 644                 } elseif( $name == 'logitem' ) {
 645                         $this->push( $name );
 646                         $this->workRevision = new WikiRevision;
 647                         xml_set_element_handler( $parser, "in_logitem", "out_logitem" );
 648                 } else {
 649                         return $this->throwXMLerror( "Expected <page>, got <$name>" );
 650                 }
 651         }
 652         function out_mediawiki( $parser, $name ) {
 653                 $name = $this->stripXmlNamespace($name);
 654                 $this->debug( "out_mediawiki $name" );
 655                 if( $name != "mediawiki" ) {
 656                         return $this->throwXMLerror( "Expected </mediawiki>, got </$name>" );
 657                 }
 658                 xml_set_element_handler( $parser, "donothing", "donothing" );
 659         }
 660
 661
 662         function in_siteinfo( $parser, $name, $attribs ) {
 663                 // no-ops for now
 664                 $name = $this->stripXmlNamespace($name);
 665                 $this->debug( "in_siteinfo $name" );
 666                 switch( $name ) {
 667                 case "sitename":
 668                 case "base":
 669                 case "generator":
 670                 case "case":
 671                 case "namespaces":
 672                 case "namespace":
 673                         break;
 674                 default:
 675                         return $this->throwXMLerror( "Element <$name> not allowed in <siteinfo>." );
 676                 }
 677         }
 678
 679         function out_siteinfo( $parser, $name ) {
 680                 $name = $this->stripXmlNamespace($name);
 681                 if( $name == "siteinfo" ) {
 682                         xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 683                 }
 684         }
 685
 686
 687         function in_page( $parser, $name, $attribs ) {
 688                 $name = $this->stripXmlNamespace($name);
 689                 $this->debug( "in_page $name" );
 690                 switch( $name ) {
 691                 case "id":
 692                 case "title":
 693                 case "restrictions":
 694                         $this->appendfield = $name;
 695                         $this->appenddata = "";
 696                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 697                         xml_set_character_data_handler( $parser, "char_append" );
 698                         break;
 699                 case "revision":
 700                         $this->push( "revision" );
 701                         if( is_object( $this->pageTitle ) ) {
 702                                 $this->workRevision = new WikiRevision;
 703                                 $this->workRevision->setTitle( $this->pageTitle );
 704                                 $this->workRevisionCount++;
 705                         } else {
 706                                 // Skipping items due to invalid page title
 707                                 $this->workRevision = null;
 708                         }
 709                         xml_set_element_handler( $parser, "in_revision", "out_revision" );
 710                         break;
 711                 case "upload":
 712                         $this->push( "upload" );
 713                         if( is_object( $this->pageTitle ) ) {
 714                                 $this->workRevision = new WikiRevision;
 715                                 $this->workRevision->setTitle( $this->pageTitle );
 716                                 $this->uploadCount++;
 717                         } else {
 718                                 // Skipping items due to invalid page title
 719                                 $this->workRevision = null;
 720                         }
 721                         xml_set_element_handler( $parser, "in_upload", "out_upload" );
 722                         break;
 723                 default:
 724                         return $this->throwXMLerror( "Element <$name> not allowed in a <page>." );
 725                 }
 726         }
 727
 728         function out_page( $parser, $name ) {
 729                 $name = $this->stripXmlNamespace($name);
 730                 $this->debug( "out_page $name" );
 731                 $this->pop();
 732                 if( $name != "page" ) {
 733                         return $this->throwXMLerror( "Expected </page>, got </$name>" );
 734                 }
 735                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 736
 737                 $this->pageOutCallback( $this->pageTitle, $this->origTitle,
 738                         $this->workRevisionCount, $this->workSuccessCount );
 739
 740                 $this->workTitle = null;
 741                 $this->workRevision = null;
 742                 $this->workRevisionCount = 0;
 743                 $this->workSuccessCount = 0;
 744                 $this->pageTitle = null;
 745                 $this->origTitle = null;
 746         }
 747
 748         function in_nothing( $parser, $name, $attribs ) {
 749                 $name = $this->stripXmlNamespace($name);
 750                 $this->debug( "in_nothing $name" );
 751                 return $this->throwXMLerror( "No child elements allowed here; got <$name>" );
 752         }
 753
 754         function char_append( $parser, $data ) {
 755                 $this->debug( "char_append '$data'" );
 756                 $this->appenddata .= $data;
 757         }
 758
 759         function out_append( $parser, $name ) {
 760                 $name = $this->stripXmlNamespace($name);
 761                 $this->debug( "out_append $name" );
 762                 if( $name != $this->appendfield ) {
 763                         return $this->throwXMLerror( "Expected </{$this->appendfield}>, got </$name>" );
 764                 }
 765
 766                 switch( $this->appendfield ) {
 767                 case "title":
 768                         $this->workTitle = $this->appenddata;
 769                         $this->origTitle = Title::newFromText( $this->workTitle );
 770                         if( !is_null( $this->mTargetNamespace ) && !is_null( $this->origTitle ) ) {
 771                                 $this->pageTitle = Title::makeTitle( $this->mTargetNamespace,
 772                                         $this->origTitle->getDBkey() );
 773                         } else {
 774                                 $this->pageTitle = Title::newFromText( $this->workTitle );
 775                         }
 776                         if( is_null( $this->pageTitle ) ) {
 777                                 // Invalid page title? Ignore the page
 778                                 $this->notice( "Skipping invalid page title '$this->workTitle'" );
 779                         } elseif( $this->pageTitle->getInterwiki() != '' ) {
 780                                 $this->notice( "Skipping interwiki page title '$this->workTitle'" );
 781                                 $this->pageTitle = null;
 782                         } else {
 783                                 $this->pageCallback( $this->workTitle );
 784                         }
 785                         break;
 786                 case "id":
 787                         if ( $this->parentTag() == 'revision' || $this->parentTag() == 'logitem' ) {
 788                                 if( $this->workRevision )
 789                                         $this->workRevision->setID( $this->appenddata );
 790                         }
 791                         break;
 792                 case "text":
 793                         if( $this->workRevision )
 794                                 $this->workRevision->setText( $this->appenddata );
 795                         break;
 796                 case "username":
 797                         if( $this->workRevision )
 798                                 $this->workRevision->setUsername( $this->appenddata );
 799                         break;
 800                 case "ip":
 801                         if( $this->workRevision )
 802                                 $this->workRevision->setUserIP( $this->appenddata );
 803                         break;
 804                 case "timestamp":
 805                         if( $this->workRevision )
 806                                 $this->workRevision->setTimestamp( $this->appenddata );
 807                         break;
 808                 case "comment":
 809                         if( $this->workRevision )
 810                                 $this->workRevision->setComment( $this->appenddata );
 811                         break;
 812                 case "type":
 813                         if( $this->workRevision )
 814                                 $this->workRevision->setType( $this->appenddata );
 815                         break;
 816                 case "action":
 817                         if( $this->workRevision )
 818                                 $this->workRevision->setAction( $this->appenddata );
 819                         break;
 820                 case "logtitle":
 821                         if( $this->workRevision )
 822                                 $this->workRevision->setTitle( Title::newFromText( $this->appenddata ) );
 823                         break;
 824                 case "params":
 825                         if( $this->workRevision )
 826                                 $this->workRevision->setParams( $this->appenddata );
 827                         break;
 828                 case "minor":
 829                         if( $this->workRevision )
 830                                 $this->workRevision->setMinor( true );
 831                         break;
 832                 case "filename":
 833                         if( $this->workRevision )
 834                                 $this->workRevision->setFilename( $this->appenddata );
 835                         break;
 836                 case "src":
 837                         if( $this->workRevision )
 838                                 $this->workRevision->setSrc( $this->appenddata );
 839                         break;
 840                 case "size":
 841                         if( $this->workRevision )
 842                                 $this->workRevision->setSize( intval( $this->appenddata ) );
 843                         break;
 844                 default:
 845                         $this->debug( "Bad append: {$this->appendfield}" );
 846                 }
 847                 $this->appendfield = "";
 848                 $this->appenddata = "";
 849
 850                 $parent = $this->parentTag();
 851                 xml_set_element_handler( $parser, "in_$parent", "out_$parent" );
 852                 xml_set_character_data_handler( $parser, "donothing" );
 853         }
 854
 855         function in_revision( $parser, $name, $attribs ) {
 856                 $name = $this->stripXmlNamespace($name);
 857                 $this->debug( "in_revision $name" );
 858                 switch( $name ) {
 859                 case "id":
 860                 case "timestamp":
 861                 case "comment":
 862                 case "minor":
 863                 case "text":
 864                         $this->appendfield = $name;
 865                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 866                         xml_set_character_data_handler( $parser, "char_append" );
 867                         break;
 868                 case "contributor":
 869                         $this->push( "contributor" );
 870                         xml_set_element_handler( $parser, "in_contributor", "out_contributor" );
 871                         break;
 872                 default:
 873                         return $this->throwXMLerror( "Element <$name> not allowed in a <revision>." );
 874                 }
 875         }
 876
 877         function out_revision( $parser, $name ) {
 878                 $name = $this->stripXmlNamespace($name);
 879                 $this->debug( "out_revision $name" );
 880                 $this->pop();
 881                 if( $name != "revision" ) {
 882                         return $this->throwXMLerror( "Expected </revision>, got </$name>" );
 883                 }
 884                 xml_set_element_handler( $parser, "in_page", "out_page" );
 885
 886                 if( $this->workRevision ) {
 887                         $ok = call_user_func_array( $this->mRevisionCallback,
 888                                 array( $this->workRevision, $this ) );
 889                         if( $ok ) {
 890                                 $this->workSuccessCount++;
 891                         }
 892                 }
 893         }
 894
 895         function in_logitem( $parser, $name, $attribs ) {
 896                 $name = $this->stripXmlNamespace($name);
 897                 $this->debug( "in_logitem $name" );
 898                 switch( $name ) {
 899                 case "id":
 900                 case "timestamp":
 901                 case "comment":
 902                 case "type":
 903                 case "action":
 904                 case "logtitle":
 905                 case "params":
 906                         $this->appendfield = $name;
 907                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 908                         xml_set_character_data_handler( $parser, "char_append" );
 909                         break;
 910                 case "contributor":
 911                         $this->push( "contributor" );
 912                         xml_set_element_handler( $parser, "in_contributor", "out_contributor" );
 913                         break;
 914                 default:
 915                         return $this->throwXMLerror( "Element <$name> not allowed in a <revision>." );
 916                 }
 917         }
 918
 919         function out_logitem( $parser, $name ) {
 920                 $name = $this->stripXmlNamespace($name);
 921                 $this->debug( "out_logitem $name" );
 922                 $this->pop();
 923                 if( $name != "logitem" ) {
 924                         return $this->throwXMLerror( "Expected </logitem>, got </$name>" );
 925                 }
 926                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 927
 928                 if( $this->workRevision ) {
 929                         $ok = call_user_func_array( $this->mLogItemCallback,
 930                                 array( $this->workRevision, $this ) );
 931                         if( $ok ) {
 932                                 $this->workSuccessCount++;
 933                         }
 934                 }
 935         }
 936
 937         function in_upload( $parser, $name, $attribs ) {
 938                 $name = $this->stripXmlNamespace($name);
 939                 $this->debug( "in_upload $name" );
 940                 switch( $name ) {
 941                 case "timestamp":
 942                 case "comment":
 943                 case "text":
 944                 case "filename":
 945                 case "src":
 946                 case "size":
 947                         $this->appendfield = $name;
 948                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 949                         xml_set_character_data_handler( $parser, "char_append" );
 950                         break;
 951                 case "contributor":
 952                         $this->push( "contributor" );
 953                         xml_set_element_handler( $parser, "in_contributor", "out_contributor" );
 954                         break;
 955                 default:
 956                         return $this->throwXMLerror( "Element <$name> not allowed in an <upload>." );
 957                 }
 958         }
 959
 960         function out_upload( $parser, $name ) {
 961                 $name = $this->stripXmlNamespace($name);
 962                 $this->debug( "out_revision $name" );
 963                 $this->pop();
 964                 if( $name != "upload" ) {
 965                         return $this->throwXMLerror( "Expected </upload>, got </$name>" );
 966                 }
 967                 xml_set_element_handler( $parser, "in_page", "out_page" );
 968
 969                 if( $this->workRevision ) {
 970                         $ok = call_user_func_array( $this->mUploadCallback,
 971                                 array( $this->workRevision, $this ) );
 972                         if( $ok ) {
 973                                 $this->workUploadSuccessCount++;
 974                         }
 975                 }
 976         }
 977
 978         function in_contributor( $parser, $name, $attribs ) {
 979                 $name = $this->stripXmlNamespace($name);
 980                 $this->debug( "in_contributor $name" );
 981                 switch( $name ) {
 982                 case "username":
 983                 case "ip":
 984                 case "id":
 985                         $this->appendfield = $name;
 986                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 987                         xml_set_character_data_handler( $parser, "char_append" );
 988                         break;
 989                 default:
 990                         $this->throwXMLerror( "Invalid tag <$name> in <contributor>" );
 991                 }
 992         }
 993
 994         function out_contributor( $parser, $name ) {
 995                 $name = $this->stripXmlNamespace($name);
 996                 $this->debug( "out_contributor $name" );
 997                 $this->pop();
 998                 if( $name != "contributor" ) {
 999                         return $this->throwXMLerror( "Expected </contributor>, got </$name>" );
1000                 }
1001                 $parent = $this->parentTag();
1002                 xml_set_element_handler( $parser, "in_$parent", "out_$parent" );
1003         }
1004
1005         private function push( $name ) {
1006                 array_push( $this->tagStack, $name );
1007                 $this->debug( "PUSH $name" );
1008         }
1009
1010         private function pop() {
1011                 $name = array_pop( $this->tagStack );
1012                 $this->debug( "POP $name" );
1013                 return $name;
1014         }
1015
1016         private function parentTag() {
1017                 $name = $this->tagStack[count( $this->tagStack ) - 1];
1018                 $this->debug( "PARENT $name" );
1019                 return $name;
1020         }
1021
1022 }
1023
1024 /**
1025  * @todo document (e.g. one-sentence class description).
1026  * @ingroup SpecialPage
1027  */
1028 class ImportStringSource {
1029         function __construct( $string ) {
1030                 $this->mString = $string;
1031                 $this->mRead = false;
1032         }
1033
1034         function atEnd() {
1035                 return $this->mRead;
1036         }
1037
1038         function readChunk() {
1039                 if( $this->atEnd() ) {
1040                         return false;
1041                 } else {
1042                         $this->mRead = true;
1043                         return $this->mString;
1044                 }
1045         }
1046 }
1047
1048 /**
1049  * @todo document (e.g. one-sentence class description).
1050  * @ingroup SpecialPage
1051  */
1052 class ImportStreamSource {
1053         function __construct( $handle ) {
1054                 $this->mHandle = $handle;
1055         }
1056
1057         function atEnd() {
1058                 return feof( $this->mHandle );
1059         }
1060
1061         function readChunk() {
1062                 return fread( $this->mHandle, 32768 );
1063         }
1064
1065         static function newFromFile( $filename ) {
1066                 $file = @fopen( $filename, 'rt' );
1067                 if( !$file ) {
1068                         return new WikiErrorMsg( "importcantopen" );
1069                 }
1070                 return new ImportStreamSource( $file );
1071         }
1072
1073         static function newFromUpload( $fieldname = "xmlimport" ) {
1074                 $upload =& $_FILES[$fieldname];
1075
1076                 if( !isset( $upload ) || !$upload['name'] ) {
1077                         return new WikiErrorMsg( 'importnofile' );
1078                 }
1079                 if( !empty( $upload['error'] ) ) {
1080                         switch($upload['error']){
1081                                 case 1: # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1082                                         return new WikiErrorMsg( 'importuploaderrorsize' );
1083                                 case 2: # The uploaded file exceeds the MAX_FILE_SIZE directive that was specified in the HTML form.
1084                                         return new WikiErrorMsg( 'importuploaderrorsize' );
1085                                 case 3: # The uploaded file was only partially uploaded
1086                                         return new WikiErrorMsg( 'importuploaderrorpartial' );
1087                             case 6: #Missing a temporary folder. Introduced in PHP 4.3.10 and PHP 5.0.3.
1088                                 return new WikiErrorMsg( 'importuploaderrortemp' );
1089                             # case else: # Currently impossible
1090                         }
1091
1092                 }
1093                 $fname = $upload['tmp_name'];
1094                 if( is_uploaded_file( $fname ) ) {
1095                         return ImportStreamSource::newFromFile( $fname );
1096                 } else {
1097                         return new WikiErrorMsg( 'importnofile' );
1098                 }
1099         }
1100
1101         static function newFromURL( $url, $method = 'GET' ) {
1102                 wfDebug( __METHOD__ . ": opening $url\n" );
1103                 # Use the standard HTTP fetch function; it times out
1104                 # quicker and sorts out user-agent problems which might
1105                 # otherwise prevent importing from large sites, such
1106                 # as the Wikimedia cluster, etc.
1107                 $data = Http::request( $method, $url );
1108                 if( $data !== false ) {
1109                         $file = tmpfile();
1110                         fwrite( $file, $data );
1111                         fflush( $file );
1112                         fseek( $file, 0 );
1113                         return new ImportStreamSource( $file );
1114                 } else {
1115                         return new WikiErrorMsg( 'importcantopen' );
1116                 }
1117         }
1118
1119         public static function newFromInterwiki( $interwiki, $page, $history=false ) {
1120                 if( $page == '' ) {
1121                         return new WikiErrorMsg( 'import-noarticle' );
1122                 }
1123                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1124                 if( is_null( $link ) || $link->getInterwiki() == '' ) {
1125                         return new WikiErrorMsg( 'importbadinterwiki' );
1126                 } else {
1127                         $params = $history ? 'history=1' : '';
1128                         $url = $link->getFullUrl( $params );
1129                         # For interwikis, use POST to avoid redirects.
1130                         return ImportStreamSource::newFromURL( $url, "POST" );
1131                 }
1132         }
1133 }