includes/SpecialImport.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  * @addtogroup SpecialPage
  23  */
  24
  25 /**
  26  * Constructor
  27  */
  28 function wfSpecialImport( $page = '' ) {
  29         global $wgUser, $wgOut, $wgRequest, $wgTitle, $wgImportSources;
  30         global $wgImportTargetNamespace;
  31
  32         $interwiki = false;
  33         $namespace = $wgImportTargetNamespace;
  34         $frompage = '';
  35         $history = true;
  36
  37         if ( wfReadOnly() ) {
  38                 $wgOut->readOnlyPage();
  39                 return;
  40         }
  41
  42         if( $wgRequest->wasPosted() && $wgRequest->getVal( 'action' ) == 'submit') {
  43                 $isUpload = false;
  44                 $namespace = $wgRequest->getIntOrNull( 'namespace' );
  45
  46                 switch( $wgRequest->getVal( "source" ) ) {
  47                 case "upload":
  48                         $isUpload = true;
  49                         if( $wgUser->isAllowed( 'importupload' ) ) {
  50                                 $source = ImportStreamSource::newFromUpload( "xmlimport" );
  51                         } else {
  52                                 return $wgOut->permissionRequired( 'importupload' );
  53                         }
  54                         break;
  55                 case "interwiki":
  56                         $interwiki = $wgRequest->getVal( 'interwiki' );
  57                         $history = $wgRequest->getCheck( 'interwikiHistory' );
  58                         $frompage = $wgRequest->getText( "frompage" );
  59                         $source = ImportStreamSource::newFromInterwiki(
  60                                 $interwiki,
  61                                 $frompage,
  62                                 $history );
  63                         break;
  64                 default:
  65                         $source = new WikiErrorMsg( "importunknownsource" );
  66                 }
  67
  68                 if( WikiError::isError( $source ) ) {
  69                         $wgOut->addWikiText( '<p class="error">' . wfMsg( 'importfailed', wfEscapeWikiText( $source->getMessage() ) ) . '</p>' );
  70                 } else {
  71                         $wgOut->addWikiText( wfMsg( "importstart" ) );
  72
  73                         $importer = new WikiImporter( $source );
  74                         if( !is_null( $namespace ) ) {
  75                                 $importer->setTargetNamespace( $namespace );
  76                         }
  77                         $reporter = new ImportReporter( $importer, $isUpload, $interwiki );
  78
  79                         $reporter->open();
  80                         $result = $importer->doImport();
  81                         $resultCount = $reporter->close();
  82
  83                         if( WikiError::isError( $result ) ) {
  84                                 # No source or XML parse error
  85                                 $wgOut->addWikiText( '<p class="error">' . wfMsg( 'importfailed', wfEscapeWikiText( $result->getMessage() ) ) . '</p>' );
  86                         } elseif( WikiError::isError( $resultCount ) ) {
  87                                 # Zero revisions
  88                                 $wgOut->addWikiText( '<p class="error">' . wfMsg( 'importfailed', wfEscapeWikiText( $resultCount->getMessage() ) ) . '</p>' );
  89                         } else {
  90                                 # Success!
  91                                 $wgOut->addWikiText( wfMsg( 'importsuccess' ) );
  92                         }
  93                         $wgOut->addWikiText( '<hr />' );
  94                 }
  95         }
  96
  97         $action = $wgTitle->getLocalUrl( 'action=submit' );
  98
  99         if( $wgUser->isAllowed( 'importupload' ) ) {
 100                 $wgOut->addWikiText( wfMsg( "importtext" ) );
 101                 $wgOut->addHTML(
 102                         Xml::openElement( 'fieldset' ).
 103                         Xml::element( 'legend', null, wfMsg( 'upload' ) ) .
 104                         Xml::openElement( 'form', array( 'enctype' => 'multipart/form-data', 'method' => 'post', 'action' => $action ) ) .
 105                         Xml::hidden( 'action', 'submit' ) .
 106                         Xml::hidden( 'source', 'upload' ) .
 107                         "<input type='file' name='xmlimport' value='' size='30' />" . // No Xml function for type=file? Todo?
 108                         Xml::submitButton( wfMsg( 'uploadbtn' ) ) .
 109                         Xml::closeElement( 'form' ) .
 110                         Xml::closeElement( 'fieldset' )
 111                 );
 112         } else {
 113                 if( empty( $wgImportSources ) ) {
 114                         $wgOut->addWikiText( wfMsg( 'importnosources' ) );
 115                 }
 116         }
 117
 118         if( !empty( $wgImportSources ) ) {
 119                 $wgOut->addHTML(
 120                         Xml::openElement( 'fieldset' ) .
 121                         Xml::element( 'legend', null, wfMsg( 'importinterwiki' ) ) .
 122                         Xml::openElement( 'form', array( 'method' => 'post', 'action' => $action ) ) .
 123                         wfMsgExt( 'import-interwiki-text', array( 'parse' ) ) .
 124                         Xml::hidden( 'action', 'submit' ) .
 125                         Xml::hidden( 'source', 'interwiki' ) .
 126                         Xml::openElement( 'table' ) .
 127                         "<tr>
 128                                 <td>" .
 129                                         Xml::openElement( 'select', array( 'name' => 'interwiki' ) )
 130                 );
 131                 foreach( $wgImportSources as $prefix ) {
 132                         $iw = htmlspecialchars( $prefix );
 133                         $selected = ($interwiki === $prefix) ? ' selected="selected"' : '';
 134                         $wgOut->addHTML( Xml::option( $iw, $iw, $selected ) );
 135                 }
 136                 $wgOut->addHTML(
 137                                         Xml::closeElement( 'select' ) .
 138                                 "</td>
 139                                 <td>" .
 140                                         Xml::input( 'frompage', 50, $frompage ) .
 141                                 "</td>
 142                         </tr>
 143                         <tr>
 144                                 <td>
 145                                 </td>
 146                                 <td>" .
 147                                         Xml::checkLabel( wfMsg( 'import-interwiki-history' ), 'interwikiHistory', 'interwikiHistory', $history ) .
 148                                 "</td>
 149                         </tr>
 150                         <tr>
 151                                 <td>
 152                                 </td>
 153                                 <td>" .
 154                                         Xml::label( wfMsg( 'import-interwiki-namespace' ), 'namespace' ) .
 155                                         Xml::namespaceSelector( $namespace, '' ) .
 156                                 "</td>
 157                         </tr>
 158                         <tr>
 159                                 <td>
 160                                 </td>
 161                                 <td>" .
 162                                         Xml::submitButton( wfMsg( 'import-interwiki-submit' ) ) .
 163                                 "</td>
 164                         </tr>" .
 165                         Xml::closeElement( 'table' ).
 166                         Xml::closeElement( 'form' ) .
 167                         Xml::closeElement( 'fieldset' )
 168                 );
 169         }
 170 }
 171
 172 /**
 173  * Reporting callback
 174  * @addtogroup SpecialPage
 175  */
 176 class ImportReporter {
 177         function __construct( $importer, $upload, $interwiki ) {
 178                 $importer->setPageOutCallback( array( $this, 'reportPage' ) );
 179                 $this->mPageCount = 0;
 180                 $this->mIsUpload = $upload;
 181                 $this->mInterwiki = $interwiki;
 182         }
 183
 184         function open() {
 185                 global $wgOut;
 186                 $wgOut->addHtml( "<ul>\n" );
 187         }
 188
 189         function reportPage( $title, $origTitle, $revisionCount, $successCount ) {
 190                 global $wgOut, $wgUser, $wgLang, $wgContLang;
 191
 192                 $skin = $wgUser->getSkin();
 193
 194                 $this->mPageCount++;
 195
 196                 $localCount = $wgLang->formatNum( $successCount );
 197                 $contentCount = $wgContLang->formatNum( $successCount );
 198
 199                 if( $successCount > 0 ) {
 200                         $wgOut->addHtml( "<li>" . $skin->makeKnownLinkObj( $title ) . " " .
 201                                 wfMsgExt( 'import-revision-count', array( 'parsemag', 'escape' ), $localCount ) .
 202                                 "</li>\n"
 203                         );
 204
 205                         $log = new LogPage( 'import' );
 206                         if( $this->mIsUpload ) {
 207                                 $detail = wfMsgExt( 'import-logentry-upload-detail', array( 'content', 'parsemag' ),
 208                                         $contentCount );
 209                                 $log->addEntry( 'upload', $title, $detail );
 210                         } else {
 211                                 $interwiki = '[[:' . $this->mInterwiki . ':' .
 212                                         $origTitle->getPrefixedText() . ']]';
 213                                 $detail = wfMsgExt( 'import-logentry-interwiki-detail', array( 'content', 'parsemag' ),
 214                                         $contentCount, $interwiki );
 215                                 $log->addEntry( 'interwiki', $title, $detail );
 216                         }
 217
 218                         $comment = $detail; // quick
 219                         $dbw = wfGetDB( DB_MASTER );
 220                         $nullRevision = Revision::newNullRevision(
 221                                 $dbw, $title->getArticleId(), $comment, true );
 222                         $nullRevision->insertOn( $dbw );
 223                         # Update page record
 224                         $article = new Article( $title );
 225                         $article->updateRevisionOn( $dbw, $nullRevision );
 226                 } else {
 227                         $wgOut->addHtml( '<li>' . wfMsgHtml( 'import-nonewrevisions' ) . '</li>' );
 228                 }
 229         }
 230
 231         function close() {
 232                 global $wgOut;
 233                 if( $this->mPageCount == 0 ) {
 234                         $wgOut->addHtml( "</ul>\n" );
 235                         return new WikiErrorMsg( "importnopages" );
 236                 }
 237                 $wgOut->addHtml( "</ul>\n" );
 238
 239                 return $this->mPageCount;
 240         }
 241 }
 242
 243 /**
 244  *
 245  * @addtogroup SpecialPage
 246  */
 247 class WikiRevision {
 248         var $title = null;
 249         var $id = 0;
 250         var $timestamp = "20010115000000";
 251         var $user = 0;
 252         var $user_text = "";
 253         var $text = "";
 254         var $comment = "";
 255         var $minor = false;
 256
 257         function setTitle( $title ) {
 258                 if( is_object( $title ) ) {
 259                         $this->title = $title;
 260                 } elseif( is_null( $title ) ) {
 261                         throw new MWException( "WikiRevision given a null title in import. You may need to adjust \$wgLegalTitleChars." );
 262                 } else {
 263                         throw new MWException( "WikiRevision given non-object title in import." );
 264                 }
 265         }
 266
 267         function setID( $id ) {
 268                 $this->id = $id;
 269         }
 270
 271         function setTimestamp( $ts ) {
 272                 # 2003-08-05T18:30:02Z
 273                 $this->timestamp = wfTimestamp( TS_MW, $ts );
 274         }
 275
 276         function setUsername( $user ) {
 277                 $this->user_text = $user;
 278         }
 279
 280         function setUserIP( $ip ) {
 281                 $this->user_text = $ip;
 282         }
 283
 284         function setText( $text ) {
 285                 $this->text = $text;
 286         }
 287
 288         function setComment( $text ) {
 289                 $this->comment = $text;
 290         }
 291
 292         function setMinor( $minor ) {
 293                 $this->minor = (bool)$minor;
 294         }
 295
 296         function getTitle() {
 297                 return $this->title;
 298         }
 299
 300         function getID() {
 301                 return $this->id;
 302         }
 303
 304         function getTimestamp() {
 305                 return $this->timestamp;
 306         }
 307
 308         function getUser() {
 309                 return $this->user_text;
 310         }
 311
 312         function getText() {
 313                 return $this->text;
 314         }
 315
 316         function getComment() {
 317                 return $this->comment;
 318         }
 319
 320         function getMinor() {
 321                 return $this->minor;
 322         }
 323
 324         function importOldRevision() {
 325                 $dbw = wfGetDB( DB_MASTER );
 326
 327                 # Sneak a single revision into place
 328                 $user = User::newFromName( $this->getUser() );
 329                 if( $user ) {
 330                         $userId = intval( $user->getId() );
 331                         $userText = $user->getName();
 332                 } else {
 333                         $userId = 0;
 334                         $userText = $this->getUser();
 335                 }
 336
 337                 // avoid memory leak...?
 338                 $linkCache =& LinkCache::singleton();
 339                 $linkCache->clear();
 340
 341                 $article = new Article( $this->title );
 342                 $pageId = $article->getId();
 343                 if( $pageId == 0 ) {
 344                         # must create the page...
 345                         $pageId = $article->insertOn( $dbw );
 346                         $created = true;
 347                 } else {
 348                         $created = false;
 349
 350                         $prior = Revision::loadFromTimestamp( $dbw, $this->title, $this->timestamp );
 351                         if( !is_null( $prior ) ) {
 352                                 // FIXME: this could fail slightly for multiple matches :P
 353                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
 354                                         $this->title->getPrefixedText() . "]], timestamp " .
 355                                         $this->timestamp . "\n" );
 356                                 return false;
 357                         }
 358                 }
 359
 360                 # FIXME: Use original rev_id optionally
 361                 # FIXME: blah blah blah
 362
 363                 #if( $numrows > 0 ) {
 364                 #       return wfMsg( "importhistoryconflict" );
 365                 #}
 366
 367                 # Insert the row
 368                 $revision = new Revision( array(
 369                         'page'       => $pageId,
 370                         'text'       => $this->getText(),
 371                         'comment'    => $this->getComment(),
 372                         'user'       => $userId,
 373                         'user_text'  => $userText,
 374                         'timestamp'  => $this->timestamp,
 375                         'minor_edit' => $this->minor,
 376                         ) );
 377                 $revId = $revision->insertOn( $dbw );
 378                 $changed = $article->updateIfNewerOn( $dbw, $revision );
 379
 380                 if( $created ) {
 381                         wfDebug( __METHOD__ . ": running onArticleCreate\n" );
 382                         Article::onArticleCreate( $this->title );
 383
 384                         wfDebug( __METHOD__ . ": running create updates\n" );
 385                         $article->createUpdates( $revision );
 386
 387                 } elseif( $changed ) {
 388                         wfDebug( __METHOD__ . ": running onArticleEdit\n" );
 389                         Article::onArticleEdit( $this->title );
 390
 391                         wfDebug( __METHOD__ . ": running edit updates\n" );
 392                         $article->editUpdates(
 393                                 $this->getText(),
 394                                 $this->getComment(),
 395                                 $this->minor,
 396                                 $this->timestamp,
 397                                 $revId );
 398                 }
 399
 400                 return true;
 401         }
 402
 403 }
 404
 405 /**
 406  * implements Special:Import
 407  * @addtogroup SpecialPage
 408  */
 409 class WikiImporter {
 410         var $mSource = null;
 411         var $mPageCallback = null;
 412         var $mPageOutCallback = null;
 413         var $mRevisionCallback = null;
 414         var $mTargetNamespace = null;
 415         var $lastfield;
 416
 417         function WikiImporter( $source ) {
 418                 $this->setRevisionCallback( array( &$this, "importRevision" ) );
 419                 $this->mSource = $source;
 420         }
 421
 422         function throwXmlError( $err ) {
 423                 $this->debug( "FAILURE: $err" );
 424                 wfDebug( "WikiImporter XML error: $err\n" );
 425         }
 426
 427         # --------------
 428
 429         function doImport() {
 430                 if( empty( $this->mSource ) ) {
 431                         return new WikiErrorMsg( "importnotext" );
 432                 }
 433
 434                 $parser = xml_parser_create( "UTF-8" );
 435
 436                 # case folding violates XML standard, turn it off
 437                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 438
 439                 xml_set_object( $parser, $this );
 440                 xml_set_element_handler( $parser, "in_start", "" );
 441
 442                 $offset = 0; // for context extraction on error reporting
 443                 do {
 444                         $chunk = $this->mSource->readChunk();
 445                         if( !xml_parse( $parser, $chunk, $this->mSource->atEnd() ) ) {
 446                                 wfDebug( "WikiImporter::doImport encountered XML parsing error\n" );
 447                                 return new WikiXmlError( $parser, wfMsgHtml( 'import-parse-failure' ), $chunk, $offset );
 448                         }
 449                         $offset += strlen( $chunk );
 450                 } while( $chunk !== false && !$this->mSource->atEnd() );
 451                 xml_parser_free( $parser );
 452
 453                 return true;
 454         }
 455
 456         function debug( $data ) {
 457                 #wfDebug( "IMPORT: $data\n" );
 458         }
 459
 460         function notice( $data ) {
 461                 global $wgCommandLineMode;
 462                 if( $wgCommandLineMode ) {
 463                         print "$data\n";
 464                 } else {
 465                         global $wgOut;
 466                         $wgOut->addHTML( "<li>" . htmlspecialchars( $data ) . "</li>\n" );
 467                 }
 468         }
 469
 470         /**
 471          * Sets the action to perform as each new page in the stream is reached.
 472          * @param callable $callback
 473          * @return callable
 474          */
 475         function setPageCallback( $callback ) {
 476                 $previous = $this->mPageCallback;
 477                 $this->mPageCallback = $callback;
 478                 return $previous;
 479         }
 480
 481         /**
 482          * Sets the action to perform as each page in the stream is completed.
 483          * Callback accepts the page title (as a Title object), a second object
 484          * with the original title form (in case it's been overridden into a
 485          * local namespace), and a count of revisions.
 486          *
 487          * @param callable $callback
 488          * @return callable
 489          */
 490         function setPageOutCallback( $callback ) {
 491                 $previous = $this->mPageOutCallback;
 492                 $this->mPageOutCallback = $callback;
 493                 return $previous;
 494         }
 495
 496         /**
 497          * Sets the action to perform as each page revision is reached.
 498          * @param callable $callback
 499          * @return callable
 500          */
 501         function setRevisionCallback( $callback ) {
 502                 $previous = $this->mRevisionCallback;
 503                 $this->mRevisionCallback = $callback;
 504                 return $previous;
 505         }
 506
 507         /**
 508          * Set a target namespace to override the defaults
 509          */
 510         function setTargetNamespace( $namespace ) {
 511                 if( is_null( $namespace ) ) {
 512                         // Don't override namespaces
 513                         $this->mTargetNamespace = null;
 514                 } elseif( $namespace >= 0 ) {
 515                         // FIXME: Check for validity
 516                         $this->mTargetNamespace = intval( $namespace );
 517                 } else {
 518                         return false;
 519                 }
 520         }
 521
 522         /**
 523          * Default per-revision callback, performs the import.
 524          * @param WikiRevision $revision
 525          * @private
 526          */
 527         function importRevision( &$revision ) {
 528                 $dbw = wfGetDB( DB_MASTER );
 529                 return $dbw->deadlockLoop( array( &$revision, 'importOldRevision' ) );
 530         }
 531
 532         /**
 533          * Alternate per-revision callback, for debugging.
 534          * @param WikiRevision $revision
 535          * @private
 536          */
 537         function debugRevisionHandler( &$revision ) {
 538                 $this->debug( "Got revision:" );
 539                 if( is_object( $revision->title ) ) {
 540                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 541                 } else {
 542                         $this->debug( "-- Title: <invalid>" );
 543                 }
 544                 $this->debug( "-- User: " . $revision->user_text );
 545                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 546                 $this->debug( "-- Comment: " . $revision->comment );
 547                 $this->debug( "-- Text: " . $revision->text );
 548         }
 549
 550         /**
 551          * Notify the callback function when a new <page> is reached.
 552          * @param Title $title
 553          * @private
 554          */
 555         function pageCallback( $title ) {
 556                 if( is_callable( $this->mPageCallback ) ) {
 557                         call_user_func( $this->mPageCallback, $title );
 558                 }
 559         }
 560
 561         /**
 562          * Notify the callback function when a </page> is closed.
 563          * @param Title $title
 564          * @param Title $origTitle
 565          * @param int $revisionCount
 566          * @param int $successCount number of revisions for which callback returned true
 567          * @private
 568          */
 569         function pageOutCallback( $title, $origTitle, $revisionCount, $successCount ) {
 570                 if( is_callable( $this->mPageOutCallback ) ) {
 571                         call_user_func( $this->mPageOutCallback, $title, $origTitle,
 572                                 $revisionCount, $successCount );
 573                 }
 574         }
 575
 576
 577         # XML parser callbacks from here out -- beware!
 578         function donothing( $parser, $x, $y="" ) {
 579                 #$this->debug( "donothing" );
 580         }
 581
 582         function in_start( $parser, $name, $attribs ) {
 583                 $this->debug( "in_start $name" );
 584                 if( $name != "mediawiki" ) {
 585                         return $this->throwXMLerror( "Expected <mediawiki>, got <$name>" );
 586                 }
 587                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 588         }
 589
 590         function in_mediawiki( $parser, $name, $attribs ) {
 591                 $this->debug( "in_mediawiki $name" );
 592                 if( $name == 'siteinfo' ) {
 593                         xml_set_element_handler( $parser, "in_siteinfo", "out_siteinfo" );
 594                 } elseif( $name == 'page' ) {
 595                         $this->workRevisionCount = 0;
 596                         $this->workSuccessCount = 0;
 597                         xml_set_element_handler( $parser, "in_page", "out_page" );
 598                 } else {
 599                         return $this->throwXMLerror( "Expected <page>, got <$name>" );
 600                 }
 601         }
 602         function out_mediawiki( $parser, $name ) {
 603                 $this->debug( "out_mediawiki $name" );
 604                 if( $name != "mediawiki" ) {
 605                         return $this->throwXMLerror( "Expected </mediawiki>, got </$name>" );
 606                 }
 607                 xml_set_element_handler( $parser, "donothing", "donothing" );
 608         }
 609
 610
 611         function in_siteinfo( $parser, $name, $attribs ) {
 612                 // no-ops for now
 613                 $this->debug( "in_siteinfo $name" );
 614                 switch( $name ) {
 615                 case "sitename":
 616                 case "base":
 617                 case "generator":
 618                 case "case":
 619                 case "namespaces":
 620                 case "namespace":
 621                         break;
 622                 default:
 623                         return $this->throwXMLerror( "Element <$name> not allowed in <siteinfo>." );
 624                 }
 625         }
 626
 627         function out_siteinfo( $parser, $name ) {
 628                 if( $name == "siteinfo" ) {
 629                         xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 630                 }
 631         }
 632
 633
 634         function in_page( $parser, $name, $attribs ) {
 635                 $this->debug( "in_page $name" );
 636                 switch( $name ) {
 637                 case "id":
 638                 case "title":
 639                 case "restrictions":
 640                         $this->appendfield = $name;
 641                         $this->appenddata = "";
 642                         $this->parenttag = "page";
 643                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 644                         xml_set_character_data_handler( $parser, "char_append" );
 645                         break;
 646                 case "revision":
 647                         if( is_object( $this->pageTitle ) ) {
 648                                 $this->workRevision = new WikiRevision;
 649                                 $this->workRevision->setTitle( $this->pageTitle );
 650                                 $this->workRevisionCount++;
 651                         } else {
 652                                 // Skipping items due to invalid page title
 653                                 $this->workRevision = null;
 654                         }
 655                         xml_set_element_handler( $parser, "in_revision", "out_revision" );
 656                         break;
 657                 default:
 658                         return $this->throwXMLerror( "Element <$name> not allowed in a <page>." );
 659                 }
 660         }
 661
 662         function out_page( $parser, $name ) {
 663                 $this->debug( "out_page $name" );
 664                 if( $name != "page" ) {
 665                         return $this->throwXMLerror( "Expected </page>, got </$name>" );
 666                 }
 667                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 668
 669                 $this->pageOutCallback( $this->pageTitle, $this->origTitle,
 670                         $this->workRevisionCount, $this->workSuccessCount );
 671
 672                 $this->workTitle = null;
 673                 $this->workRevision = null;
 674                 $this->workRevisionCount = 0;
 675                 $this->workSuccessCount = 0;
 676                 $this->pageTitle = null;
 677                 $this->origTitle = null;
 678         }
 679
 680         function in_nothing( $parser, $name, $attribs ) {
 681                 $this->debug( "in_nothing $name" );
 682                 return $this->throwXMLerror( "No child elements allowed here; got <$name>" );
 683         }
 684         function char_append( $parser, $data ) {
 685                 $this->debug( "char_append '$data'" );
 686                 $this->appenddata .= $data;
 687         }
 688         function out_append( $parser, $name ) {
 689                 $this->debug( "out_append $name" );
 690                 if( $name != $this->appendfield ) {
 691                         return $this->throwXMLerror( "Expected </{$this->appendfield}>, got </$name>" );
 692                 }
 693                 xml_set_element_handler( $parser, "in_$this->parenttag", "out_$this->parenttag" );
 694                 xml_set_character_data_handler( $parser, "donothing" );
 695
 696                 switch( $this->appendfield ) {
 697                 case "title":
 698                         $this->workTitle = $this->appenddata;
 699                         $this->origTitle = Title::newFromText( $this->workTitle );
 700                         if( !is_null( $this->mTargetNamespace ) && !is_null( $this->origTitle ) ) {
 701                                 $this->pageTitle = Title::makeTitle( $this->mTargetNamespace,
 702                                         $this->origTitle->getDBkey() );
 703                         } else {
 704                                 $this->pageTitle = Title::newFromText( $this->workTitle );
 705                         }
 706                         if( is_null( $this->pageTitle ) ) {
 707                                 // Invalid page title? Ignore the page
 708                                 $this->notice( "Skipping invalid page title '$this->workTitle'" );
 709                         } else {
 710                                 $this->pageCallback( $this->workTitle );
 711                         }
 712                         break;
 713                 case "id":
 714                         if ( $this->parenttag == 'revision' ) {
 715                                 if( $this->workRevision )
 716                                         $this->workRevision->setID( $this->appenddata );
 717                         }
 718                         break;
 719                 case "text":
 720                         if( $this->workRevision )
 721                                 $this->workRevision->setText( $this->appenddata );
 722                         break;
 723                 case "username":
 724                         if( $this->workRevision )
 725                                 $this->workRevision->setUsername( $this->appenddata );
 726                         break;
 727                 case "ip":
 728                         if( $this->workRevision )
 729                                 $this->workRevision->setUserIP( $this->appenddata );
 730                         break;
 731                 case "timestamp":
 732                         if( $this->workRevision )
 733                                 $this->workRevision->setTimestamp( $this->appenddata );
 734                         break;
 735                 case "comment":
 736                         if( $this->workRevision )
 737                                 $this->workRevision->setComment( $this->appenddata );
 738                         break;
 739                 case "minor":
 740                         if( $this->workRevision )
 741                                 $this->workRevision->setMinor( true );
 742                         break;
 743                 default:
 744                         $this->debug( "Bad append: {$this->appendfield}" );
 745                 }
 746                 $this->appendfield = "";
 747                 $this->appenddata = "";
 748         }
 749
 750         function in_revision( $parser, $name, $attribs ) {
 751                 $this->debug( "in_revision $name" );
 752                 switch( $name ) {
 753                 case "id":
 754                 case "timestamp":
 755                 case "comment":
 756                 case "minor":
 757                 case "text":
 758                         $this->parenttag = "revision";
 759                         $this->appendfield = $name;
 760                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 761                         xml_set_character_data_handler( $parser, "char_append" );
 762                         break;
 763                 case "contributor":
 764                         xml_set_element_handler( $parser, "in_contributor", "out_contributor" );
 765                         break;
 766                 default:
 767                         return $this->throwXMLerror( "Element <$name> not allowed in a <revision>." );
 768                 }
 769         }
 770
 771         function out_revision( $parser, $name ) {
 772                 $this->debug( "out_revision $name" );
 773                 if( $name != "revision" ) {
 774                         return $this->throwXMLerror( "Expected </revision>, got </$name>" );
 775                 }
 776                 xml_set_element_handler( $parser, "in_page", "out_page" );
 777
 778                 if( $this->workRevision ) {
 779                         $ok = call_user_func_array( $this->mRevisionCallback,
 780                                 array( &$this->workRevision, &$this ) );
 781                         if( $ok ) {
 782                                 $this->workSuccessCount++;
 783                         }
 784                 }
 785         }
 786
 787         function in_contributor( $parser, $name, $attribs ) {
 788                 $this->debug( "in_contributor $name" );
 789                 switch( $name ) {
 790                 case "username":
 791                 case "ip":
 792                 case "id":
 793                         $this->parenttag = "contributor";
 794                         $this->appendfield = $name;
 795                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 796                         xml_set_character_data_handler( $parser, "char_append" );
 797                         break;
 798                 default:
 799                         $this->throwXMLerror( "Invalid tag <$name> in <contributor>" );
 800                 }
 801         }
 802
 803         function out_contributor( $parser, $name ) {
 804                 $this->debug( "out_contributor $name" );
 805                 if( $name != "contributor" ) {
 806                         return $this->throwXMLerror( "Expected </contributor>, got </$name>" );
 807                 }
 808                 xml_set_element_handler( $parser, "in_revision", "out_revision" );
 809         }
 810
 811 }
 812
 813 /**
 814  * @todo document (e.g. one-sentence class description).
 815  * @addtogroup SpecialPage
 816  */
 817 class ImportStringSource {
 818         function ImportStringSource( $string ) {
 819                 $this->mString = $string;
 820                 $this->mRead = false;
 821         }
 822
 823         function atEnd() {
 824                 return $this->mRead;
 825         }
 826
 827         function readChunk() {
 828                 if( $this->atEnd() ) {
 829                         return false;
 830                 } else {
 831                         $this->mRead = true;
 832                         return $this->mString;
 833                 }
 834         }
 835 }
 836
 837 /**
 838  * @todo document (e.g. one-sentence class description).
 839  * @addtogroup SpecialPage
 840  */
 841 class ImportStreamSource {
 842         function ImportStreamSource( $handle ) {
 843                 $this->mHandle = $handle;
 844         }
 845
 846         function atEnd() {
 847                 return feof( $this->mHandle );
 848         }
 849
 850         function readChunk() {
 851                 return fread( $this->mHandle, 32768 );
 852         }
 853
 854         static function newFromFile( $filename ) {
 855                 $file = @fopen( $filename, 'rt' );
 856                 if( !$file ) {
 857                         return new WikiErrorMsg( "importcantopen" );
 858                 }
 859                 return new ImportStreamSource( $file );
 860         }
 861
 862         static function newFromUpload( $fieldname = "xmlimport" ) {
 863                 $upload =& $_FILES[$fieldname];
 864
 865                 if( !isset( $upload ) || !$upload['name'] ) {
 866                         return new WikiErrorMsg( 'importnofile' );
 867                 }
 868                 if( !empty( $upload['error'] ) ) {
 869                         switch($upload['error']){
 870                                 case 1: # The uploaded file exceeds the upload_max_filesize directive in php.ini.
 871                                         return new WikiErrorMsg( 'importuploaderrorsize' );
 872                                 case 2: # The uploaded file exceeds the MAX_FILE_SIZE directive that was specified in the HTML form.
 873                                         return new WikiErrorMsg( 'importuploaderrorsize' );
 874                                 case 3: # The uploaded file was only partially uploaded
 875                                         return new WikiErrorMsg( 'importuploaderrorpartial' );
 876                             case 6: #Missing a temporary folder. Introduced in PHP 4.3.10 and PHP 5.0.3.
 877                                 return new WikiErrorMsg( 'importuploaderrortemp' );
 878                             # case else: # Currently impossible
 879                         }
 880
 881                 }
 882                 $fname = $upload['tmp_name'];
 883                 if( is_uploaded_file( $fname ) ) {
 884                         return ImportStreamSource::newFromFile( $fname );
 885                 } else {
 886                         return new WikiErrorMsg( 'importnofile' );
 887                 }
 888         }
 889
 890         function newFromURL( $url, $method = 'GET' ) {
 891                 wfDebug( __METHOD__ . ": opening $url\n" );
 892                 # Use the standard HTTP fetch function; it times out
 893                 # quicker and sorts out user-agent problems which might
 894                 # otherwise prevent importing from large sites, such
 895                 # as the Wikimedia cluster, etc.
 896                 $data = Http::request( $method, $url );
 897                 if( $data !== false ) {
 898                         $file = tmpfile();
 899                         fwrite( $file, $data );
 900                         fflush( $file );
 901                         fseek( $file, 0 );
 902                         return new ImportStreamSource( $file );
 903                 } else {
 904                         return new WikiErrorMsg( 'importcantopen' );
 905                 }
 906         }
 907
 908         public static function newFromInterwiki( $interwiki, $page, $history=false ) {
 909                 if( $page == '' ) {
 910                         return new WikiErrorMsg( 'import-noarticle' );
 911                 }
 912                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
 913                 if( is_null( $link ) || $link->getInterwiki() == '' ) {
 914                         return new WikiErrorMsg( 'importbadinterwiki' );
 915                 } else {
 916                         $params = $history ? 'history=1' : '';
 917                         $url = $link->getFullUrl( $params );
 918                         # For interwikis, use POST to avoid redirects.
 919                         return ImportStreamSource::newFromURL( $url, "POST" );
 920                 }
 921         }
 922 }