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( $wgRequest->wasPosted() && $wgRequest->getVal( 'action' ) == 'submit') {
  38                 $isUpload = false;
  39                 $namespace = $wgRequest->getIntOrNull( 'namespace' );
  40
  41                 switch( $wgRequest->getVal( "source" ) ) {
  42                 case "upload":
  43                         $isUpload = true;
  44                         if( $wgUser->isAllowed( 'importupload' ) ) {
  45                                 $source = ImportStreamSource::newFromUpload( "xmlimport" );
  46                         } else {
  47                                 return $wgOut->permissionRequired( 'importupload' );
  48                         }
  49                         break;
  50                 case "interwiki":
  51                         $interwiki = $wgRequest->getVal( 'interwiki' );
  52                         $history = $wgRequest->getCheck( 'interwikiHistory' );
  53                         $frompage = $wgRequest->getText( "frompage" );
  54                         $source = ImportStreamSource::newFromInterwiki(
  55                                 $interwiki,
  56                                 $frompage,
  57                                 $history );
  58                         break;
  59                 default:
  60                         $source = new WikiErrorMsg( "importunknownsource" );
  61                 }
  62
  63                 if( WikiError::isError( $source ) ) {
  64                         $wgOut->addWikiText( wfEscapeWikiText( $source->getMessage() ) );
  65                 } else {
  66                         $wgOut->addWikiText( wfMsg( "importstart" ) );
  67
  68                         $importer = new WikiImporter( $source );
  69                         if( !is_null( $namespace ) ) {
  70                                 $importer->setTargetNamespace( $namespace );
  71                         }
  72                         $reporter = new ImportReporter( $importer, $isUpload, $interwiki );
  73
  74                         $reporter->open();
  75                         $result = $importer->doImport();
  76                         $reporter->close();
  77
  78                         if( WikiError::isError( $result ) ) {
  79                                 $wgOut->addWikiText( wfMsg( "importfailed",
  80                                         wfEscapeWikiText( $result->getMessage() ) ) );
  81                         } else {
  82                                 # Success!
  83                                 $wgOut->addWikiText( wfMsg( "importsuccess" ) );
  84                         }
  85                 }
  86         }
  87
  88         $action = $wgTitle->getLocalUrl( 'action=submit' );
  89
  90         if( $wgUser->isAllowed( 'importupload' ) ) {
  91                 $wgOut->addWikiText( wfMsg( "importtext" ) );
  92                 $wgOut->addHTML( "
  93 <fieldset>
  94         <legend>" . wfMsgHtml('upload') . "</legend>
  95         <form enctype='multipart/form-data' method='post' action=\"$action\">
  96                 <input type='hidden' name='action' value='submit' />
  97                 <input type='hidden' name='source' value='upload' />
  98                 <input type='hidden' name='MAX_FILE_SIZE' value='2000000' />
  99                 <input type='file' name='xmlimport' value='' size='30' />
 100                 <input type='submit' value=\"" . wfMsgHtml( "uploadbtn" ) . "\" />
 101         </form>
 102 </fieldset>
 103 " );
 104         } else {
 105                 if( empty( $wgImportSources ) ) {
 106                         $wgOut->addWikiText( wfMsg( 'importnosources' ) );
 107                 }
 108         }
 109
 110         if( !empty( $wgImportSources ) ) {
 111                 $wgOut->addHTML( "
 112 <fieldset>
 113         <legend>" . wfMsgHtml('importinterwiki') . "</legend>
 114         <form method='post' action=\"$action\">" .
 115                 $wgOut->parse( wfMsg( 'import-interwiki-text' ) ) . "
 116                 <input type='hidden' name='action' value='submit' />
 117                 <input type='hidden' name='source' value='interwiki' />
 118                 <table>
 119                         <tr>
 120                                 <td>
 121                                         <select name='interwiki'>" );
 122                 foreach( $wgImportSources as $prefix ) {
 123                         $iw = htmlspecialchars( $prefix );
 124                         $selected = ($interwiki === $prefix) ? ' selected="selected"' : '';
 125                         $wgOut->addHTML( "<option value=\"$iw\"$selected>$iw</option>\n" );
 126                 }
 127                 $wgOut->addHTML( "
 128                                         </select>
 129                                 </td>
 130                                 <td>" .
 131                                         wfInput( 'frompage', 50, $frompage ) .
 132                                 "</td>
 133                         </tr>
 134                         <tr>
 135                                 <td></td>
 136                                 <td>" .
 137                                         wfCheckLabel( wfMsg( 'import-interwiki-history' ),
 138                                                 'interwikiHistory', 'interwikiHistory', $history ) .
 139                                 "</td>
 140                         </tr>
 141                         <tr>
 142                                 <td></td>
 143                                 <td>
 144                                         " . wfMsgHtml( 'import-interwiki-namespace' ) . " " .
 145                                                 HTMLnamespaceselector( $namespace, '' ) . "
 146                                 </td>
 147                         </tr>
 148                         <tr>
 149                                 <td></td>
 150                                 <td>" .
 151                                         wfSubmitButton( wfMsg( 'import-interwiki-submit' ) ) .
 152                                 "</td>
 153                         </tr>
 154                 </table>
 155         </form>
 156 </fieldset>
 157 " );
 158         }
 159 }
 160
 161 /**
 162  * Reporting callback
 163  * @addtogroup SpecialPage
 164  */
 165 class ImportReporter {
 166         function __construct( $importer, $upload, $interwiki ) {
 167                 $importer->setPageOutCallback( array( $this, 'reportPage' ) );
 168                 $this->mPageCount = 0;
 169                 $this->mIsUpload = $upload;
 170                 $this->mInterwiki = $interwiki;
 171         }
 172
 173         function open() {
 174                 global $wgOut;
 175                 $wgOut->addHtml( "<ul>\n" );
 176         }
 177
 178         function reportPage( $title, $origTitle, $revisionCount, $successCount ) {
 179                 global $wgOut, $wgUser, $wgLang, $wgContLang;
 180
 181                 $skin = $wgUser->getSkin();
 182
 183                 $this->mPageCount++;
 184
 185                 $localCount = $wgLang->formatNum( $successCount );
 186                 $contentCount = $wgContLang->formatNum( $successCount );
 187
 188                 $wgOut->addHtml( "<li>" . $skin->makeKnownLinkObj( $title ) .
 189                         " " .
 190                         wfMsgExt( 'import-revision-count', array( 'parsemag', 'escape' ), $localCount ) .
 191                         "</li>\n" );
 192
 193                 if( $successCount > 0 ) {
 194                         $log = new LogPage( 'import' );
 195                         if( $this->mIsUpload ) {
 196                                 $detail = wfMsgForContent( 'import-logentry-upload-detail',
 197                                         $contentCount );
 198                                 $log->addEntry( 'upload', $title, $detail );
 199                         } else {
 200                                 $interwiki = '[[:' . $this->mInterwiki . ':' .
 201                                         $origTitle->getPrefixedText() . ']]';
 202                                 $detail = wfMsgForContent( 'import-logentry-interwiki-detail',
 203                                         $contentCount, $interwiki );
 204                                 $log->addEntry( 'interwiki', $title, $detail );
 205                         }
 206
 207                         $comment = $detail; // quick
 208                         $dbw = wfGetDB( DB_MASTER );
 209                         $nullRevision = Revision::newNullRevision(
 210                                 $dbw, $title->getArticleId(), $comment, true );
 211                         $nullRevision->insertOn( $dbw );
 212                         # Update page record
 213                         $article = new Article( $title );
 214                         $article->updateRevisionOn( $dbw, $nullRevision );
 215                 }
 216         }
 217
 218         function close() {
 219                 global $wgOut;
 220                 if( $this->mPageCount == 0 ) {
 221                         $wgOut->addHtml( "<li>" . wfMsgHtml( 'importnopages' ) . "</li>\n" );
 222                 }
 223                 $wgOut->addHtml( "</ul>\n" );
 224         }
 225 }
 226
 227 /**
 228  *
 229  * @addtogroup SpecialPage
 230  */
 231 class WikiRevision {
 232         var $title = null;
 233         var $id = 0;
 234         var $timestamp = "20010115000000";
 235         var $user = 0;
 236         var $user_text = "";
 237         var $text = "";
 238         var $comment = "";
 239         var $minor = false;
 240
 241         function setTitle( $title ) {
 242                 if( is_object( $title ) ) {
 243                         $this->title = $title;
 244                 } elseif( is_null( $title ) ) {
 245                         throw new MWException( "WikiRevision given a null title in import. You may need to adjust \$wgLegalTitleChars." );
 246                 } else {
 247                         throw new MWException( "WikiRevision given non-object title in import." );
 248                 }
 249         }
 250
 251         function setID( $id ) {
 252                 $this->id = $id;
 253         }
 254
 255         function setTimestamp( $ts ) {
 256                 # 2003-08-05T18:30:02Z
 257                 $this->timestamp = wfTimestamp( TS_MW, $ts );
 258         }
 259
 260         function setUsername( $user ) {
 261                 $this->user_text = $user;
 262         }
 263
 264         function setUserIP( $ip ) {
 265                 $this->user_text = $ip;
 266         }
 267
 268         function setText( $text ) {
 269                 $this->text = $text;
 270         }
 271
 272         function setComment( $text ) {
 273                 $this->comment = $text;
 274         }
 275
 276         function setMinor( $minor ) {
 277                 $this->minor = (bool)$minor;
 278         }
 279
 280         function getTitle() {
 281                 return $this->title;
 282         }
 283
 284         function getID() {
 285                 return $this->id;
 286         }
 287
 288         function getTimestamp() {
 289                 return $this->timestamp;
 290         }
 291
 292         function getUser() {
 293                 return $this->user_text;
 294         }
 295
 296         function getText() {
 297                 return $this->text;
 298         }
 299
 300         function getComment() {
 301                 return $this->comment;
 302         }
 303
 304         function getMinor() {
 305                 return $this->minor;
 306         }
 307
 308         function importOldRevision() {
 309                 $dbw = wfGetDB( DB_MASTER );
 310
 311                 # Sneak a single revision into place
 312                 $user = User::newFromName( $this->getUser() );
 313                 if( $user ) {
 314                         $userId = intval( $user->getId() );
 315                         $userText = $user->getName();
 316                 } else {
 317                         $userId = 0;
 318                         $userText = $this->getUser();
 319                 }
 320
 321                 // avoid memory leak...?
 322                 $linkCache =& LinkCache::singleton();
 323                 $linkCache->clear();
 324
 325                 $article = new Article( $this->title );
 326                 $pageId = $article->getId();
 327                 if( $pageId == 0 ) {
 328                         # must create the page...
 329                         $pageId = $article->insertOn( $dbw );
 330                         $created = true;
 331                 } else {
 332                         $created = false;
 333
 334                         $prior = Revision::loadFromTimestamp( $dbw, $this->title, $this->timestamp );
 335                         if( !is_null( $prior ) ) {
 336                                 // FIXME: this could fail slightly for multiple matches :P
 337                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
 338                                         $this->title->getPrefixedText() . "]], timestamp " .
 339                                         $this->timestamp . "\n" );
 340                                 return false;
 341                         }
 342                 }
 343
 344                 # FIXME: Use original rev_id optionally
 345                 # FIXME: blah blah blah
 346
 347                 #if( $numrows > 0 ) {
 348                 #       return wfMsg( "importhistoryconflict" );
 349                 #}
 350
 351                 # Insert the row
 352                 $revision = new Revision( array(
 353                         'page'       => $pageId,
 354                         'text'       => $this->getText(),
 355                         'comment'    => $this->getComment(),
 356                         'user'       => $userId,
 357                         'user_text'  => $userText,
 358                         'timestamp'  => $this->timestamp,
 359                         'minor_edit' => $this->minor,
 360                         ) );
 361                 $revId = $revision->insertOn( $dbw );
 362                 $changed = $article->updateIfNewerOn( $dbw, $revision );
 363
 364                 if( $created ) {
 365                         wfDebug( __METHOD__ . ": running onArticleCreate\n" );
 366                         Article::onArticleCreate( $this->title );
 367
 368                         wfDebug( __METHOD__ . ": running create updates\n" );
 369                         $article->createUpdates( $revision );
 370
 371                 } elseif( $changed ) {
 372                         wfDebug( __METHOD__ . ": running onArticleEdit\n" );
 373                         Article::onArticleEdit( $this->title );
 374
 375                         wfDebug( __METHOD__ . ": running edit updates\n" );
 376                         $article->editUpdates(
 377                                 $this->getText(),
 378                                 $this->getComment(),
 379                                 $this->minor,
 380                                 $this->timestamp,
 381                                 $revId );
 382                 }
 383
 384                 return true;
 385         }
 386
 387 }
 388
 389 /**
 390  * implements Special:Import
 391  * @addtogroup SpecialPage
 392  */
 393 class WikiImporter {
 394         var $mSource = null;
 395         var $mPageCallback = null;
 396         var $mPageOutCallback = null;
 397         var $mRevisionCallback = null;
 398         var $mTargetNamespace = null;
 399         var $lastfield;
 400
 401         function WikiImporter( $source ) {
 402                 $this->setRevisionCallback( array( &$this, "importRevision" ) );
 403                 $this->mSource = $source;
 404         }
 405
 406         function throwXmlError( $err ) {
 407                 $this->debug( "FAILURE: $err" );
 408                 wfDebug( "WikiImporter XML error: $err\n" );
 409         }
 410
 411         # --------------
 412
 413         function doImport() {
 414                 if( empty( $this->mSource ) ) {
 415                         return new WikiErrorMsg( "importnotext" );
 416                 }
 417
 418                 $parser = xml_parser_create( "UTF-8" );
 419
 420                 # case folding violates XML standard, turn it off
 421                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 422
 423                 xml_set_object( $parser, $this );
 424                 xml_set_element_handler( $parser, "in_start", "" );
 425
 426                 $offset = 0; // for context extraction on error reporting
 427                 do {
 428                         $chunk = $this->mSource->readChunk();
 429                         if( !xml_parse( $parser, $chunk, $this->mSource->atEnd() ) ) {
 430                                 wfDebug( "WikiImporter::doImport encountered XML parsing error\n" );
 431                                 return new WikiXmlError( $parser, 'XML import parse failure', $chunk, $offset );
 432                         }
 433                         $offset += strlen( $chunk );
 434                 } while( $chunk !== false && !$this->mSource->atEnd() );
 435                 xml_parser_free( $parser );
 436
 437                 return true;
 438         }
 439
 440         function debug( $data ) {
 441                 #wfDebug( "IMPORT: $data\n" );
 442         }
 443
 444         function notice( $data ) {
 445                 global $wgCommandLineMode;
 446                 if( $wgCommandLineMode ) {
 447                         print "$data\n";
 448                 } else {
 449                         global $wgOut;
 450                         $wgOut->addHTML( "<li>" . htmlspecialchars( $data ) . "</li>\n" );
 451                 }
 452         }
 453
 454         /**
 455          * Sets the action to perform as each new page in the stream is reached.
 456          * @param callable $callback
 457          * @return callable
 458          */
 459         function setPageCallback( $callback ) {
 460                 $previous = $this->mPageCallback;
 461                 $this->mPageCallback = $callback;
 462                 return $previous;
 463         }
 464
 465         /**
 466          * Sets the action to perform as each page in the stream is completed.
 467          * Callback accepts the page title (as a Title object), a second object
 468          * with the original title form (in case it's been overridden into a
 469          * local namespace), and a count of revisions.
 470          *
 471          * @param callable $callback
 472          * @return callable
 473          */
 474         function setPageOutCallback( $callback ) {
 475                 $previous = $this->mPageOutCallback;
 476                 $this->mPageOutCallback = $callback;
 477                 return $previous;
 478         }
 479
 480         /**
 481          * Sets the action to perform as each page revision is reached.
 482          * @param callable $callback
 483          * @return callable
 484          */
 485         function setRevisionCallback( $callback ) {
 486                 $previous = $this->mRevisionCallback;
 487                 $this->mRevisionCallback = $callback;
 488                 return $previous;
 489         }
 490
 491         /**
 492          * Set a target namespace to override the defaults
 493          */
 494         function setTargetNamespace( $namespace ) {
 495                 if( is_null( $namespace ) ) {
 496                         // Don't override namespaces
 497                         $this->mTargetNamespace = null;
 498                 } elseif( $namespace >= 0 ) {
 499                         // FIXME: Check for validity
 500                         $this->mTargetNamespace = intval( $namespace );
 501                 } else {
 502                         return false;
 503                 }
 504         }
 505
 506         /**
 507          * Default per-revision callback, performs the import.
 508          * @param WikiRevision $revision
 509          * @private
 510          */
 511         function importRevision( &$revision ) {
 512                 $dbw = wfGetDB( DB_MASTER );
 513                 return $dbw->deadlockLoop( array( &$revision, 'importOldRevision' ) );
 514         }
 515
 516         /**
 517          * Alternate per-revision callback, for debugging.
 518          * @param WikiRevision $revision
 519          * @private
 520          */
 521         function debugRevisionHandler( &$revision ) {
 522                 $this->debug( "Got revision:" );
 523                 if( is_object( $revision->title ) ) {
 524                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 525                 } else {
 526                         $this->debug( "-- Title: <invalid>" );
 527                 }
 528                 $this->debug( "-- User: " . $revision->user_text );
 529                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 530                 $this->debug( "-- Comment: " . $revision->comment );
 531                 $this->debug( "-- Text: " . $revision->text );
 532         }
 533
 534         /**
 535          * Notify the callback function when a new <page> is reached.
 536          * @param Title $title
 537          * @private
 538          */
 539         function pageCallback( $title ) {
 540                 if( is_callable( $this->mPageCallback ) ) {
 541                         call_user_func( $this->mPageCallback, $title );
 542                 }
 543         }
 544
 545         /**
 546          * Notify the callback function when a </page> is closed.
 547          * @param Title $title
 548          * @param Title $origTitle
 549          * @param int $revisionCount
 550          * @param int $successCount number of revisions for which callback returned true
 551          * @private
 552          */
 553         function pageOutCallback( $title, $origTitle, $revisionCount, $successCount ) {
 554                 if( is_callable( $this->mPageOutCallback ) ) {
 555                         call_user_func( $this->mPageOutCallback, $title, $origTitle,
 556                                 $revisionCount, $successCount );
 557                 }
 558         }
 559
 560
 561         # XML parser callbacks from here out -- beware!
 562         function donothing( $parser, $x, $y="" ) {
 563                 #$this->debug( "donothing" );
 564         }
 565
 566         function in_start( $parser, $name, $attribs ) {
 567                 $this->debug( "in_start $name" );
 568                 if( $name != "mediawiki" ) {
 569                         return $this->throwXMLerror( "Expected <mediawiki>, got <$name>" );
 570                 }
 571                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 572         }
 573
 574         function in_mediawiki( $parser, $name, $attribs ) {
 575                 $this->debug( "in_mediawiki $name" );
 576                 if( $name == 'siteinfo' ) {
 577                         xml_set_element_handler( $parser, "in_siteinfo", "out_siteinfo" );
 578                 } elseif( $name == 'page' ) {
 579                         $this->workRevisionCount = 0;
 580                         $this->workSuccessCount = 0;
 581                         xml_set_element_handler( $parser, "in_page", "out_page" );
 582                 } else {
 583                         return $this->throwXMLerror( "Expected <page>, got <$name>" );
 584                 }
 585         }
 586         function out_mediawiki( $parser, $name ) {
 587                 $this->debug( "out_mediawiki $name" );
 588                 if( $name != "mediawiki" ) {
 589                         return $this->throwXMLerror( "Expected </mediawiki>, got </$name>" );
 590                 }
 591                 xml_set_element_handler( $parser, "donothing", "donothing" );
 592         }
 593
 594
 595         function in_siteinfo( $parser, $name, $attribs ) {
 596                 // no-ops for now
 597                 $this->debug( "in_siteinfo $name" );
 598                 switch( $name ) {
 599                 case "sitename":
 600                 case "base":
 601                 case "generator":
 602                 case "case":
 603                 case "namespaces":
 604                 case "namespace":
 605                         break;
 606                 default:
 607                         return $this->throwXMLerror( "Element <$name> not allowed in <siteinfo>." );
 608                 }
 609         }
 610
 611         function out_siteinfo( $parser, $name ) {
 612                 if( $name == "siteinfo" ) {
 613                         xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 614                 }
 615         }
 616
 617
 618         function in_page( $parser, $name, $attribs ) {
 619                 $this->debug( "in_page $name" );
 620                 switch( $name ) {
 621                 case "id":
 622                 case "title":
 623                 case "restrictions":
 624                         $this->appendfield = $name;
 625                         $this->appenddata = "";
 626                         $this->parenttag = "page";
 627                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 628                         xml_set_character_data_handler( $parser, "char_append" );
 629                         break;
 630                 case "revision":
 631                         if( is_object( $this->pageTitle ) ) {
 632                                 $this->workRevision = new WikiRevision;
 633                                 $this->workRevision->setTitle( $this->pageTitle );
 634                                 $this->workRevisionCount++;
 635                         } else {
 636                                 // Skipping items due to invalid page title
 637                                 $this->workRevision = null;
 638                         }
 639                         xml_set_element_handler( $parser, "in_revision", "out_revision" );
 640                         break;
 641                 default:
 642                         return $this->throwXMLerror( "Element <$name> not allowed in a <page>." );
 643                 }
 644         }
 645
 646         function out_page( $parser, $name ) {
 647                 $this->debug( "out_page $name" );
 648                 if( $name != "page" ) {
 649                         return $this->throwXMLerror( "Expected </page>, got </$name>" );
 650                 }
 651                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 652
 653                 $this->pageOutCallback( $this->pageTitle, $this->origTitle,
 654                         $this->workRevisionCount, $this->workSuccessCount );
 655
 656                 $this->workTitle = null;
 657                 $this->workRevision = null;
 658                 $this->workRevisionCount = 0;
 659                 $this->workSuccessCount = 0;
 660                 $this->pageTitle = null;
 661                 $this->origTitle = null;
 662         }
 663
 664         function in_nothing( $parser, $name, $attribs ) {
 665                 $this->debug( "in_nothing $name" );
 666                 return $this->throwXMLerror( "No child elements allowed here; got <$name>" );
 667         }
 668         function char_append( $parser, $data ) {
 669                 $this->debug( "char_append '$data'" );
 670                 $this->appenddata .= $data;
 671         }
 672         function out_append( $parser, $name ) {
 673                 $this->debug( "out_append $name" );
 674                 if( $name != $this->appendfield ) {
 675                         return $this->throwXMLerror( "Expected </{$this->appendfield}>, got </$name>" );
 676                 }
 677                 xml_set_element_handler( $parser, "in_$this->parenttag", "out_$this->parenttag" );
 678                 xml_set_character_data_handler( $parser, "donothing" );
 679
 680                 switch( $this->appendfield ) {
 681                 case "title":
 682                         $this->workTitle = $this->appenddata;
 683                         $this->origTitle = Title::newFromText( $this->workTitle );
 684                         if( !is_null( $this->mTargetNamespace ) && !is_null( $this->origTitle ) ) {
 685                                 $this->pageTitle = Title::makeTitle( $this->mTargetNamespace,
 686                                         $this->origTitle->getDbKey() );
 687                         } else {
 688                                 $this->pageTitle = Title::newFromText( $this->workTitle );
 689                         }
 690                         if( is_null( $this->pageTitle ) ) {
 691                                 // Invalid page title? Ignore the page
 692                                 $this->notice( "Skipping invalid page title '$this->workTitle'" );
 693                         } else {
 694                                 $this->pageCallback( $this->workTitle );
 695                         }
 696                         break;
 697                 case "id":
 698                         if ( $this->parenttag == 'revision' ) {
 699                                 if( $this->workRevision )
 700                                         $this->workRevision->setID( $this->appenddata );
 701                         }
 702                         break;
 703                 case "text":
 704                         if( $this->workRevision )
 705                                 $this->workRevision->setText( $this->appenddata );
 706                         break;
 707                 case "username":
 708                         if( $this->workRevision )
 709                                 $this->workRevision->setUsername( $this->appenddata );
 710                         break;
 711                 case "ip":
 712                         if( $this->workRevision )
 713                                 $this->workRevision->setUserIP( $this->appenddata );
 714                         break;
 715                 case "timestamp":
 716                         if( $this->workRevision )
 717                                 $this->workRevision->setTimestamp( $this->appenddata );
 718                         break;
 719                 case "comment":
 720                         if( $this->workRevision )
 721                                 $this->workRevision->setComment( $this->appenddata );
 722                         break;
 723                 case "minor":
 724                         if( $this->workRevision )
 725                                 $this->workRevision->setMinor( true );
 726                         break;
 727                 default:
 728                         $this->debug( "Bad append: {$this->appendfield}" );
 729                 }
 730                 $this->appendfield = "";
 731                 $this->appenddata = "";
 732         }
 733
 734         function in_revision( $parser, $name, $attribs ) {
 735                 $this->debug( "in_revision $name" );
 736                 switch( $name ) {
 737                 case "id":
 738                 case "timestamp":
 739                 case "comment":
 740                 case "minor":
 741                 case "text":
 742                         $this->parenttag = "revision";
 743                         $this->appendfield = $name;
 744                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 745                         xml_set_character_data_handler( $parser, "char_append" );
 746                         break;
 747                 case "contributor":
 748                         xml_set_element_handler( $parser, "in_contributor", "out_contributor" );
 749                         break;
 750                 default:
 751                         return $this->throwXMLerror( "Element <$name> not allowed in a <revision>." );
 752                 }
 753         }
 754
 755         function out_revision( $parser, $name ) {
 756                 $this->debug( "out_revision $name" );
 757                 if( $name != "revision" ) {
 758                         return $this->throwXMLerror( "Expected </revision>, got </$name>" );
 759                 }
 760                 xml_set_element_handler( $parser, "in_page", "out_page" );
 761
 762                 if( $this->workRevision ) {
 763                         $ok = call_user_func_array( $this->mRevisionCallback,
 764                                 array( &$this->workRevision, &$this ) );
 765                         if( $ok ) {
 766                                 $this->workSuccessCount++;
 767                         }
 768                 }
 769         }
 770
 771         function in_contributor( $parser, $name, $attribs ) {
 772                 $this->debug( "in_contributor $name" );
 773                 switch( $name ) {
 774                 case "username":
 775                 case "ip":
 776                 case "id":
 777                         $this->parenttag = "contributor";
 778                         $this->appendfield = $name;
 779                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 780                         xml_set_character_data_handler( $parser, "char_append" );
 781                         break;
 782                 default:
 783                         $this->throwXMLerror( "Invalid tag <$name> in <contributor>" );
 784                 }
 785         }
 786
 787         function out_contributor( $parser, $name ) {
 788                 $this->debug( "out_contributor $name" );
 789                 if( $name != "contributor" ) {
 790                         return $this->throwXMLerror( "Expected </contributor>, got </$name>" );
 791                 }
 792                 xml_set_element_handler( $parser, "in_revision", "out_revision" );
 793         }
 794
 795 }
 796
 797 /**
 798  * @todo document (e.g. one-sentence class description).
 799  * @addtogroup SpecialPage
 800  */
 801 class ImportStringSource {
 802         function ImportStringSource( $string ) {
 803                 $this->mString = $string;
 804                 $this->mRead = false;
 805         }
 806
 807         function atEnd() {
 808                 return $this->mRead;
 809         }
 810
 811         function readChunk() {
 812                 if( $this->atEnd() ) {
 813                         return false;
 814                 } else {
 815                         $this->mRead = true;
 816                         return $this->mString;
 817                 }
 818         }
 819 }
 820
 821 /**
 822  * @todo document (e.g. one-sentence class description).
 823  * @addtogroup SpecialPage
 824  */
 825 class ImportStreamSource {
 826         function ImportStreamSource( $handle ) {
 827                 $this->mHandle = $handle;
 828         }
 829
 830         function atEnd() {
 831                 return feof( $this->mHandle );
 832         }
 833
 834         function readChunk() {
 835                 return fread( $this->mHandle, 32768 );
 836         }
 837
 838         static function newFromFile( $filename ) {
 839                 $file = @fopen( $filename, 'rt' );
 840                 if( !$file ) {
 841                         return new WikiErrorMsg( "importcantopen" );
 842                 }
 843                 return new ImportStreamSource( $file );
 844         }
 845
 846         static function newFromUpload( $fieldname = "xmlimport" ) {
 847                 $upload =& $_FILES[$fieldname];
 848
 849                 if( !isset( $upload ) || !$upload['name'] ) {
 850                         return new WikiErrorMsg( 'importnofile' );
 851                 }
 852                 if( !empty( $upload['error'] ) ) {
 853                         return new WikiErrorMsg( 'importuploaderror', $upload['error'] );
 854                 }
 855                 $fname = $upload['tmp_name'];
 856                 if( is_uploaded_file( $fname ) ) {
 857                         return ImportStreamSource::newFromFile( $fname );
 858                 } else {
 859                         return new WikiErrorMsg( 'importnofile' );
 860                 }
 861         }
 862
 863         function newFromURL( $url, $method = 'GET' ) {
 864                 wfDebug( __METHOD__ . ": opening $url\n" );
 865                 # Use the standard HTTP fetch function; it times out
 866                 # quicker and sorts out user-agent problems which might
 867                 # otherwise prevent importing from large sites, such
 868                 # as the Wikimedia cluster, etc.
 869                 $data = Http::request( $method, $url );
 870                 if( $data !== false ) {
 871                         $file = tmpfile();
 872                         fwrite( $file, $data );
 873                         fflush( $file );
 874                         fseek( $file, 0 );
 875                         return new ImportStreamSource( $file );
 876                 } else {
 877                         return new WikiErrorMsg( 'importcantopen' );
 878                 }
 879         }
 880
 881         public static function newFromInterwiki( $interwiki, $page, $history=false ) {
 882                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
 883                 if( is_null( $link ) || $link->getInterwiki() == '' ) {
 884                         return new WikiErrorMsg( 'importbadinterwiki' );
 885                 } else {
 886                         $params = $history ? 'history=1' : '';
 887                         $url = $link->getFullUrl( $params );
 888                         # For interwikis, use POST to avoid redirects.
 889                         return ImportStreamSource::newFromURL( $url, "POST" );
 890                 }
 891         }
 892 }
 893
 894
 895