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