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  * @file
  23  * @ingroup SpecialPage
  24  */
  25
  26 /**
  27  * Constructor
  28  */
  29 function wfSpecialImport( $page = '' ) {
  30         global $wgUser, $wgOut, $wgRequest, $wgTitle, $wgImportSources;
  31         global $wgImportTargetNamespace;
  32
  33         $interwiki = false;
  34         $namespace = $wgImportTargetNamespace;
  35         $frompage = '';
  36         $history = true;
  37
  38         if ( wfReadOnly() ) {
  39                 $wgOut->readOnlyPage();
  40                 return;
  41         }
  42
  43         if( $wgRequest->wasPosted() && $wgRequest->getVal( 'action' ) == 'submit') {
  44                 $isUpload = false;
  45                 $namespace = $wgRequest->getIntOrNull( 'namespace' );
  46
  47                 switch( $wgRequest->getVal( "source" ) ) {
  48                 case "upload":
  49                         $isUpload = true;
  50                         if( $wgUser->isAllowed( 'importupload' ) ) {
  51                                 $source = ImportStreamSource::newFromUpload( "xmlimport" );
  52                         } else {
  53                                 return $wgOut->permissionRequired( 'importupload' );
  54                         }
  55                         break;
  56                 case "interwiki":
  57                         $interwiki = $wgRequest->getVal( 'interwiki' );
  58                         $history = $wgRequest->getCheck( 'interwikiHistory' );
  59                         $frompage = $wgRequest->getText( "frompage" );
  60                         $source = ImportStreamSource::newFromInterwiki(
  61                                 $interwiki,
  62                                 $frompage,
  63                                 $history );
  64                         break;
  65                 default:
  66                         $source = new WikiErrorMsg( "importunknownsource" );
  67                 }
  68
  69                 if( WikiError::isError( $source ) ) {
  70                         $wgOut->wrapWikiMsg( '<p class="error">$1</p>', array( 'importfailed', $source->getMessage() ) );
  71                 } else {
  72                         $wgOut->addWikiMsg( "importstart" );
  73
  74                         $importer = new WikiImporter( $source );
  75                         if( !is_null( $namespace ) ) {
  76                                 $importer->setTargetNamespace( $namespace );
  77                         }
  78                         $reporter = new ImportReporter( $importer, $isUpload, $interwiki );
  79
  80                         $reporter->open();
  81                         $result = $importer->doImport();
  82                         $resultCount = $reporter->close();
  83
  84                         if( WikiError::isError( $result ) ) {
  85                                 # No source or XML parse error
  86                                 $wgOut->wrapWikiMsg( '<p class="error">$1</p>', array( 'importfailed', $result->getMessage() ) );
  87                         } elseif( WikiError::isError( $resultCount ) ) {
  88                                 # Zero revisions
  89                                 $wgOut->wrapWikiMsg( '<p class="error">$1</p>', array( 'importfailed', $resultCount->getMessage() ) );
  90                         } else {
  91                                 # Success!
  92                                 $wgOut->addWikiMsg( 'importsuccess' );
  93                         }
  94                         $wgOut->addWikiText( '<hr />' );
  95                 }
  96         }
  97
  98         $action = $wgTitle->getLocalUrl( 'action=submit' );
  99
 100         if( $wgUser->isAllowed( 'importupload' ) ) {
 101                 $wgOut->addWikiMsg( "importtext" );
 102                 $wgOut->addHTML(
 103                         Xml::openElement( 'fieldset' ).
 104                         Xml::element( 'legend', null, wfMsg( 'import-upload' ) ) .
 105                         Xml::openElement( 'form', array( 'enctype' => 'multipart/form-data', 'method' => 'post', 'action' => $action ) ) .
 106                         Xml::hidden( 'action', 'submit' ) .
 107                         Xml::hidden( 'source', 'upload' ) .
 108                         Xml::input( 'xmlimport', 50, '', array( 'type' => 'file' ) ) . ' ' .
 109                         Xml::submitButton( wfMsg( 'uploadbtn' ) ) .
 110                         Xml::closeElement( 'form' ) .
 111                         Xml::closeElement( 'fieldset' )
 112                 );
 113         } else {
 114                 if( empty( $wgImportSources ) ) {
 115                         $wgOut->addWikiMsg( 'importnosources' );
 116                 }
 117         }
 118
 119         if( !empty( $wgImportSources ) ) {
 120                 $wgOut->addHTML(
 121                         Xml::openElement( 'fieldset' ) .
 122                         Xml::element( 'legend', null, wfMsg( 'importinterwiki' ) ) .
 123                         Xml::openElement( 'form', array( 'method' => 'post', 'action' => $action ) ) .
 124                         wfMsgExt( 'import-interwiki-text', array( 'parse' ) ) .
 125                         Xml::hidden( 'action', 'submit' ) .
 126                         Xml::hidden( 'source', 'interwiki' ) .
 127                         Xml::openElement( 'table', array( 'id' => 'mw-import-table' ) ) .
 128                         "<tr>
 129                                 <td>" .
 130                                         Xml::openElement( 'select', array( 'name' => 'interwiki' ) )
 131                 );
 132                 foreach( $wgImportSources as $prefix ) {
 133                         $selected = ( $interwiki === $prefix ) ? ' selected="selected"' : '';
 134                         $wgOut->addHTML( Xml::option( $prefix, $prefix, $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  * @ingroup 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( $dbw, $title->getArticleId(), $comment, true );
 221                         $nullRevision->insertOn( $dbw );
 222                         $article = new Article( $title );
 223                         # Update page record
 224                         $article->updateRevisionOn( $dbw, $nullRevision );
 225                         wfRunHooks( 'NewRevisionFromEditComplete', array($article, $nullRevision, false) );
 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  * @ingroup 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 setSrc( $src ) {
 297                 $this->src = $src;
 298         }
 299
 300         function setFilename( $filename ) {
 301                 $this->filename = $filename;
 302         }
 303
 304         function setSize( $size ) {
 305                 $this->size = intval( $size );
 306         }
 307
 308         function getTitle() {
 309                 return $this->title;
 310         }
 311
 312         function getID() {
 313                 return $this->id;
 314         }
 315
 316         function getTimestamp() {
 317                 return $this->timestamp;
 318         }
 319
 320         function getUser() {
 321                 return $this->user_text;
 322         }
 323
 324         function getText() {
 325                 return $this->text;
 326         }
 327
 328         function getComment() {
 329                 return $this->comment;
 330         }
 331
 332         function getMinor() {
 333                 return $this->minor;
 334         }
 335
 336         function getSrc() {
 337                 return $this->src;
 338         }
 339
 340         function getFilename() {
 341                 return $this->filename;
 342         }
 343
 344         function getSize() {
 345                 return $this->size;
 346         }
 347
 348         function importOldRevision() {
 349                 $dbw = wfGetDB( DB_MASTER );
 350
 351                 # Sneak a single revision into place
 352                 $user = User::newFromName( $this->getUser() );
 353                 if( $user ) {
 354                         $userId = intval( $user->getId() );
 355                         $userText = $user->getName();
 356                 } else {
 357                         $userId = 0;
 358                         $userText = $this->getUser();
 359                 }
 360
 361                 // avoid memory leak...?
 362                 $linkCache = LinkCache::singleton();
 363                 $linkCache->clear();
 364
 365                 $article = new Article( $this->title );
 366                 $pageId = $article->getId();
 367                 if( $pageId == 0 ) {
 368                         # must create the page...
 369                         $pageId = $article->insertOn( $dbw );
 370                         $created = true;
 371                 } else {
 372                         $created = false;
 373
 374                         $prior = Revision::loadFromTimestamp( $dbw, $this->title, $this->timestamp );
 375                         if( !is_null( $prior ) ) {
 376                                 // FIXME: this could fail slightly for multiple matches :P
 377                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
 378                                         $this->title->getPrefixedText() . "]], timestamp " .
 379                                         $this->timestamp . "\n" );
 380                                 return false;
 381                         }
 382                 }
 383
 384                 # FIXME: Use original rev_id optionally
 385                 # FIXME: blah blah blah
 386
 387                 #if( $numrows > 0 ) {
 388                 #       return wfMsg( "importhistoryconflict" );
 389                 #}
 390
 391                 # Insert the row
 392                 $revision = new Revision( array(
 393                         'page'       => $pageId,
 394                         'text'       => $this->getText(),
 395                         'comment'    => $this->getComment(),
 396                         'user'       => $userId,
 397                         'user_text'  => $userText,
 398                         'timestamp'  => $this->timestamp,
 399                         'minor_edit' => $this->minor,
 400                         ) );
 401                 $revId = $revision->insertOn( $dbw );
 402                 $changed = $article->updateIfNewerOn( $dbw, $revision );
 403
 404                 if( $created ) {
 405                         wfDebug( __METHOD__ . ": running onArticleCreate\n" );
 406                         Article::onArticleCreate( $this->title );
 407
 408                         wfDebug( __METHOD__ . ": running create updates\n" );
 409                         $article->createUpdates( $revision );
 410
 411                 } elseif( $changed ) {
 412                         wfDebug( __METHOD__ . ": running onArticleEdit\n" );
 413                         Article::onArticleEdit( $this->title );
 414
 415                         wfDebug( __METHOD__ . ": running edit updates\n" );
 416                         $article->editUpdates(
 417                                 $this->getText(),
 418                                 $this->getComment(),
 419                                 $this->minor,
 420                                 $this->timestamp,
 421                                 $revId );
 422                 }
 423
 424                 return true;
 425         }
 426
 427         function importUpload() {
 428                 wfDebug( __METHOD__ . ": STUB\n" );
 429
 430                 /**
 431                         // from file revert...
 432                         $source = $this->file->getArchiveVirtualUrl( $this->oldimage );
 433                         $comment = $wgRequest->getText( 'wpComment' );
 434                         // TODO: Preserve file properties from database instead of reloading from file
 435                         $status = $this->file->upload( $source, $comment, $comment );
 436                         if( $status->isGood() ) {
 437                 */
 438
 439                 /**
 440                         // from file upload...
 441                 $this->mLocalFile = wfLocalFile( $nt );
 442                 $this->mDestName = $this->mLocalFile->getName();
 443                 //....
 444                         $status = $this->mLocalFile->upload( $this->mTempPath, $this->mComment, $pageText,
 445                         File::DELETE_SOURCE, $this->mFileProps );
 446                         if ( !$status->isGood() ) {
 447                                 $resultDetails = array( 'internal' => $status->getWikiText() );
 448                 */
 449
 450                 // @fixme upload() uses $wgUser, which is wrong here
 451                 // it may also create a page without our desire, also wrong potentially.
 452                 // and, it will record a *current* upload, but we might want an archive version here
 453
 454                 $file = wfLocalFile( $this->getTitle() );
 455                 if( !$file ) {
 456                         var_dump( $file );
 457                         wfDebug( "IMPORT: Bad file. :(\n" );
 458                         return false;
 459                 }
 460
 461                 $source = $this->downloadSource();
 462                 if( !$source ) {
 463                         wfDebug( "IMPORT: Could not fetch remote file. :(\n" );
 464                         return false;
 465                 }
 466
 467                 $status = $file->upload( $source,
 468                         $this->getComment(),
 469                         $this->getComment(), // Initial page, if none present...
 470                         File::DELETE_SOURCE,
 471                         false, // props...
 472                         $this->getTimestamp() );
 473
 474                 if( $status->isGood() ) {
 475                         // yay?
 476                         wfDebug( "IMPORT: is ok?\n" );
 477                         return true;
 478                 }
 479
 480                 wfDebug( "IMPORT: is bad? " . $status->getXml() . "\n" );
 481                 return false;
 482
 483         }
 484
 485         function downloadSource() {
 486                 global $wgEnableUploads;
 487                 if( !$wgEnableUploads ) {
 488                         return false;
 489                 }
 490
 491                 $tempo = tempnam( wfTempDir(), 'download' );
 492                 $f = fopen( $tempo, 'wb' );
 493                 if( !$f ) {
 494                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
 495                         return false;
 496                 }
 497
 498                 // @fixme!
 499                 $src = $this->getSrc();
 500                 $data = Http::get( $src );
 501                 if( !$data ) {
 502                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
 503                         fclose( $f );
 504                         unlink( $tempo );
 505                         return false;
 506                 }
 507
 508                 fwrite( $f, $data );
 509                 fclose( $f );
 510
 511                 return $tempo;
 512         }
 513
 514 }
 515
 516 /**
 517  * implements Special:Import
 518  * @ingroup SpecialPage
 519  */
 520 class WikiImporter {
 521         var $mDebug = false;
 522         var $mSource = null;
 523         var $mPageCallback = null;
 524         var $mPageOutCallback = null;
 525         var $mRevisionCallback = null;
 526         var $mUploadCallback = null;
 527         var $mTargetNamespace = null;
 528         var $lastfield;
 529         var $tagStack = array();
 530
 531         function __construct( $source ) {
 532                 $this->setRevisionCallback( array( $this, "importRevision" ) );
 533                 $this->setUploadCallback( array( $this, "importUpload" ) );
 534                 $this->mSource = $source;
 535         }
 536
 537         function throwXmlError( $err ) {
 538                 $this->debug( "FAILURE: $err" );
 539                 wfDebug( "WikiImporter XML error: $err\n" );
 540         }
 541
 542         # --------------
 543
 544         function doImport() {
 545                 if( empty( $this->mSource ) ) {
 546                         return new WikiErrorMsg( "importnotext" );
 547                 }
 548
 549                 $parser = xml_parser_create( "UTF-8" );
 550
 551                 # case folding violates XML standard, turn it off
 552                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 553
 554                 xml_set_object( $parser, $this );
 555                 xml_set_element_handler( $parser, "in_start", "" );
 556
 557                 $offset = 0; // for context extraction on error reporting
 558                 do {
 559                         $chunk = $this->mSource->readChunk();
 560                         if( !xml_parse( $parser, $chunk, $this->mSource->atEnd() ) ) {
 561                                 wfDebug( "WikiImporter::doImport encountered XML parsing error\n" );
 562                                 return new WikiXmlError( $parser, wfMsgHtml( 'import-parse-failure' ), $chunk, $offset );
 563                         }
 564                         $offset += strlen( $chunk );
 565                 } while( $chunk !== false && !$this->mSource->atEnd() );
 566                 xml_parser_free( $parser );
 567
 568                 return true;
 569         }
 570
 571         function debug( $data ) {
 572                 if( $this->mDebug ) {
 573                         wfDebug( "IMPORT: $data\n" );
 574                 }
 575         }
 576
 577         function notice( $data ) {
 578                 global $wgCommandLineMode;
 579                 if( $wgCommandLineMode ) {
 580                         print "$data\n";
 581                 } else {
 582                         global $wgOut;
 583                         $wgOut->addHTML( "<li>" . htmlspecialchars( $data ) . "</li>\n" );
 584                 }
 585         }
 586
 587         /**
 588          * Set debug mode...
 589          */
 590         function setDebug( $debug ) {
 591                 $this->mDebug = $debug;
 592         }
 593
 594         /**
 595          * Sets the action to perform as each new page in the stream is reached.
 596          * @param $callback callback
 597          * @return callback
 598          */
 599         function setPageCallback( $callback ) {
 600                 $previous = $this->mPageCallback;
 601                 $this->mPageCallback = $callback;
 602                 return $previous;
 603         }
 604
 605         /**
 606          * Sets the action to perform as each page in the stream is completed.
 607          * Callback accepts the page title (as a Title object), a second object
 608          * with the original title form (in case it's been overridden into a
 609          * local namespace), and a count of revisions.
 610          *
 611          * @param $callback callback
 612          * @return callback
 613          */
 614         function setPageOutCallback( $callback ) {
 615                 $previous = $this->mPageOutCallback;
 616                 $this->mPageOutCallback = $callback;
 617                 return $previous;
 618         }
 619
 620         /**
 621          * Sets the action to perform as each page revision is reached.
 622          * @param $callback callback
 623          * @return callback
 624          */
 625         function setRevisionCallback( $callback ) {
 626                 $previous = $this->mRevisionCallback;
 627                 $this->mRevisionCallback = $callback;
 628                 return $previous;
 629         }
 630
 631         /**
 632          * Sets the action to perform as each file upload version is reached.
 633          * @param $callback callback
 634          * @return callback
 635          */
 636         function setUploadCallback( $callback ) {
 637                 $previous = $this->mUploadCallback;
 638                 $this->mUploadCallback = $callback;
 639                 return $previous;
 640         }
 641
 642         /**
 643          * Set a target namespace to override the defaults
 644          */
 645         function setTargetNamespace( $namespace ) {
 646                 if( is_null( $namespace ) ) {
 647                         // Don't override namespaces
 648                         $this->mTargetNamespace = null;
 649                 } elseif( $namespace >= 0 ) {
 650                         // FIXME: Check for validity
 651                         $this->mTargetNamespace = intval( $namespace );
 652                 } else {
 653                         return false;
 654                 }
 655         }
 656
 657         /**
 658          * Default per-revision callback, performs the import.
 659          * @param $revision WikiRevision
 660          * @private
 661          */
 662         function importRevision( $revision ) {
 663                 $dbw = wfGetDB( DB_MASTER );
 664                 return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
 665         }
 666
 667         /**
 668          * Dummy for now...
 669          */
 670         function importUpload( $revision ) {
 671                 //$dbw = wfGetDB( DB_MASTER );
 672                 //return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
 673                 return false;
 674         }
 675
 676         /**
 677          * Alternate per-revision callback, for debugging.
 678          * @param $revision WikiRevision
 679          * @private
 680          */
 681         function debugRevisionHandler( &$revision ) {
 682                 $this->debug( "Got revision:" );
 683                 if( is_object( $revision->title ) ) {
 684                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 685                 } else {
 686                         $this->debug( "-- Title: <invalid>" );
 687                 }
 688                 $this->debug( "-- User: " . $revision->user_text );
 689                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 690                 $this->debug( "-- Comment: " . $revision->comment );
 691                 $this->debug( "-- Text: " . $revision->text );
 692         }
 693
 694         /**
 695          * Notify the callback function when a new <page> is reached.
 696          * @param $title Title
 697          * @private
 698          */
 699         function pageCallback( $title ) {
 700                 if( is_callable( $this->mPageCallback ) ) {
 701                         call_user_func( $this->mPageCallback, $title );
 702                 }
 703         }
 704
 705         /**
 706          * Notify the callback function when a </page> is closed.
 707          * @param $title Title
 708          * @param $origTitle Title
 709          * @param $revisionCount int
 710          * @param $successCount Int: number of revisions for which callback returned true
 711          * @private
 712          */
 713         function pageOutCallback( $title, $origTitle, $revisionCount, $successCount ) {
 714                 if( is_callable( $this->mPageOutCallback ) ) {
 715                         call_user_func( $this->mPageOutCallback, $title, $origTitle,
 716                                 $revisionCount, $successCount );
 717                 }
 718         }
 719
 720
 721         # XML parser callbacks from here out -- beware!
 722         function donothing( $parser, $x, $y="" ) {
 723                 #$this->debug( "donothing" );
 724         }
 725
 726         function in_start( $parser, $name, $attribs ) {
 727                 $this->debug( "in_start $name" );
 728                 if( $name != "mediawiki" ) {
 729                         return $this->throwXMLerror( "Expected <mediawiki>, got <$name>" );
 730                 }
 731                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 732         }
 733
 734         function in_mediawiki( $parser, $name, $attribs ) {
 735                 $this->debug( "in_mediawiki $name" );
 736                 if( $name == 'siteinfo' ) {
 737                         xml_set_element_handler( $parser, "in_siteinfo", "out_siteinfo" );
 738                 } elseif( $name == 'page' ) {
 739                         $this->push( $name );
 740                         $this->workRevisionCount = 0;
 741                         $this->workSuccessCount = 0;
 742                         $this->uploadCount = 0;
 743                         $this->uploadSuccessCount = 0;
 744                         xml_set_element_handler( $parser, "in_page", "out_page" );
 745                 } else {
 746                         return $this->throwXMLerror( "Expected <page>, got <$name>" );
 747                 }
 748         }
 749         function out_mediawiki( $parser, $name ) {
 750                 $this->debug( "out_mediawiki $name" );
 751                 if( $name != "mediawiki" ) {
 752                         return $this->throwXMLerror( "Expected </mediawiki>, got </$name>" );
 753                 }
 754                 xml_set_element_handler( $parser, "donothing", "donothing" );
 755         }
 756
 757
 758         function in_siteinfo( $parser, $name, $attribs ) {
 759                 // no-ops for now
 760                 $this->debug( "in_siteinfo $name" );
 761                 switch( $name ) {
 762                 case "sitename":
 763                 case "base":
 764                 case "generator":
 765                 case "case":
 766                 case "namespaces":
 767                 case "namespace":
 768                         break;
 769                 default:
 770                         return $this->throwXMLerror( "Element <$name> not allowed in <siteinfo>." );
 771                 }
 772         }
 773
 774         function out_siteinfo( $parser, $name ) {
 775                 if( $name == "siteinfo" ) {
 776                         xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 777                 }
 778         }
 779
 780
 781         function in_page( $parser, $name, $attribs ) {
 782                 $this->debug( "in_page $name" );
 783                 switch( $name ) {
 784                 case "id":
 785                 case "title":
 786                 case "restrictions":
 787                         $this->appendfield = $name;
 788                         $this->appenddata = "";
 789                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 790                         xml_set_character_data_handler( $parser, "char_append" );
 791                         break;
 792                 case "revision":
 793                         $this->push( "revision" );
 794                         if( is_object( $this->pageTitle ) ) {
 795                                 $this->workRevision = new WikiRevision;
 796                                 $this->workRevision->setTitle( $this->pageTitle );
 797                                 $this->workRevisionCount++;
 798                         } else {
 799                                 // Skipping items due to invalid page title
 800                                 $this->workRevision = null;
 801                         }
 802                         xml_set_element_handler( $parser, "in_revision", "out_revision" );
 803                         break;
 804                 case "upload":
 805                         $this->push( "upload" );
 806                         if( is_object( $this->pageTitle ) ) {
 807                                 $this->workRevision = new WikiRevision;
 808                                 $this->workRevision->setTitle( $this->pageTitle );
 809                                 $this->uploadCount++;
 810                         } else {
 811                                 // Skipping items due to invalid page title
 812                                 $this->workRevision = null;
 813                         }
 814                         xml_set_element_handler( $parser, "in_upload", "out_upload" );
 815                         break;
 816                 default:
 817                         return $this->throwXMLerror( "Element <$name> not allowed in a <page>." );
 818                 }
 819         }
 820
 821         function out_page( $parser, $name ) {
 822                 $this->debug( "out_page $name" );
 823                 $this->pop();
 824                 if( $name != "page" ) {
 825                         return $this->throwXMLerror( "Expected </page>, got </$name>" );
 826                 }
 827                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 828
 829                 $this->pageOutCallback( $this->pageTitle, $this->origTitle,
 830                         $this->workRevisionCount, $this->workSuccessCount );
 831
 832                 $this->workTitle = null;
 833                 $this->workRevision = null;
 834                 $this->workRevisionCount = 0;
 835                 $this->workSuccessCount = 0;
 836                 $this->pageTitle = null;
 837                 $this->origTitle = null;
 838         }
 839
 840         function in_nothing( $parser, $name, $attribs ) {
 841                 $this->debug( "in_nothing $name" );
 842                 return $this->throwXMLerror( "No child elements allowed here; got <$name>" );
 843         }
 844         function char_append( $parser, $data ) {
 845                 $this->debug( "char_append '$data'" );
 846                 $this->appenddata .= $data;
 847         }
 848         function out_append( $parser, $name ) {
 849                 $this->debug( "out_append $name" );
 850                 if( $name != $this->appendfield ) {
 851                         return $this->throwXMLerror( "Expected </{$this->appendfield}>, got </$name>" );
 852                 }
 853
 854                 switch( $this->appendfield ) {
 855                 case "title":
 856                         $this->workTitle = $this->appenddata;
 857                         $this->origTitle = Title::newFromText( $this->workTitle );
 858                         if( !is_null( $this->mTargetNamespace ) && !is_null( $this->origTitle ) ) {
 859                                 $this->pageTitle = Title::makeTitle( $this->mTargetNamespace,
 860                                         $this->origTitle->getDBkey() );
 861                         } else {
 862                                 $this->pageTitle = Title::newFromText( $this->workTitle );
 863                         }
 864                         if( is_null( $this->pageTitle ) ) {
 865                                 // Invalid page title? Ignore the page
 866                                 $this->notice( "Skipping invalid page title '$this->workTitle'" );
 867                         } else {
 868                                 $this->pageCallback( $this->workTitle );
 869                         }
 870                         break;
 871                 case "id":
 872                         if ( $this->parentTag() == 'revision' ) {
 873                                 if( $this->workRevision )
 874                                         $this->workRevision->setID( $this->appenddata );
 875                         }
 876                         break;
 877                 case "text":
 878                         if( $this->workRevision )
 879                                 $this->workRevision->setText( $this->appenddata );
 880                         break;
 881                 case "username":
 882                         if( $this->workRevision )
 883                                 $this->workRevision->setUsername( $this->appenddata );
 884                         break;
 885                 case "ip":
 886                         if( $this->workRevision )
 887                                 $this->workRevision->setUserIP( $this->appenddata );
 888                         break;
 889                 case "timestamp":
 890                         if( $this->workRevision )
 891                                 $this->workRevision->setTimestamp( $this->appenddata );
 892                         break;
 893                 case "comment":
 894                         if( $this->workRevision )
 895                                 $this->workRevision->setComment( $this->appenddata );
 896                         break;
 897                 case "minor":
 898                         if( $this->workRevision )
 899                                 $this->workRevision->setMinor( true );
 900                         break;
 901                 case "filename":
 902                         if( $this->workRevision )
 903                                 $this->workRevision->setFilename( $this->appenddata );
 904                         break;
 905                 case "src":
 906                         if( $this->workRevision )
 907                                 $this->workRevision->setSrc( $this->appenddata );
 908                         break;
 909                 case "size":
 910                         if( $this->workRevision )
 911                                 $this->workRevision->setSize( intval( $this->appenddata ) );
 912                         break;
 913                 default:
 914                         $this->debug( "Bad append: {$this->appendfield}" );
 915                 }
 916                 $this->appendfield = "";
 917                 $this->appenddata = "";
 918
 919                 $parent = $this->parentTag();
 920                 xml_set_element_handler( $parser, "in_$parent", "out_$parent" );
 921                 xml_set_character_data_handler( $parser, "donothing" );
 922         }
 923
 924         function in_revision( $parser, $name, $attribs ) {
 925                 $this->debug( "in_revision $name" );
 926                 switch( $name ) {
 927                 case "id":
 928                 case "timestamp":
 929                 case "comment":
 930                 case "minor":
 931                 case "text":
 932                         $this->appendfield = $name;
 933                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 934                         xml_set_character_data_handler( $parser, "char_append" );
 935                         break;
 936                 case "contributor":
 937                         $this->push( "contributor" );
 938                         xml_set_element_handler( $parser, "in_contributor", "out_contributor" );
 939                         break;
 940                 default:
 941                         return $this->throwXMLerror( "Element <$name> not allowed in a <revision>." );
 942                 }
 943         }
 944
 945         function out_revision( $parser, $name ) {
 946                 $this->debug( "out_revision $name" );
 947                 $this->pop();
 948                 if( $name != "revision" ) {
 949                         return $this->throwXMLerror( "Expected </revision>, got </$name>" );
 950                 }
 951                 xml_set_element_handler( $parser, "in_page", "out_page" );
 952
 953                 if( $this->workRevision ) {
 954                         $ok = call_user_func_array( $this->mRevisionCallback,
 955                                 array( $this->workRevision, $this ) );
 956                         if( $ok ) {
 957                                 $this->workSuccessCount++;
 958                         }
 959                 }
 960         }
 961
 962         function in_upload( $parser, $name, $attribs ) {
 963                 $this->debug( "in_upload $name" );
 964                 switch( $name ) {
 965                 case "timestamp":
 966                 case "comment":
 967                 case "text":
 968                 case "filename":
 969                 case "src":
 970                 case "size":
 971                         $this->appendfield = $name;
 972                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 973                         xml_set_character_data_handler( $parser, "char_append" );
 974                         break;
 975                 case "contributor":
 976                         $this->push( "contributor" );
 977                         xml_set_element_handler( $parser, "in_contributor", "out_contributor" );
 978                         break;
 979                 default:
 980                         return $this->throwXMLerror( "Element <$name> not allowed in an <upload>." );
 981                 }
 982         }
 983
 984         function out_upload( $parser, $name ) {
 985                 $this->debug( "out_revision $name" );
 986                 $this->pop();
 987                 if( $name != "upload" ) {
 988                         return $this->throwXMLerror( "Expected </upload>, got </$name>" );
 989                 }
 990                 xml_set_element_handler( $parser, "in_page", "out_page" );
 991
 992                 if( $this->workRevision ) {
 993                         $ok = call_user_func_array( $this->mUploadCallback,
 994                                 array( $this->workRevision, $this ) );
 995                         if( $ok ) {
 996                                 $this->workUploadSuccessCount++;
 997                         }
 998                 }
 999         }
1000
1001         function in_contributor( $parser, $name, $attribs ) {
1002                 $this->debug( "in_contributor $name" );
1003                 switch( $name ) {
1004                 case "username":
1005                 case "ip":
1006                 case "id":
1007                         $this->appendfield = $name;
1008                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
1009                         xml_set_character_data_handler( $parser, "char_append" );
1010                         break;
1011                 default:
1012                         $this->throwXMLerror( "Invalid tag <$name> in <contributor>" );
1013                 }
1014         }
1015
1016         function out_contributor( $parser, $name ) {
1017                 $this->debug( "out_contributor $name" );
1018                 $this->pop();
1019                 if( $name != "contributor" ) {
1020                         return $this->throwXMLerror( "Expected </contributor>, got </$name>" );
1021                 }
1022                 $parent = $this->parentTag();
1023                 xml_set_element_handler( $parser, "in_$parent", "out_$parent" );
1024         }
1025
1026         private function push( $name ) {
1027                 array_push( $this->tagStack, $name );
1028                 $this->debug( "PUSH $name" );
1029         }
1030
1031         private function pop() {
1032                 $name = array_pop( $this->tagStack );
1033                 $this->debug( "POP $name" );
1034                 return $name;
1035         }
1036
1037         private function parentTag() {
1038                 $name = $this->tagStack[count( $this->tagStack ) - 1];
1039                 $this->debug( "PARENT $name" );
1040                 return $name;
1041         }
1042
1043 }
1044
1045 /**
1046  * @todo document (e.g. one-sentence class description).
1047  * @ingroup SpecialPage
1048  */
1049 class ImportStringSource {
1050         function __construct( $string ) {
1051                 $this->mString = $string;
1052                 $this->mRead = false;
1053         }
1054
1055         function atEnd() {
1056                 return $this->mRead;
1057         }
1058
1059         function readChunk() {
1060                 if( $this->atEnd() ) {
1061                         return false;
1062                 } else {
1063                         $this->mRead = true;
1064                         return $this->mString;
1065                 }
1066         }
1067 }
1068
1069 /**
1070  * @todo document (e.g. one-sentence class description).
1071  * @ingroup SpecialPage
1072  */
1073 class ImportStreamSource {
1074         function __construct( $handle ) {
1075                 $this->mHandle = $handle;
1076         }
1077
1078         function atEnd() {
1079                 return feof( $this->mHandle );
1080         }
1081
1082         function readChunk() {
1083                 return fread( $this->mHandle, 32768 );
1084         }
1085
1086         static function newFromFile( $filename ) {
1087                 $file = @fopen( $filename, 'rt' );
1088                 if( !$file ) {
1089                         return new WikiErrorMsg( "importcantopen" );
1090                 }
1091                 return new ImportStreamSource( $file );
1092         }
1093
1094         static function newFromUpload( $fieldname = "xmlimport" ) {
1095                 $upload =& $_FILES[$fieldname];
1096
1097                 if( !isset( $upload ) || !$upload['name'] ) {
1098                         return new WikiErrorMsg( 'importnofile' );
1099                 }
1100                 if( !empty( $upload['error'] ) ) {
1101                         switch($upload['error']){
1102                                 case 1: # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1103                                         return new WikiErrorMsg( 'importuploaderrorsize' );
1104                                 case 2: # The uploaded file exceeds the MAX_FILE_SIZE directive that was specified in the HTML form.
1105                                         return new WikiErrorMsg( 'importuploaderrorsize' );
1106                                 case 3: # The uploaded file was only partially uploaded
1107                                         return new WikiErrorMsg( 'importuploaderrorpartial' );
1108                             case 6: #Missing a temporary folder. Introduced in PHP 4.3.10 and PHP 5.0.3.
1109                                 return new WikiErrorMsg( 'importuploaderrortemp' );
1110                             # case else: # Currently impossible
1111                         }
1112
1113                 }
1114                 $fname = $upload['tmp_name'];
1115                 if( is_uploaded_file( $fname ) ) {
1116                         return ImportStreamSource::newFromFile( $fname );
1117                 } else {
1118                         return new WikiErrorMsg( 'importnofile' );
1119                 }
1120         }
1121
1122         static function newFromURL( $url, $method = 'GET' ) {
1123                 wfDebug( __METHOD__ . ": opening $url\n" );
1124                 # Use the standard HTTP fetch function; it times out
1125                 # quicker and sorts out user-agent problems which might
1126                 # otherwise prevent importing from large sites, such
1127                 # as the Wikimedia cluster, etc.
1128                 $data = Http::request( $method, $url );
1129                 if( $data !== false ) {
1130                         $file = tmpfile();
1131                         fwrite( $file, $data );
1132                         fflush( $file );
1133                         fseek( $file, 0 );
1134                         return new ImportStreamSource( $file );
1135                 } else {
1136                         return new WikiErrorMsg( 'importcantopen' );
1137                 }
1138         }
1139
1140         public static function newFromInterwiki( $interwiki, $page, $history=false ) {
1141                 if( $page == '' ) {
1142                         return new WikiErrorMsg( 'import-noarticle' );
1143                 }
1144                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1145                 if( is_null( $link ) || $link->getInterwiki() == '' ) {
1146                         return new WikiErrorMsg( 'importbadinterwiki' );
1147                 } else {
1148                         $params = $history ? 'history=1' : '';
1149                         $url = $link->getFullUrl( $params );
1150                         # For interwikis, use POST to avoid redirects.
1151                         return ImportStreamSource::newFromURL( $url, "POST" );
1152                 }
1153         }
1154 }