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->wrapWikiMsg( '<p class="error">$1</p>', array( 'importfailed', $source->getMessage() ) );
  70                 } else {
  71                         $wgOut->addWikiMsg( "importstart" );
  72
  73                         $importer = new WikiImporter( $source );
  74                         if( !is_null( $namespace ) ) {
  75                                 $importer->setTargetNamespace( $namespace );
  76                         }
  77                         $reporter = new ImportReporter( $importer, $isUpload, $interwiki );
  78
  79                         $reporter->open();
  80                         $result = $importer->doImport();
  81                         $resultCount = $reporter->close();
  82
  83                         if( WikiError::isError( $result ) ) {
  84                                 # No source or XML parse error
  85                                 $wgOut->wrapWikiMsg( '<p class="error">$1</p>', array( 'importfailed', $result->getMessage() ) );
  86                         } elseif( WikiError::isError( $resultCount ) ) {
  87                                 # Zero revisions
  88                                 $wgOut->wrapWikiMsg( '<p class="error">$1</p>', array( 'importfailed', $resultCount->getMessage() ) );
  89                         } else {
  90                                 # Success!
  91                                 $wgOut->addWikiMsg( 'importsuccess' );
  92                         }
  93                         $wgOut->addWikiText( '<hr />' );
  94                 }
  95         }
  96
  97         $action = $wgTitle->getLocalUrl( 'action=submit' );
  98
  99         if( $wgUser->isAllowed( 'importupload' ) ) {
 100                 $wgOut->addWikiMsg( "importtext" );
 101                 $wgOut->addHTML(
 102                         Xml::openElement( 'fieldset' ).
 103                         Xml::element( 'legend', null, wfMsg( 'upload' ) ) .
 104                         Xml::openElement( 'form', array( 'enctype' => 'multipart/form-data', 'method' => 'post', 'action' => $action ) ) .
 105                         Xml::hidden( 'action', 'submit' ) .
 106                         Xml::hidden( 'source', 'upload' ) .
 107                         Xml::input( 'xmlimport', 50, '', array( 'type' => 'file' ) ) . ' ' .
 108                         Xml::submitButton( wfMsg( 'uploadbtn' ) ) .
 109                         Xml::closeElement( 'form' ) .
 110                         Xml::closeElement( 'fieldset' )
 111                 );
 112         } else {
 113                 if( empty( $wgImportSources ) ) {
 114                         $wgOut->addWikiMsg( 'importnosources' );
 115                 }
 116         }
 117
 118         if( !empty( $wgImportSources ) ) {
 119                 $wgOut->addHTML(
 120                         Xml::openElement( 'fieldset' ) .
 121                         Xml::element( 'legend', null, wfMsg( 'importinterwiki' ) ) .
 122                         Xml::openElement( 'form', array( 'method' => 'post', 'action' => $action ) ) .
 123                         wfMsgExt( 'import-interwiki-text', array( 'parse' ) ) .
 124                         Xml::hidden( 'action', 'submit' ) .
 125                         Xml::hidden( 'source', 'interwiki' ) .
 126                         Xml::openElement( 'table', array( 'id' => 'mw-import-table' ) ) .
 127                         "<tr>
 128                                 <td>" .
 129                                         Xml::openElement( 'select', array( 'name' => 'interwiki' ) )
 130                 );
 131                 foreach( $wgImportSources as $prefix ) {
 132                         $selected = ( $interwiki === $prefix ) ? ' selected="selected"' : '';
 133                         $wgOut->addHTML( Xml::option( $prefix, $prefix, $selected ) );
 134                 }
 135                 $wgOut->addHTML(
 136                                         Xml::closeElement( 'select' ) .
 137                                 "</td>
 138                                 <td>" .
 139                                         Xml::input( 'frompage', 50, $frompage ) .
 140                                 "</td>
 141                         </tr>
 142                         <tr>
 143                                 <td>
 144                                 </td>
 145                                 <td>" .
 146                                         Xml::checkLabel( wfMsg( 'import-interwiki-history' ), 'interwikiHistory', 'interwikiHistory', $history ) .
 147                                 "</td>
 148                         </tr>
 149                         <tr>
 150                                 <td>
 151                                 </td>
 152                                 <td>" .
 153                                         Xml::label( wfMsg( 'import-interwiki-namespace' ), 'namespace' ) .
 154                                         Xml::namespaceSelector( $namespace, '' ) .
 155                                 "</td>
 156                         </tr>
 157                         <tr>
 158                                 <td>
 159                                 </td>
 160                                 <td>" .
 161                                         Xml::submitButton( wfMsg( 'import-interwiki-submit' ) ) .
 162                                 "</td>
 163                         </tr>" .
 164                         Xml::closeElement( 'table' ).
 165                         Xml::closeElement( 'form' ) .
 166                         Xml::closeElement( 'fieldset' )
 167                 );
 168         }
 169 }
 170
 171 /**
 172  * Reporting callback
 173  * @addtogroup SpecialPage
 174  */
 175 class ImportReporter {
 176         function __construct( $importer, $upload, $interwiki ) {
 177                 $importer->setPageOutCallback( array( $this, 'reportPage' ) );
 178                 $this->mPageCount = 0;
 179                 $this->mIsUpload = $upload;
 180                 $this->mInterwiki = $interwiki;
 181         }
 182
 183         function open() {
 184                 global $wgOut;
 185                 $wgOut->addHtml( "<ul>\n" );
 186         }
 187
 188         function reportPage( $title, $origTitle, $revisionCount, $successCount ) {
 189                 global $wgOut, $wgUser, $wgLang, $wgContLang;
 190
 191                 $skin = $wgUser->getSkin();
 192
 193                 $this->mPageCount++;
 194
 195                 $localCount = $wgLang->formatNum( $successCount );
 196                 $contentCount = $wgContLang->formatNum( $successCount );
 197
 198                 if( $successCount > 0 ) {
 199                         $wgOut->addHtml( "<li>" . $skin->makeKnownLinkObj( $title ) . " " .
 200                                 wfMsgExt( 'import-revision-count', array( 'parsemag', 'escape' ), $localCount ) .
 201                                 "</li>\n"
 202                         );
 203
 204                         $log = new LogPage( 'import' );
 205                         if( $this->mIsUpload ) {
 206                                 $detail = wfMsgExt( 'import-logentry-upload-detail', array( 'content', 'parsemag' ),
 207                                         $contentCount );
 208                                 $log->addEntry( 'upload', $title, $detail );
 209                         } else {
 210                                 $interwiki = '[[:' . $this->mInterwiki . ':' .
 211                                         $origTitle->getPrefixedText() . ']]';
 212                                 $detail = wfMsgExt( 'import-logentry-interwiki-detail', array( 'content', 'parsemag' ),
 213                                         $contentCount, $interwiki );
 214                                 $log->addEntry( 'interwiki', $title, $detail );
 215                         }
 216
 217                         $comment = $detail; // quick
 218                         $dbw = wfGetDB( DB_MASTER );
 219                         $nullRevision = Revision::newNullRevision(
 220                                 $dbw, $title->getArticleId(), $comment, true );
 221                         $nullRevision->insertOn( $dbw );
 222                         # Update page record
 223                         $article = new Article( $title );
 224                         $article->updateRevisionOn( $dbw, $nullRevision );
 225                 } else {
 226                         $wgOut->addHtml( '<li>' . wfMsgHtml( 'import-nonewrevisions' ) . '</li>' );
 227                 }
 228         }
 229
 230         function close() {
 231                 global $wgOut;
 232                 if( $this->mPageCount == 0 ) {
 233                         $wgOut->addHtml( "</ul>\n" );
 234                         return new WikiErrorMsg( "importnopages" );
 235                 }
 236                 $wgOut->addHtml( "</ul>\n" );
 237
 238                 return $this->mPageCount;
 239         }
 240 }
 241
 242 /**
 243  *
 244  * @addtogroup SpecialPage
 245  */
 246 class WikiRevision {
 247         var $title = null;
 248         var $id = 0;
 249         var $timestamp = "20010115000000";
 250         var $user = 0;
 251         var $user_text = "";
 252         var $text = "";
 253         var $comment = "";
 254         var $minor = false;
 255
 256         function setTitle( $title ) {
 257                 if( is_object( $title ) ) {
 258                         $this->title = $title;
 259                 } elseif( is_null( $title ) ) {
 260                         throw new MWException( "WikiRevision given a null title in import. You may need to adjust \$wgLegalTitleChars." );
 261                 } else {
 262                         throw new MWException( "WikiRevision given non-object title in import." );
 263                 }
 264         }
 265
 266         function setID( $id ) {
 267                 $this->id = $id;
 268         }
 269
 270         function setTimestamp( $ts ) {
 271                 # 2003-08-05T18:30:02Z
 272                 $this->timestamp = wfTimestamp( TS_MW, $ts );
 273         }
 274
 275         function setUsername( $user ) {
 276                 $this->user_text = $user;
 277         }
 278
 279         function setUserIP( $ip ) {
 280                 $this->user_text = $ip;
 281         }
 282
 283         function setText( $text ) {
 284                 $this->text = $text;
 285         }
 286
 287         function setComment( $text ) {
 288                 $this->comment = $text;
 289         }
 290
 291         function setMinor( $minor ) {
 292                 $this->minor = (bool)$minor;
 293         }
 294
 295         function setSrc( $src ) {
 296                 $this->src = $src;
 297         }
 298
 299         function setFilename( $filename ) {
 300                 $this->filename = $filename;
 301         }
 302
 303         function setSize( $size ) {
 304                 $this->size = intval( $size );
 305         }
 306
 307         function getTitle() {
 308                 return $this->title;
 309         }
 310
 311         function getID() {
 312                 return $this->id;
 313         }
 314
 315         function getTimestamp() {
 316                 return $this->timestamp;
 317         }
 318
 319         function getUser() {
 320                 return $this->user_text;
 321         }
 322
 323         function getText() {
 324                 return $this->text;
 325         }
 326
 327         function getComment() {
 328                 return $this->comment;
 329         }
 330
 331         function getMinor() {
 332                 return $this->minor;
 333         }
 334
 335         function getSrc() {
 336                 return $this->src;
 337         }
 338
 339         function getFilename() {
 340                 return $this->filename;
 341         }
 342
 343         function getSize() {
 344                 return $this->size;
 345         }
 346
 347         function importOldRevision() {
 348                 $dbw = wfGetDB( DB_MASTER );
 349
 350                 # Sneak a single revision into place
 351                 $user = User::newFromName( $this->getUser() );
 352                 if( $user ) {
 353                         $userId = intval( $user->getId() );
 354                         $userText = $user->getName();
 355                 } else {
 356                         $userId = 0;
 357                         $userText = $this->getUser();
 358                 }
 359
 360                 // avoid memory leak...?
 361                 $linkCache = LinkCache::singleton();
 362                 $linkCache->clear();
 363
 364                 $article = new Article( $this->title );
 365                 $pageId = $article->getId();
 366                 if( $pageId == 0 ) {
 367                         # must create the page...
 368                         $pageId = $article->insertOn( $dbw );
 369                         $created = true;
 370                 } else {
 371                         $created = false;
 372
 373                         $prior = Revision::loadFromTimestamp( $dbw, $this->title, $this->timestamp );
 374                         if( !is_null( $prior ) ) {
 375                                 // FIXME: this could fail slightly for multiple matches :P
 376                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
 377                                         $this->title->getPrefixedText() . "]], timestamp " .
 378                                         $this->timestamp . "\n" );
 379                                 return false;
 380                         }
 381                 }
 382
 383                 # FIXME: Use original rev_id optionally
 384                 # FIXME: blah blah blah
 385
 386                 #if( $numrows > 0 ) {
 387                 #       return wfMsg( "importhistoryconflict" );
 388                 #}
 389
 390                 # Insert the row
 391                 $revision = new Revision( array(
 392                         'page'       => $pageId,
 393                         'text'       => $this->getText(),
 394                         'comment'    => $this->getComment(),
 395                         'user'       => $userId,
 396                         'user_text'  => $userText,
 397                         'timestamp'  => $this->timestamp,
 398                         'minor_edit' => $this->minor,
 399                         ) );
 400                 $revId = $revision->insertOn( $dbw );
 401                 $changed = $article->updateIfNewerOn( $dbw, $revision );
 402
 403                 if( $created ) {
 404                         wfDebug( __METHOD__ . ": running onArticleCreate\n" );
 405                         Article::onArticleCreate( $this->title );
 406
 407                         wfDebug( __METHOD__ . ": running create updates\n" );
 408                         $article->createUpdates( $revision );
 409
 410                 } elseif( $changed ) {
 411                         wfDebug( __METHOD__ . ": running onArticleEdit\n" );
 412                         Article::onArticleEdit( $this->title );
 413
 414                         wfDebug( __METHOD__ . ": running edit updates\n" );
 415                         $article->editUpdates(
 416                                 $this->getText(),
 417                                 $this->getComment(),
 418                                 $this->minor,
 419                                 $this->timestamp,
 420                                 $revId );
 421                 }
 422
 423                 return true;
 424         }
 425
 426         function importUpload() {
 427                 wfDebug( __METHOD__ . ": STUB\n" );
 428
 429                 /**
 430                         // from file revert...
 431                         $source = $this->file->getArchiveVirtualUrl( $this->oldimage );
 432                         $comment = $wgRequest->getText( 'wpComment' );
 433                         // TODO: Preserve file properties from database instead of reloading from file
 434                         $status = $this->file->upload( $source, $comment, $comment );
 435                         if( $status->isGood() ) {
 436                 */
 437
 438                 /**
 439                         // from file upload...
 440                 $this->mLocalFile = wfLocalFile( $nt );
 441                 $this->mDestName = $this->mLocalFile->getName();
 442                 //....
 443                         $status = $this->mLocalFile->upload( $this->mTempPath, $this->mComment, $pageText,
 444                         File::DELETE_SOURCE, $this->mFileProps );
 445                         if ( !$status->isGood() ) {
 446                                 $resultDetails = array( 'internal' => $status->getWikiText() );
 447                 */
 448
 449                 // @fixme upload() uses $wgUser, which is wrong here
 450                 // it may also create a page without our desire, also wrong potentially.
 451                 // and, it will record a *current* upload, but we might want an archive version here
 452
 453                 $file = wfLocalFile( $this->getTitle() );
 454                 if( !$file ) {
 455                         var_dump( $file );
 456                         wfDebug( "IMPORT: Bad file. :(\n" );
 457                         return false;
 458                 }
 459
 460                 $source = $this->downloadSource();
 461                 if( !$source ) {
 462                         wfDebug( "IMPORT: Could not fetch remote file. :(\n" );
 463                         return false;
 464                 }
 465
 466                 $status = $file->upload( $source,
 467                         $this->getComment(),
 468                         $this->getComment(), // Initial page, if none present...
 469                         File::DELETE_SOURCE,
 470                         false, // props...
 471                         $this->getTimestamp() );
 472
 473                 if( $status->isGood() ) {
 474                         // yay?
 475                         wfDebug( "IMPORT: is ok?\n" );
 476                         return true;
 477                 }
 478
 479                 wfDebug( "IMPORT: is bad? " . $status->getXml() . "\n" );
 480                 return false;
 481
 482         }
 483
 484         function downloadSource() {
 485                 global $wgEnableUploads;
 486                 if( !$wgEnableUploads ) {
 487                         return false;
 488                 }
 489
 490                 $tempo = tempnam( wfTempDir(), 'download' );
 491                 $f = fopen( $tempo, 'wb' );
 492                 if( !$f ) {
 493                         wfDebug( "IMPORT: couldn't write to temp file $tempo\n" );
 494                         return false;
 495                 }
 496
 497                 // @fixme!
 498                 $src = $this->getSrc();
 499                 $data = Http::get( $src );
 500                 if( !$data ) {
 501                         wfDebug( "IMPORT: couldn't fetch source $src\n" );
 502                         fclose( $f );
 503                         unlink( $tempo );
 504                         return false;
 505                 }
 506
 507                 fwrite( $f, $data );
 508                 fclose( $f );
 509
 510                 return $tempo;
 511         }
 512
 513 }
 514
 515 /**
 516  * implements Special:Import
 517  * @addtogroup SpecialPage
 518  */
 519 class WikiImporter {
 520         var $mDebug = false;
 521         var $mSource = null;
 522         var $mPageCallback = null;
 523         var $mPageOutCallback = null;
 524         var $mRevisionCallback = null;
 525         var $mUploadCallback = null;
 526         var $mTargetNamespace = null;
 527         var $lastfield;
 528         var $tagStack = array();
 529
 530         function __construct( $source ) {
 531                 $this->setRevisionCallback( array( $this, "importRevision" ) );
 532                 $this->setUploadCallback( array( $this, "importUpload" ) );
 533                 $this->mSource = $source;
 534         }
 535
 536         function throwXmlError( $err ) {
 537                 $this->debug( "FAILURE: $err" );
 538                 wfDebug( "WikiImporter XML error: $err\n" );
 539         }
 540
 541         # --------------
 542
 543         function doImport() {
 544                 if( empty( $this->mSource ) ) {
 545                         return new WikiErrorMsg( "importnotext" );
 546                 }
 547
 548                 $parser = xml_parser_create( "UTF-8" );
 549
 550                 # case folding violates XML standard, turn it off
 551                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 552
 553                 xml_set_object( $parser, $this );
 554                 xml_set_element_handler( $parser, "in_start", "" );
 555
 556                 $offset = 0; // for context extraction on error reporting
 557                 do {
 558                         $chunk = $this->mSource->readChunk();
 559                         if( !xml_parse( $parser, $chunk, $this->mSource->atEnd() ) ) {
 560                                 wfDebug( "WikiImporter::doImport encountered XML parsing error\n" );
 561                                 return new WikiXmlError( $parser, wfMsgHtml( 'import-parse-failure' ), $chunk, $offset );
 562                         }
 563                         $offset += strlen( $chunk );
 564                 } while( $chunk !== false && !$this->mSource->atEnd() );
 565                 xml_parser_free( $parser );
 566
 567                 return true;
 568         }
 569
 570         function debug( $data ) {
 571                 if( $this->mDebug ) {
 572                         wfDebug( "IMPORT: $data\n" );
 573                 }
 574         }
 575
 576         function notice( $data ) {
 577                 global $wgCommandLineMode;
 578                 if( $wgCommandLineMode ) {
 579                         print "$data\n";
 580                 } else {
 581                         global $wgOut;
 582                         $wgOut->addHTML( "<li>" . htmlspecialchars( $data ) . "</li>\n" );
 583                 }
 584         }
 585
 586         /**
 587          * Set debug mode...
 588          */
 589         function setDebug( $debug ) {
 590                 $this->mDebug = $debug;
 591         }
 592
 593         /**
 594          * Sets the action to perform as each new page in the stream is reached.
 595          * @param callable $callback
 596          * @return callable
 597          */
 598         function setPageCallback( $callback ) {
 599                 $previous = $this->mPageCallback;
 600                 $this->mPageCallback = $callback;
 601                 return $previous;
 602         }
 603
 604         /**
 605          * Sets the action to perform as each page in the stream is completed.
 606          * Callback accepts the page title (as a Title object), a second object
 607          * with the original title form (in case it's been overridden into a
 608          * local namespace), and a count of revisions.
 609          *
 610          * @param callable $callback
 611          * @return callable
 612          */
 613         function setPageOutCallback( $callback ) {
 614                 $previous = $this->mPageOutCallback;
 615                 $this->mPageOutCallback = $callback;
 616                 return $previous;
 617         }
 618
 619         /**
 620          * Sets the action to perform as each page revision is reached.
 621          * @param callable $callback
 622          * @return callable
 623          */
 624         function setRevisionCallback( $callback ) {
 625                 $previous = $this->mRevisionCallback;
 626                 $this->mRevisionCallback = $callback;
 627                 return $previous;
 628         }
 629
 630         /**
 631          * Sets the action to perform as each file upload version is reached.
 632          * @param callable callback
 633          * @return callable
 634          */
 635         function setUploadCallback( $callback ) {
 636                 $previous = $this->mUploadCallback;
 637                 $this->mUploadCallback = $callback;
 638                 return $previous;
 639         }
 640
 641         /**
 642          * Set a target namespace to override the defaults
 643          */
 644         function setTargetNamespace( $namespace ) {
 645                 if( is_null( $namespace ) ) {
 646                         // Don't override namespaces
 647                         $this->mTargetNamespace = null;
 648                 } elseif( $namespace >= 0 ) {
 649                         // FIXME: Check for validity
 650                         $this->mTargetNamespace = intval( $namespace );
 651                 } else {
 652                         return false;
 653                 }
 654         }
 655
 656         /**
 657          * Default per-revision callback, performs the import.
 658          * @param WikiRevision $revision
 659          * @private
 660          */
 661         function importRevision( $revision ) {
 662                 $dbw = wfGetDB( DB_MASTER );
 663                 return $dbw->deadlockLoop( array( $revision, 'importOldRevision' ) );
 664         }
 665
 666         /**
 667          * Dummy for now...
 668          */
 669         function importUpload( $revision ) {
 670                 //$dbw = wfGetDB( DB_MASTER );
 671                 //return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
 672                 return false;
 673         }
 674
 675         /**
 676          * Alternate per-revision callback, for debugging.
 677          * @param WikiRevision $revision
 678          * @private
 679          */
 680         function debugRevisionHandler( &$revision ) {
 681                 $this->debug( "Got revision:" );
 682                 if( is_object( $revision->title ) ) {
 683                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 684                 } else {
 685                         $this->debug( "-- Title: <invalid>" );
 686                 }
 687                 $this->debug( "-- User: " . $revision->user_text );
 688                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 689                 $this->debug( "-- Comment: " . $revision->comment );
 690                 $this->debug( "-- Text: " . $revision->text );
 691         }
 692
 693         /**
 694          * Notify the callback function when a new <page> is reached.
 695          * @param Title $title
 696          * @private
 697          */
 698         function pageCallback( $title ) {
 699                 if( is_callable( $this->mPageCallback ) ) {
 700                         call_user_func( $this->mPageCallback, $title );
 701                 }
 702         }
 703
 704         /**
 705          * Notify the callback function when a </page> is closed.
 706          * @param Title $title
 707          * @param Title $origTitle
 708          * @param int $revisionCount
 709          * @param int $successCount number of revisions for which callback returned true
 710          * @private
 711          */
 712         function pageOutCallback( $title, $origTitle, $revisionCount, $successCount ) {
 713                 if( is_callable( $this->mPageOutCallback ) ) {
 714                         call_user_func( $this->mPageOutCallback, $title, $origTitle,
 715                                 $revisionCount, $successCount );
 716                 }
 717         }
 718
 719
 720         # XML parser callbacks from here out -- beware!
 721         function donothing( $parser, $x, $y="" ) {
 722                 #$this->debug( "donothing" );
 723         }
 724
 725         function in_start( $parser, $name, $attribs ) {
 726                 $this->debug( "in_start $name" );
 727                 if( $name != "mediawiki" ) {
 728                         return $this->throwXMLerror( "Expected <mediawiki>, got <$name>" );
 729                 }
 730                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 731         }
 732
 733         function in_mediawiki( $parser, $name, $attribs ) {
 734                 $this->debug( "in_mediawiki $name" );
 735                 if( $name == 'siteinfo' ) {
 736                         xml_set_element_handler( $parser, "in_siteinfo", "out_siteinfo" );
 737                 } elseif( $name == 'page' ) {
 738                         $this->push( $name );
 739                         $this->workRevisionCount = 0;
 740                         $this->workSuccessCount = 0;
 741                         $this->uploadCount = 0;
 742                         $this->uploadSuccessCount = 0;
 743                         xml_set_element_handler( $parser, "in_page", "out_page" );
 744                 } else {
 745                         return $this->throwXMLerror( "Expected <page>, got <$name>" );
 746                 }
 747         }
 748         function out_mediawiki( $parser, $name ) {
 749                 $this->debug( "out_mediawiki $name" );
 750                 if( $name != "mediawiki" ) {
 751                         return $this->throwXMLerror( "Expected </mediawiki>, got </$name>" );
 752                 }
 753                 xml_set_element_handler( $parser, "donothing", "donothing" );
 754         }
 755
 756
 757         function in_siteinfo( $parser, $name, $attribs ) {
 758                 // no-ops for now
 759                 $this->debug( "in_siteinfo $name" );
 760                 switch( $name ) {
 761                 case "sitename":
 762                 case "base":
 763                 case "generator":
 764                 case "case":
 765                 case "namespaces":
 766                 case "namespace":
 767                         break;
 768                 default:
 769                         return $this->throwXMLerror( "Element <$name> not allowed in <siteinfo>." );
 770                 }
 771         }
 772
 773         function out_siteinfo( $parser, $name ) {
 774                 if( $name == "siteinfo" ) {
 775                         xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 776                 }
 777         }
 778
 779
 780         function in_page( $parser, $name, $attribs ) {
 781                 $this->debug( "in_page $name" );
 782                 switch( $name ) {
 783                 case "id":
 784                 case "title":
 785                 case "restrictions":
 786                         $this->appendfield = $name;
 787                         $this->appenddata = "";
 788                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 789                         xml_set_character_data_handler( $parser, "char_append" );
 790                         break;
 791                 case "revision":
 792                         $this->push( "revision" );
 793                         if( is_object( $this->pageTitle ) ) {
 794                                 $this->workRevision = new WikiRevision;
 795                                 $this->workRevision->setTitle( $this->pageTitle );
 796                                 $this->workRevisionCount++;
 797                         } else {
 798                                 // Skipping items due to invalid page title
 799                                 $this->workRevision = null;
 800                         }
 801                         xml_set_element_handler( $parser, "in_revision", "out_revision" );
 802                         break;
 803                 case "upload":
 804                         $this->push( "upload" );
 805                         if( is_object( $this->pageTitle ) ) {
 806                                 $this->workRevision = new WikiRevision;
 807                                 $this->workRevision->setTitle( $this->pageTitle );
 808                                 $this->uploadCount++;
 809                         } else {
 810                                 // Skipping items due to invalid page title
 811                                 $this->workRevision = null;
 812                         }
 813                         xml_set_element_handler( $parser, "in_upload", "out_upload" );
 814                         break;
 815                 default:
 816                         return $this->throwXMLerror( "Element <$name> not allowed in a <page>." );
 817                 }
 818         }
 819
 820         function out_page( $parser, $name ) {
 821                 $this->debug( "out_page $name" );
 822                 $this->pop();
 823                 if( $name != "page" ) {
 824                         return $this->throwXMLerror( "Expected </page>, got </$name>" );
 825                 }
 826                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 827
 828                 $this->pageOutCallback( $this->pageTitle, $this->origTitle,
 829                         $this->workRevisionCount, $this->workSuccessCount );
 830
 831                 $this->workTitle = null;
 832                 $this->workRevision = null;
 833                 $this->workRevisionCount = 0;
 834                 $this->workSuccessCount = 0;
 835                 $this->pageTitle = null;
 836                 $this->origTitle = null;
 837         }
 838
 839         function in_nothing( $parser, $name, $attribs ) {
 840                 $this->debug( "in_nothing $name" );
 841                 return $this->throwXMLerror( "No child elements allowed here; got <$name>" );
 842         }
 843         function char_append( $parser, $data ) {
 844                 $this->debug( "char_append '$data'" );
 845                 $this->appenddata .= $data;
 846         }
 847         function out_append( $parser, $name ) {
 848                 $this->debug( "out_append $name" );
 849                 if( $name != $this->appendfield ) {
 850                         return $this->throwXMLerror( "Expected </{$this->appendfield}>, got </$name>" );
 851                 }
 852
 853                 switch( $this->appendfield ) {
 854                 case "title":
 855                         $this->workTitle = $this->appenddata;
 856                         $this->origTitle = Title::newFromText( $this->workTitle );
 857                         if( !is_null( $this->mTargetNamespace ) && !is_null( $this->origTitle ) ) {
 858                                 $this->pageTitle = Title::makeTitle( $this->mTargetNamespace,
 859                                         $this->origTitle->getDBkey() );
 860                         } else {
 861                                 $this->pageTitle = Title::newFromText( $this->workTitle );
 862                         }
 863                         if( is_null( $this->pageTitle ) ) {
 864                                 // Invalid page title? Ignore the page
 865                                 $this->notice( "Skipping invalid page title '$this->workTitle'" );
 866                         } else {
 867                                 $this->pageCallback( $this->workTitle );
 868                         }
 869                         break;
 870                 case "id":
 871                         if ( $this->parentTag() == 'revision' ) {
 872                                 if( $this->workRevision )
 873                                         $this->workRevision->setID( $this->appenddata );
 874                         }
 875                         break;
 876                 case "text":
 877                         if( $this->workRevision )
 878                                 $this->workRevision->setText( $this->appenddata );
 879                         break;
 880                 case "username":
 881                         if( $this->workRevision )
 882                                 $this->workRevision->setUsername( $this->appenddata );
 883                         break;
 884                 case "ip":
 885                         if( $this->workRevision )
 886                                 $this->workRevision->setUserIP( $this->appenddata );
 887                         break;
 888                 case "timestamp":
 889                         if( $this->workRevision )
 890                                 $this->workRevision->setTimestamp( $this->appenddata );
 891                         break;
 892                 case "comment":
 893                         if( $this->workRevision )
 894                                 $this->workRevision->setComment( $this->appenddata );
 895                         break;
 896                 case "minor":
 897                         if( $this->workRevision )
 898                                 $this->workRevision->setMinor( true );
 899                         break;
 900                 case "filename":
 901                         if( $this->workRevision )
 902                                 $this->workRevision->setFilename( $this->appenddata );
 903                         break;
 904                 case "src":
 905                         if( $this->workRevision )
 906                                 $this->workRevision->setSrc( $this->appenddata );
 907                         break;
 908                 case "size":
 909                         if( $this->workRevision )
 910                                 $this->workRevision->setSize( intval( $this->appenddata ) );
 911                         break;
 912                 default:
 913                         $this->debug( "Bad append: {$this->appendfield}" );
 914                 }
 915                 $this->appendfield = "";
 916                 $this->appenddata = "";
 917
 918                 $parent = $this->parentTag();
 919                 xml_set_element_handler( $parser, "in_$parent", "out_$parent" );
 920                 xml_set_character_data_handler( $parser, "donothing" );
 921         }
 922
 923         function in_revision( $parser, $name, $attribs ) {
 924                 $this->debug( "in_revision $name" );
 925                 switch( $name ) {
 926                 case "id":
 927                 case "timestamp":
 928                 case "comment":
 929                 case "minor":
 930                 case "text":
 931                         $this->appendfield = $name;
 932                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 933                         xml_set_character_data_handler( $parser, "char_append" );
 934                         break;
 935                 case "contributor":
 936                         $this->push( "contributor" );
 937                         xml_set_element_handler( $parser, "in_contributor", "out_contributor" );
 938                         break;
 939                 default:
 940                         return $this->throwXMLerror( "Element <$name> not allowed in a <revision>." );
 941                 }
 942         }
 943
 944         function out_revision( $parser, $name ) {
 945                 $this->debug( "out_revision $name" );
 946                 $this->pop();
 947                 if( $name != "revision" ) {
 948                         return $this->throwXMLerror( "Expected </revision>, got </$name>" );
 949                 }
 950                 xml_set_element_handler( $parser, "in_page", "out_page" );
 951
 952                 if( $this->workRevision ) {
 953                         $ok = call_user_func_array( $this->mRevisionCallback,
 954                                 array( $this->workRevision, $this ) );
 955                         if( $ok ) {
 956                                 $this->workSuccessCount++;
 957                         }
 958                 }
 959         }
 960
 961         function in_upload( $parser, $name, $attribs ) {
 962                 $this->debug( "in_upload $name" );
 963                 switch( $name ) {
 964                 case "timestamp":
 965                 case "comment":
 966                 case "text":
 967                 case "filename":
 968                 case "src":
 969                 case "size":
 970                         $this->appendfield = $name;
 971                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 972                         xml_set_character_data_handler( $parser, "char_append" );
 973                         break;
 974                 case "contributor":
 975                         $this->push( "contributor" );
 976                         xml_set_element_handler( $parser, "in_contributor", "out_contributor" );
 977                         break;
 978                 default:
 979                         return $this->throwXMLerror( "Element <$name> not allowed in an <upload>." );
 980                 }
 981         }
 982
 983         function out_upload( $parser, $name ) {
 984                 $this->debug( "out_revision $name" );
 985                 $this->pop();
 986                 if( $name != "upload" ) {
 987                         return $this->throwXMLerror( "Expected </upload>, got </$name>" );
 988                 }
 989                 xml_set_element_handler( $parser, "in_page", "out_page" );
 990
 991                 if( $this->workRevision ) {
 992                         $ok = call_user_func_array( $this->mUploadCallback,
 993                                 array( $this->workRevision, $this ) );
 994                         if( $ok ) {
 995                                 $this->workUploadSuccessCount++;
 996                         }
 997                 }
 998         }
 999
1000         function in_contributor( $parser, $name, $attribs ) {
1001                 $this->debug( "in_contributor $name" );
1002                 switch( $name ) {
1003                 case "username":
1004                 case "ip":
1005                 case "id":
1006                         $this->appendfield = $name;
1007                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
1008                         xml_set_character_data_handler( $parser, "char_append" );
1009                         break;
1010                 default:
1011                         $this->throwXMLerror( "Invalid tag <$name> in <contributor>" );
1012                 }
1013         }
1014
1015         function out_contributor( $parser, $name ) {
1016                 $this->debug( "out_contributor $name" );
1017                 $this->pop();
1018                 if( $name != "contributor" ) {
1019                         return $this->throwXMLerror( "Expected </contributor>, got </$name>" );
1020                 }
1021                 $parent = $this->parentTag();
1022                 xml_set_element_handler( $parser, "in_$parent", "out_$parent" );
1023         }
1024
1025         private function push( $name ) {
1026                 array_push( $this->tagStack, $name );
1027                 $this->debug( "PUSH $name" );
1028         }
1029
1030         private function pop() {
1031                 $name = array_pop( $this->tagStack );
1032                 $this->debug( "POP $name" );
1033                 return $name;
1034         }
1035
1036         private function parentTag() {
1037                 $name = $this->tagStack[count( $this->tagStack ) - 1];
1038                 $this->debug( "PARENT $name" );
1039                 return $name;
1040         }
1041
1042 }
1043
1044 /**
1045  * @todo document (e.g. one-sentence class description).
1046  * @addtogroup SpecialPage
1047  */
1048 class ImportStringSource {
1049         function __construct( $string ) {
1050                 $this->mString = $string;
1051                 $this->mRead = false;
1052         }
1053
1054         function atEnd() {
1055                 return $this->mRead;
1056         }
1057
1058         function readChunk() {
1059                 if( $this->atEnd() ) {
1060                         return false;
1061                 } else {
1062                         $this->mRead = true;
1063                         return $this->mString;
1064                 }
1065         }
1066 }
1067
1068 /**
1069  * @todo document (e.g. one-sentence class description).
1070  * @addtogroup SpecialPage
1071  */
1072 class ImportStreamSource {
1073         function __construct( $handle ) {
1074                 $this->mHandle = $handle;
1075         }
1076
1077         function atEnd() {
1078                 return feof( $this->mHandle );
1079         }
1080
1081         function readChunk() {
1082                 return fread( $this->mHandle, 32768 );
1083         }
1084
1085         static function newFromFile( $filename ) {
1086                 $file = @fopen( $filename, 'rt' );
1087                 if( !$file ) {
1088                         return new WikiErrorMsg( "importcantopen" );
1089                 }
1090                 return new ImportStreamSource( $file );
1091         }
1092
1093         static function newFromUpload( $fieldname = "xmlimport" ) {
1094                 $upload =& $_FILES[$fieldname];
1095
1096                 if( !isset( $upload ) || !$upload['name'] ) {
1097                         return new WikiErrorMsg( 'importnofile' );
1098                 }
1099                 if( !empty( $upload['error'] ) ) {
1100                         switch($upload['error']){
1101                                 case 1: # The uploaded file exceeds the upload_max_filesize directive in php.ini.
1102                                         return new WikiErrorMsg( 'importuploaderrorsize' );
1103                                 case 2: # The uploaded file exceeds the MAX_FILE_SIZE directive that was specified in the HTML form.
1104                                         return new WikiErrorMsg( 'importuploaderrorsize' );
1105                                 case 3: # The uploaded file was only partially uploaded
1106                                         return new WikiErrorMsg( 'importuploaderrorpartial' );
1107                             case 6: #Missing a temporary folder. Introduced in PHP 4.3.10 and PHP 5.0.3.
1108                                 return new WikiErrorMsg( 'importuploaderrortemp' );
1109                             # case else: # Currently impossible
1110                         }
1111
1112                 }
1113                 $fname = $upload['tmp_name'];
1114                 if( is_uploaded_file( $fname ) ) {
1115                         return ImportStreamSource::newFromFile( $fname );
1116                 } else {
1117                         return new WikiErrorMsg( 'importnofile' );
1118                 }
1119         }
1120
1121         function newFromURL( $url, $method = 'GET' ) {
1122                 wfDebug( __METHOD__ . ": opening $url\n" );
1123                 # Use the standard HTTP fetch function; it times out
1124                 # quicker and sorts out user-agent problems which might
1125                 # otherwise prevent importing from large sites, such
1126                 # as the Wikimedia cluster, etc.
1127                 $data = Http::request( $method, $url );
1128                 if( $data !== false ) {
1129                         $file = tmpfile();
1130                         fwrite( $file, $data );
1131                         fflush( $file );
1132                         fseek( $file, 0 );
1133                         return new ImportStreamSource( $file );
1134                 } else {
1135                         return new WikiErrorMsg( 'importcantopen' );
1136                 }
1137         }
1138
1139         public static function newFromInterwiki( $interwiki, $page, $history=false ) {
1140                 if( $page == '' ) {
1141                         return new WikiErrorMsg( 'import-noarticle' );
1142                 }
1143                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
1144                 if( is_null( $link ) || $link->getInterwiki() == '' ) {
1145                         return new WikiErrorMsg( 'importbadinterwiki' );
1146                 } else {
1147                         $params = $history ? 'history=1' : '';
1148                         $url = $link->getFullUrl( $params );
1149                         # For interwikis, use POST to avoid redirects.
1150                         return ImportStreamSource::newFromURL( $url, "POST" );
1151                 }
1152         }
1153 }