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         $articleName = '';
  34         $namespace = $wgImportTargetNamespace;
  35         $frompage = '';
  36         $history = true;
  37
  38         if( $wgRequest->wasPosted() && $wgRequest->getVal( 'action' ) == 'submit') {
  39                 $isUpload = false;
  40                 $articleName = htmlspecialchars( $wgRequest->getText( 'articlename' ) );
  41                 $namespace = $wgRequest->getIntOrNull( 'namespace' );
  42
  43                 switch( $wgRequest->getVal( "source" ) ) {
  44                 case "upload":
  45                         $isUpload = true;
  46                         if( $wgUser->isAllowed( 'importupload' ) ) {
  47                                 $source = ImportStreamSource::newFromUpload( "xmlimport" );
  48                         } else {
  49                                 return $wgOut->permissionRequired( 'importupload' );
  50                         }
  51                         break;
  52                 case "interwiki":
  53                         $interwiki = $wgRequest->getVal( 'interwiki' );
  54                         $history = $wgRequest->getCheck( 'interwikiHistory' );
  55                         $frompage = $wgRequest->getText( "frompage" );
  56                         $source = ImportStreamSource::newFromInterwiki(
  57                                 $interwiki,
  58                                 $frompage,
  59                                 $history );
  60                         break;
  61                 default:
  62                         $source = new WikiErrorMsg( "importunknownsource" );
  63                 }
  64
  65                 if( WikiError::isError( $source ) ) {
  66                         $wgOut->addWikiText( '<p class="error">' . wfMsg( "importfailed", wfEscapeWikiText( $source->getMessage() ) ) . '</p>' );
  67                 } else {
  68                         $wgOut->addWikiText( wfMsg( "importstart" ) );
  69
  70                         $importer = new WikiImporter( $source );
  71                         if( !is_null( $namespace ) ) {
  72                                 $importer->setTargetNamespace( $namespace );
  73                         }
  74                         if( !is_null( $articleName ) ) {
  75                                 $importer->setTargetArticleName( $articleName );
  76                         }
  77                         $reporter = new ImportReporter( $importer, $isUpload, $interwiki );
  78
  79                         $reporter->open();
  80                         $result = $importer->doImport();
  81                         $resultCount = $reporter->close();
  82
  83                         if( WikiError::isError( $result ) ) {
  84                                 # No source or XML parse error
  85                                 $wgOut->addWikiText( '<p class="error">' . wfMsg( "importfailed", wfEscapeWikiText( $result->getMessage() ) ) . '</p>' );
  86                         } elseif( WikiError::isError( $resultCount ) ) {
  87                                 # Zero revisions
  88                                 $wgOut->addWikiText( '<p class="error">' . wfMsg( "importfailed", wfEscapeWikiText( $resultCount->getMessage() ) ) . '</p>' );
  89                         } else {
  90                                 # Success!
  91                                 $wgOut->addWikiText( wfMsg( "importsuccess" ) );
  92                         }
  93                         $wgOut->addWikiText( '<hr />' );
  94                 }
  95         }
  96
  97         $action = $wgTitle->escapeLocalUrl( 'action=submit' );
  98
  99         if( $wgUser->isAllowed( 'importupload' ) ) {
 100                 $wgOut->addWikiText( wfMsg( "importtext" ) );
 101                 $wgOut->addHTML(
 102                         Xml::openElement( 'fieldset' ).
 103                         Xml::element( 'legend', null, wfMsg( 'upload' ) ) .
 104                         Xml::openElement( 'form', array( 'enctype' => 'multipart/form-data', 'method' => 'post', 'action' => $action ) ) .
 105                         Xml::hidden( 'action', 'submit' ) .
 106                         Xml::hidden( 'source', 'upload' ) .
 107                         Xml::hidden( 'MAX_FILE_SIZE', '2000000' ) .
 108                         "<input type='file' name='xmlimport' value='' size='30' />" . // No Xml function for type=file? Todo?
 109                         Xml::submitButton( wfMsg( 'uploadbtn' ) ) .
 110                         Xml::closeElement( 'form' ) .
 111                         Xml::closeElement( 'fieldset' ) );
 112         } else {
 113                 if( empty( $wgImportSources ) ) {
 114                         $wgOut->addWikiText( wfMsg( 'importnosources' ) );
 115                 }
 116         }
 117
 118         if( !empty( $wgImportSources ) ) {
 119                 $wgOut->addHTML(
 120                         Xml::openElement( 'fieldset' ) .
 121                         Xml::element( 'legend', null, wfMsg( 'importinterwiki' ) ) .
 122                         Xml::openElement( 'form', array( 'method' => 'post', 'action' => $action ) ) .
 123                         wfMsgExt( 'import-interwiki-text', array( 'parse' ) ) .
 124                         Xml::hidden( 'action', 'submit' ) .
 125                         Xml::hidden( 'source', 'interwiki' ) .
 126                         Xml::openElement( 'table' ) .
 127                         "<tr>
 128                                 <td>" .
 129                                         Xml::openElement( 'select', array( 'name' => 'interwiki' ) ) );
 130                 foreach( $wgImportSources as $prefix ) {
 131                         $iw = htmlspecialchars( $prefix );
 132                         $selected = ($interwiki === $prefix) ? ' selected="selected"' : '';
 133                         $wgOut->addHTML( Xml::option( $iw, $iw, $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></td>
 144                                 <td>" .
 145                                         Xml::checkLabel( wfMsg( 'import-interwiki-history' ), 'interwikiHistory', 'interwikiHistory', $history ) .
 146                                 "</td>
 147                         </tr>
 148                         <tr>
 149                                 <td></td>
 150                                 <td>" .
 151                                         Xml::inputLabel( wfMsg( 'import-articlename' ), 'articlename', 'articlename', 50, $articleName ) .
 152                                 "</td>
 153                         </tr>
 154                         <tr>
 155                                 <td></td>
 156                                 <td>" .
 157                                         Xml::label( wfMsg( 'import-interwiki-namespace' ), 'namespace' ) .
 158                                         Xml::namespaceSelector( $namespace, '' ) .
 159                                 "</td>
 160                         </tr>
 161                         <tr>
 162                                 <td></td>
 163                                 <td>" .
 164                                         Xml::submitButton( wfMsg( 'import-interwiki-submit' ) ) .
 165                                 "</td>
 166                         </tr>" .
 167                         Xml::closeElement( 'table' ).
 168                         Xml::closeElement( 'form' ) .
 169                         Xml::closeElement( 'fieldset' ) );
 170         }
 171 }
 172
 173 /**
 174  * Reporting callback
 175  * @addtogroup SpecialPage
 176  */
 177 class ImportReporter {
 178         function __construct( $importer, $upload, $interwiki ) {
 179                 $importer->setPageOutCallback( array( $this, 'reportPage' ) );
 180                 $this->mPageCount = 0;
 181                 $this->mIsUpload = $upload;
 182                 $this->mInterwiki = $interwiki;
 183         }
 184
 185         function open() {
 186                 global $wgOut;
 187                 $wgOut->addHtml( "<ul>\n" );
 188         }
 189
 190         function reportPage( $title, $origTitle, $revisionCount, $successCount ) {
 191                 global $wgOut, $wgUser, $wgLang, $wgContLang;
 192
 193                 $skin = $wgUser->getSkin();
 194
 195                 $this->mPageCount++;
 196
 197                 $localCount = $wgLang->formatNum( $successCount );
 198                 $contentCount = $wgContLang->formatNum( $successCount );
 199
 200                 if( $successCount > 0 ) {
 201                         $wgOut->addHtml( "<li>" . $skin->makeKnownLinkObj( $title ) .
 202                                 " " .
 203                                 wfMsgExt( 'import-revision-count', array( 'parsemag', 'escape' ), $localCount ) .
 204                                 "</li>\n" );
 205
 206                         $log = new LogPage( 'import' );
 207                         if( $this->mIsUpload ) {
 208                                 $detail = wfMsgForContent( 'import-logentry-upload-detail',
 209                                         $contentCount );
 210                                 $log->addEntry( 'upload', $title, $detail );
 211                         } else {
 212                                 $interwiki = '[[:' . $this->mInterwiki . ':' .
 213                                         $origTitle->getPrefixedText() . ']]';
 214                                 $detail = wfMsgForContent( 'import-logentry-interwiki-detail',
 215                                         $contentCount, $interwiki );
 216                                 $log->addEntry( 'interwiki', $title, $detail );
 217                         }
 218
 219                         $comment = $detail; // quick
 220                         $dbw = wfGetDB( DB_MASTER );
 221                         $nullRevision = Revision::newNullRevision(
 222                                 $dbw, $title->getArticleId(), $comment, true );
 223                         $nullRevision->insertOn( $dbw );
 224                 } else {
 225                         $wgOut->addHtml( '<li>' . wfMsgExt( 'import-nonewrevisions', array( 'parse', ) ) . '</li>' );
 226                 }
 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 getTitle() {
 296                 return $this->title;
 297         }
 298
 299         function getID() {
 300                 return $this->id;
 301         }
 302
 303         function getTimestamp() {
 304                 return $this->timestamp;
 305         }
 306
 307         function getUser() {
 308                 return $this->user_text;
 309         }
 310
 311         function getText() {
 312                 return $this->text;
 313         }
 314
 315         function getComment() {
 316                 return $this->comment;
 317         }
 318
 319         function getMinor() {
 320                 return $this->minor;
 321         }
 322
 323         function importOldRevision() {
 324                 $dbw = wfGetDB( DB_MASTER );
 325
 326                 # Sneak a single revision into place
 327                 $user = User::newFromName( $this->getUser() );
 328                 if( $user ) {
 329                         $userId = intval( $user->getId() );
 330                         $userText = $user->getName();
 331                 } else {
 332                         $userId = 0;
 333                         $userText = $this->getUser();
 334                 }
 335
 336                 // avoid memory leak...?
 337                 $linkCache =& LinkCache::singleton();
 338                 $linkCache->clear();
 339
 340                 $article = new Article( $this->title );
 341                 $pageId = $article->getId();
 342                 if( $pageId == 0 ) {
 343                         # must create the page...
 344                         $pageId = $article->insertOn( $dbw );
 345                         $created = true;
 346                 } else {
 347                         $created = false;
 348
 349                         $prior = Revision::loadFromTimestamp( $dbw, $this->title, $this->timestamp );
 350                         if( !is_null( $prior ) ) {
 351                                 // FIXME: this could fail slightly for multiple matches :P
 352                                 wfDebug( __METHOD__ . ": skipping existing revision for [[" .
 353                                         $this->title->getPrefixedText() . "]], timestamp " .
 354                                         $this->timestamp . "\n" );
 355                                 return false;
 356                         }
 357                 }
 358
 359                 # FIXME: Use original rev_id optionally
 360                 # FIXME: blah blah blah
 361
 362                 #if( $numrows > 0 ) {
 363                 #       return wfMsg( "importhistoryconflict" );
 364                 #}
 365
 366                 # Insert the row
 367                 $revision = new Revision( array(
 368                         'page'       => $pageId,
 369                         'text'       => $this->getText(),
 370                         'comment'    => $this->getComment(),
 371                         'user'       => $userId,
 372                         'user_text'  => $userText,
 373                         'timestamp'  => $this->timestamp,
 374                         'minor_edit' => $this->minor,
 375                         ) );
 376                 $revId = $revision->insertOn( $dbw );
 377                 $changed = $article->updateIfNewerOn( $dbw, $revision );
 378
 379                 if( $created ) {
 380                         wfDebug( __METHOD__ . ": running onArticleCreate\n" );
 381                         Article::onArticleCreate( $this->title );
 382
 383                         wfDebug( __METHOD__ . ": running create updates\n" );
 384                         $article->createUpdates( $revision );
 385
 386                 } elseif( $changed ) {
 387                         wfDebug( __METHOD__ . ": running onArticleEdit\n" );
 388                         Article::onArticleEdit( $this->title );
 389
 390                         wfDebug( __METHOD__ . ": running edit updates\n" );
 391                         $article->editUpdates(
 392                                 $this->getText(),
 393                                 $this->getComment(),
 394                                 $this->minor,
 395                                 $this->timestamp,
 396                                 $revId );
 397                 }
 398
 399                 return true;
 400         }
 401
 402 }
 403
 404 /**
 405  * implements Special:Import
 406  * @addtogroup SpecialPage
 407  */
 408 class WikiImporter {
 409         var $mSource = null;
 410         var $mPageCallback = null;
 411         var $mPageOutCallback = null;
 412         var $mRevisionCallback = null;
 413         var $mTargetNamespace = null;
 414         var $mTargetArticleName = null;
 415         var $lastfield;
 416
 417         function WikiImporter( $source ) {
 418                 $this->setRevisionCallback( array( &$this, "importRevision" ) );
 419                 $this->mSource = $source;
 420         }
 421
 422         function throwXmlError( $err ) {
 423                 $this->debug( "FAILURE: $err" );
 424                 wfDebug( "WikiImporter XML error: $err\n" );
 425         }
 426
 427         # --------------
 428
 429         function doImport() {
 430                 if( empty( $this->mSource ) ) {
 431                         return new WikiErrorMsg( "importnotext" );
 432                 }
 433
 434                 $parser = xml_parser_create( "UTF-8" );
 435
 436                 # case folding violates XML standard, turn it off
 437                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 438
 439                 xml_set_object( $parser, $this );
 440                 xml_set_element_handler( $parser, "in_start", "" );
 441
 442                 $offset = 0; // for context extraction on error reporting
 443                 do {
 444                         $chunk = $this->mSource->readChunk();
 445                         if( !xml_parse( $parser, $chunk, $this->mSource->atEnd() ) ) {
 446                                 wfDebug( "WikiImporter::doImport encountered XML parsing error\n" );
 447                                 return new WikiXmlError( $parser, wfMsgHtml( 'import-parse-failure' ), $chunk, $offset );
 448                         }
 449                         $offset += strlen( $chunk );
 450                 } while( $chunk !== false && !$this->mSource->atEnd() );
 451                 xml_parser_free( $parser );
 452
 453                 return true;
 454         }
 455
 456         function debug( $data ) {
 457                 #wfDebug( "IMPORT: $data\n" );
 458         }
 459
 460         function notice( $data ) {
 461                 global $wgCommandLineMode;
 462                 if( $wgCommandLineMode ) {
 463                         print "$data\n";
 464                 } else {
 465                         global $wgOut;
 466                         $wgOut->addHTML( "<li>" . htmlspecialchars( $data ) . "</li>\n" );
 467                 }
 468         }
 469
 470         /**
 471          * Sets the action to perform as each new page in the stream is reached.
 472          * @param callable $callback
 473          * @return callable
 474          */
 475         function setPageCallback( $callback ) {
 476                 $previous = $this->mPageCallback;
 477                 $this->mPageCallback = $callback;
 478                 return $previous;
 479         }
 480
 481         /**
 482          * Sets the action to perform as each page in the stream is completed.
 483          * Callback accepts the page title (as a Title object), a second object
 484          * with the original title form (in case it's been overridden into a
 485          * local namespace), and a count of revisions.
 486          *
 487          * @param callable $callback
 488          * @return callable
 489          */
 490         function setPageOutCallback( $callback ) {
 491                 $previous = $this->mPageOutCallback;
 492                 $this->mPageOutCallback = $callback;
 493                 return $previous;
 494         }
 495
 496         /**
 497          * Sets the action to perform as each page revision is reached.
 498          * @param callable $callback
 499          * @return callable
 500          */
 501         function setRevisionCallback( $callback ) {
 502                 $previous = $this->mRevisionCallback;
 503                 $this->mRevisionCallback = $callback;
 504                 return $previous;
 505         }
 506
 507         /**
 508          * Set a target namespace to override the defaults
 509          */
 510         function setTargetNamespace( $namespace ) {
 511                 if( is_null( $namespace ) ) {
 512                         // Don't override namespaces
 513                         $this->mTargetNamespace = null;
 514                 } elseif( $namespace >= 0 ) {
 515                         // FIXME: Check for validity
 516                         $this->mTargetNamespace = intval( $namespace );
 517                 } else {
 518                         return false;
 519                 }
 520         }
 521
 522         /**
 523          * Set a target articlename to override the defaults
 524          */
 525         function setTargetArticleName( $articleName ) {
 526                 $this->mTargetArticleName = is_null( $articleName ) ? null : $articleName;
 527         }
 528
 529         /**
 530          * Default per-revision callback, performs the import.
 531          * @param WikiRevision $revision
 532          * @private
 533          */
 534         function importRevision( &$revision ) {
 535                 $dbw = wfGetDB( DB_MASTER );
 536                 return $dbw->deadlockLoop( array( &$revision, 'importOldRevision' ) );
 537         }
 538
 539         /**
 540          * Alternate per-revision callback, for debugging.
 541          * @param WikiRevision $revision
 542          * @private
 543          */
 544         function debugRevisionHandler( &$revision ) {
 545                 $this->debug( "Got revision:" );
 546                 if( is_object( $revision->title ) ) {
 547                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 548                 } else {
 549                         $this->debug( "-- Title: <invalid>" );
 550                 }
 551                 $this->debug( "-- User: " . $revision->user_text );
 552                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 553                 $this->debug( "-- Comment: " . $revision->comment );
 554                 $this->debug( "-- Text: " . $revision->text );
 555         }
 556
 557         /**
 558          * Notify the callback function when a new <page> is reached.
 559          * @param Title $title
 560          * @private
 561          */
 562         function pageCallback( $title ) {
 563                 if( is_callable( $this->mPageCallback ) ) {
 564                         call_user_func( $this->mPageCallback, $title );
 565                 }
 566         }
 567
 568         /**
 569          * Notify the callback function when a </page> is closed.
 570          * @param Title $title
 571          * @param Title $origTitle
 572          * @param int $revisionCount
 573          * @param int $successCount number of revisions for which callback returned true
 574          * @private
 575          */
 576         function pageOutCallback( $title, $origTitle, $revisionCount, $successCount ) {
 577                 if( is_callable( $this->mPageOutCallback ) ) {
 578                         call_user_func( $this->mPageOutCallback, $title, $origTitle,
 579                                 $revisionCount, $successCount );
 580                 }
 581         }
 582
 583
 584         # XML parser callbacks from here out -- beware!
 585         function donothing( $parser, $x, $y="" ) {
 586                 #$this->debug( "donothing" );
 587         }
 588
 589         function in_start( $parser, $name, $attribs ) {
 590                 $this->debug( "in_start $name" );
 591                 if( $name != "mediawiki" ) {
 592                         return $this->throwXMLerror( "Expected <mediawiki>, got <$name>" );
 593                 }
 594                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 595         }
 596
 597         function in_mediawiki( $parser, $name, $attribs ) {
 598                 $this->debug( "in_mediawiki $name" );
 599                 if( $name == 'siteinfo' ) {
 600                         xml_set_element_handler( $parser, "in_siteinfo", "out_siteinfo" );
 601                 } elseif( $name == 'page' ) {
 602                         $this->workRevisionCount = 0;
 603                         $this->workSuccessCount = 0;
 604                         xml_set_element_handler( $parser, "in_page", "out_page" );
 605                 } else {
 606                         return $this->throwXMLerror( "Expected <page>, got <$name>" );
 607                 }
 608         }
 609         function out_mediawiki( $parser, $name ) {
 610                 $this->debug( "out_mediawiki $name" );
 611                 if( $name != "mediawiki" ) {
 612                         return $this->throwXMLerror( "Expected </mediawiki>, got </$name>" );
 613                 }
 614                 xml_set_element_handler( $parser, "donothing", "donothing" );
 615         }
 616
 617
 618         function in_siteinfo( $parser, $name, $attribs ) {
 619                 // no-ops for now
 620                 $this->debug( "in_siteinfo $name" );
 621                 switch( $name ) {
 622                 case "sitename":
 623                 case "base":
 624                 case "generator":
 625                 case "case":
 626                 case "namespaces":
 627                 case "namespace":
 628                         break;
 629                 default:
 630                         return $this->throwXMLerror( "Element <$name> not allowed in <siteinfo>." );
 631                 }
 632         }
 633
 634         function out_siteinfo( $parser, $name ) {
 635                 if( $name == "siteinfo" ) {
 636                         xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 637                 }
 638         }
 639
 640
 641         function in_page( $parser, $name, $attribs ) {
 642                 $this->debug( "in_page $name" );
 643                 switch( $name ) {
 644                 case "id":
 645                 case "title":
 646                 case "restrictions":
 647                         $this->appendfield = $name;
 648                         $this->appenddata = "";
 649                         $this->parenttag = "page";
 650                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 651                         xml_set_character_data_handler( $parser, "char_append" );
 652                         break;
 653                 case "revision":
 654                         if( is_object( $this->pageTitle ) ) {
 655                                 $this->workRevision = new WikiRevision;
 656                                 $this->workRevision->setTitle( $this->pageTitle );
 657                                 $this->workRevisionCount++;
 658                         } else {
 659                                 // Skipping items due to invalid page title
 660                                 $this->workRevision = null;
 661                         }
 662                         xml_set_element_handler( $parser, "in_revision", "out_revision" );
 663                         break;
 664                 default:
 665                         return $this->throwXMLerror( "Element <$name> not allowed in a <page>." );
 666                 }
 667         }
 668
 669         function out_page( $parser, $name ) {
 670                 $this->debug( "out_page $name" );
 671                 if( $name != "page" ) {
 672                         return $this->throwXMLerror( "Expected </page>, got </$name>" );
 673                 }
 674                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 675
 676                 $this->pageOutCallback( $this->pageTitle, $this->origTitle,
 677                         $this->workRevisionCount, $this->workSuccessCount );
 678
 679                 $this->workTitle = null;
 680                 $this->workRevision = null;
 681                 $this->workRevisionCount = 0;
 682                 $this->workSuccessCount = 0;
 683                 $this->pageTitle = null;
 684                 $this->origTitle = null;
 685         }
 686
 687         function in_nothing( $parser, $name, $attribs ) {
 688                 $this->debug( "in_nothing $name" );
 689                 return $this->throwXMLerror( "No child elements allowed here; got <$name>" );
 690         }
 691         function char_append( $parser, $data ) {
 692                 $this->debug( "char_append '$data'" );
 693                 $this->appenddata .= $data;
 694         }
 695         function out_append( $parser, $name ) {
 696                 $this->debug( "out_append $name" );
 697                 if( $name != $this->appendfield ) {
 698                         return $this->throwXMLerror( "Expected </{$this->appendfield}>, got </$name>" );
 699                 }
 700                 xml_set_element_handler( $parser, "in_$this->parenttag", "out_$this->parenttag" );
 701                 xml_set_character_data_handler( $parser, "donothing" );
 702
 703                 switch( $this->appendfield ) {
 704                 case "title":
 705                         $this->workTitle = $this->appenddata;
 706
 707                         if( $this->mTargetArticleName != '' ) {
 708                                 // Import article with a different target article name
 709                                 $this->workTitle = $this->mTargetArticleName;
 710                         }
 711                         $this->origTitle = Title::newFromText( $this->workTitle );
 712                         if( !is_null( $this->mTargetNamespace ) && !is_null( $this->origTitle ) ) {
 713                                 $this->pageTitle = Title::makeTitle( $this->mTargetNamespace,
 714                                         $this->origTitle->getDbKey() );
 715                         } else {
 716                                 $this->pageTitle = Title::newFromText( $this->workTitle );
 717                         }
 718                         if( is_null( $this->pageTitle ) ) {
 719                                 // Invalid page title? Ignore the page
 720                                 $this->notice( "Skipping invalid page title '$this->workTitle'" );
 721                         } else {
 722                                 $this->pageCallback( $this->workTitle );
 723                         }
 724                         break;
 725                 case "id":
 726                         if ( $this->parenttag == 'revision' ) {
 727                                 if( $this->workRevision )
 728                                         $this->workRevision->setID( $this->appenddata );
 729                         }
 730                         break;
 731                 case "text":
 732                         if( $this->workRevision )
 733                                 $this->workRevision->setText( $this->appenddata );
 734                         break;
 735                 case "username":
 736                         if( $this->workRevision )
 737                                 $this->workRevision->setUsername( $this->appenddata );
 738                         break;
 739                 case "ip":
 740                         if( $this->workRevision )
 741                                 $this->workRevision->setUserIP( $this->appenddata );
 742                         break;
 743                 case "timestamp":
 744                         if( $this->workRevision )
 745                                 $this->workRevision->setTimestamp( $this->appenddata );
 746                         break;
 747                 case "comment":
 748                         if( $this->workRevision )
 749                                 $this->workRevision->setComment( $this->appenddata );
 750                         break;
 751                 case "minor":
 752                         if( $this->workRevision )
 753                                 $this->workRevision->setMinor( true );
 754                         break;
 755                 default:
 756                         $this->debug( "Bad append: {$this->appendfield}" );
 757                 }
 758                 $this->appendfield = "";
 759                 $this->appenddata = "";
 760         }
 761
 762         function in_revision( $parser, $name, $attribs ) {
 763                 $this->debug( "in_revision $name" );
 764                 switch( $name ) {
 765                 case "id":
 766                 case "timestamp":
 767                 case "comment":
 768                 case "minor":
 769                 case "text":
 770                         $this->parenttag = "revision";
 771                         $this->appendfield = $name;
 772                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 773                         xml_set_character_data_handler( $parser, "char_append" );
 774                         break;
 775                 case "contributor":
 776                         xml_set_element_handler( $parser, "in_contributor", "out_contributor" );
 777                         break;
 778                 default:
 779                         return $this->throwXMLerror( "Element <$name> not allowed in a <revision>." );
 780                 }
 781         }
 782
 783         function out_revision( $parser, $name ) {
 784                 $this->debug( "out_revision $name" );
 785                 if( $name != "revision" ) {
 786                         return $this->throwXMLerror( "Expected </revision>, got </$name>" );
 787                 }
 788                 xml_set_element_handler( $parser, "in_page", "out_page" );
 789
 790                 if( $this->workRevision ) {
 791                         $ok = call_user_func_array( $this->mRevisionCallback,
 792                                 array( &$this->workRevision, &$this ) );
 793                         if( $ok ) {
 794                                 $this->workSuccessCount++;
 795                         }
 796                 }
 797         }
 798
 799         function in_contributor( $parser, $name, $attribs ) {
 800                 $this->debug( "in_contributor $name" );
 801                 switch( $name ) {
 802                 case "username":
 803                 case "ip":
 804                 case "id":
 805                         $this->parenttag = "contributor";
 806                         $this->appendfield = $name;
 807                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 808                         xml_set_character_data_handler( $parser, "char_append" );
 809                         break;
 810                 default:
 811                         $this->throwXMLerror( "Invalid tag <$name> in <contributor>" );
 812                 }
 813         }
 814
 815         function out_contributor( $parser, $name ) {
 816                 $this->debug( "out_contributor $name" );
 817                 if( $name != "contributor" ) {
 818                         return $this->throwXMLerror( "Expected </contributor>, got </$name>" );
 819                 }
 820                 xml_set_element_handler( $parser, "in_revision", "out_revision" );
 821         }
 822
 823 }
 824
 825 /**
 826  * @todo document (e.g. one-sentence class description).
 827  * @addtogroup SpecialPage
 828  */
 829 class ImportStringSource {
 830         function ImportStringSource( $string ) {
 831                 $this->mString = $string;
 832                 $this->mRead = false;
 833         }
 834
 835         function atEnd() {
 836                 return $this->mRead;
 837         }
 838
 839         function readChunk() {
 840                 if( $this->atEnd() ) {
 841                         return false;
 842                 } else {
 843                         $this->mRead = true;
 844                         return $this->mString;
 845                 }
 846         }
 847 }
 848
 849 /**
 850  * @todo document (e.g. one-sentence class description).
 851  * @addtogroup SpecialPage
 852  */
 853 class ImportStreamSource {
 854         function ImportStreamSource( $handle ) {
 855                 $this->mHandle = $handle;
 856         }
 857
 858         function atEnd() {
 859                 return feof( $this->mHandle );
 860         }
 861
 862         function readChunk() {
 863                 return fread( $this->mHandle, 32768 );
 864         }
 865
 866         static function newFromFile( $filename ) {
 867                 $file = @fopen( $filename, 'rt' );
 868                 if( !$file ) {
 869                         return new WikiErrorMsg( "importcantopen" );
 870                 }
 871                 return new ImportStreamSource( $file );
 872         }
 873
 874         static function newFromUpload( $fieldname = "xmlimport" ) {
 875                 $upload =& $_FILES[$fieldname];
 876
 877                 if( !isset( $upload ) || !$upload['name'] ) {
 878                         return new WikiErrorMsg( 'importnofile' );
 879                 }
 880                 if( !empty( $upload['error'] ) ) {
 881                         return new WikiErrorMsg( 'importuploaderror', $upload['error'] );
 882                 }
 883                 $fname = $upload['tmp_name'];
 884                 if( is_uploaded_file( $fname ) ) {
 885                         return ImportStreamSource::newFromFile( $fname );
 886                 } else {
 887                         return new WikiErrorMsg( 'importnofile' );
 888                 }
 889         }
 890
 891         function newFromURL( $url, $method = 'GET' ) {
 892                 wfDebug( __METHOD__ . ": opening $url\n" );
 893                 # Use the standard HTTP fetch function; it times out
 894                 # quicker and sorts out user-agent problems which might
 895                 # otherwise prevent importing from large sites, such
 896                 # as the Wikimedia cluster, etc.
 897                 $data = Http::request( $method, $url );
 898                 if( $data !== false ) {
 899                         $file = tmpfile();
 900                         fwrite( $file, $data );
 901                         fflush( $file );
 902                         fseek( $file, 0 );
 903                         return new ImportStreamSource( $file );
 904                 } else {
 905                         return new WikiErrorMsg( 'importcantopen' );
 906                 }
 907         }
 908
 909         public static function newFromInterwiki( $interwiki, $page, $history=false ) {
 910                 if( $page == '' ) {
 911                         return new WikiErrorMsg( 'import-noarticle' );
 912                 }
 913                 $link = Title::newFromText( "$interwiki:Special:Export/$page" );
 914                 if( is_null( $link ) || $link->getInterwiki() == '' ) {
 915                         return new WikiErrorMsg( 'importbadinterwiki' );
 916                 } else {
 917                         $params = $history ? 'history=1' : '';
 918                         $url = $link->getFullUrl( $params );
 919                         # For interwikis, use POST to avoid redirects.
 920                         return ImportStreamSource::newFromURL( $url, "POST" );
 921                 }
 922         }
 923 }
 924
 925
 926