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  * 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
  20  * http://www.gnu.org/copyleft/gpl.html
  21  *
  22  * @package MediaWiki
  23  * @subpackage SpecialPage
  24  */
  25
  26 /** */
  27 require_once( 'WikiError.php' );
  28
  29 /**
  30  * Constructor
  31  */
  32 function wfSpecialImport( $page = '' ) {
  33         global $wgUser, $wgOut, $wgRequest, $wgTitle, $wgImportSources;
  34
  35         ###
  36 #       $wgOut->addWikiText( "Special:Import is not ready for this beta release, sorry." );
  37 #       return;
  38         ###
  39
  40         if( $wgRequest->wasPosted() && $wgRequest->getVal( 'action' ) == 'submit') {
  41                 switch( $wgRequest->getVal( "source" ) ) {
  42                 case "upload":
  43                         if( $wgUser->isAllowed( 'importupload' ) ) {
  44                                 $source = ImportStreamSource::newFromUpload( "xmlimport" );
  45                         } else {
  46                                 return $wgOut->permissionRequired( 'importupload' );
  47                         }
  48                         break;
  49                 case "interwiki":
  50                         $source = ImportStreamSource::newFromInterwiki(
  51                                 $wgRequest->getVal( "interwiki" ),
  52                                 $wgRequest->getText( "frompage" ) );
  53                         break;
  54                 default:
  55                         $source = new WikiError( "Unknown import source type" );
  56                 }
  57
  58                 if( WikiError::isError( $source ) ) {
  59                         $wgOut->addWikiText( wfEscapeWikiText( $source->getMessage() ) );
  60                 } else {
  61                         $importer = new WikiImporter( $source );
  62                         $result = $importer->doImport();
  63                         if( WikiError::isError( $result ) ) {
  64                                 $wgOut->addWikiText( wfMsg( "importfailed",
  65                                         wfEscapeWikiText( $result->getMessage() ) ) );
  66                         } else {
  67                                 # Success!
  68                                 $wgOut->addWikiText( wfMsg( "importsuccess" ) );
  69                         }
  70                 }
  71         }
  72
  73         $action = $wgTitle->escapeLocalUrl( 'action=submit' );
  74
  75         if( $wgUser->isAllowed( 'importupload' ) ) {
  76                 $wgOut->addWikiText( wfMsg( "importtext" ) );
  77                 $wgOut->addHTML( "
  78 <fieldset>
  79         <legend>" . wfMsgHtml('upload') . "</legend>
  80         <form enctype='multipart/form-data' method='post' action=\"$action\">
  81                 <input type='hidden' name='action' value='submit' />
  82                 <input type='hidden' name='source' value='upload' />
  83                 <input type='hidden' name='MAX_FILE_SIZE' value='2000000' />
  84                 <input type='file' name='xmlimport' value='' size='30' />
  85                 <input type='submit' value='" . wfMsgHtml( "uploadbtn" ) . "'/>
  86         </form>
  87 </fieldset>
  88 " );
  89         } else {
  90                 if( empty( $wgImportSources ) ) {
  91                         $wgOut->addWikiText( wfMsg( 'importnosources' ) );
  92                 }
  93         }
  94
  95         if( !empty( $wgImportSources ) ) {
  96                 $wgOut->addHTML( "
  97 <fieldset>
  98         <legend>" . wfMsgHtml('importinterwiki') . "</legend>
  99         <form method='post' action=\"$action\">
 100                 <input type='hidden' name='action' value='submit' />
 101                 <input type='hidden' name='source' value='interwiki' />
 102                 <select name='interwiki'>
 103 " );
 104                 foreach( $wgImportSources as $interwiki ) {
 105                         $iw = htmlspecialchars( $interwiki );
 106                         $wgOut->addHTML( "<option value=\"$iw\">$iw</option>\n" );
 107                 }
 108                 $wgOut->addHTML( "
 109                 </select>
 110                 <input name='frompage' />
 111                 <input type='submit' />
 112         </form>
 113 </fieldset>
 114 " );
 115         }
 116 }
 117
 118 /**
 119  *
 120  * @package MediaWiki
 121  * @subpackage SpecialPage
 122  */
 123 class WikiRevision {
 124         var $title = NULL;
 125         var $timestamp = "20010115000000";
 126         var $user = 0;
 127         var $user_text = "";
 128         var $text = "";
 129         var $comment = "";
 130         var $minor = false;
 131
 132         function setTitle( $text ) {
 133                 $this->title = Title::newFromText( $text );
 134         }
 135
 136         function setTimestamp( $ts ) {
 137                 # 2003-08-05T18:30:02Z
 138                 $this->timestamp = wfTimestamp( TS_MW, $ts );
 139         }
 140
 141         function setUsername( $user ) {
 142                 $this->user_text = $user;
 143         }
 144
 145         function setUserIP( $ip ) {
 146                 $this->user_text = $ip;
 147         }
 148
 149         function setText( $text ) {
 150                 $this->text = $text;
 151         }
 152
 153         function setComment( $text ) {
 154                 $this->comment = $text;
 155         }
 156
 157         function setMinor( $minor ) {
 158                 $this->minor = (bool)$minor;
 159         }
 160
 161         function getTitle() {
 162                 return $this->title;
 163         }
 164
 165         function getTimestamp() {
 166                 return $this->timestamp;
 167         }
 168
 169         function getUser() {
 170                 return $this->user_text;
 171         }
 172
 173         function getText() {
 174                 return $this->text;
 175         }
 176
 177         function getComment() {
 178                 return $this->comment;
 179         }
 180
 181         function getMinor() {
 182                 return $this->minor;
 183         }
 184
 185         function importOldRevision() {
 186                 $fname = "WikiImporter::importOldRevision";
 187                 $dbw =& wfGetDB( DB_MASTER );
 188
 189                 # Sneak a single revision into place
 190                 $user = User::newFromName( $this->getUser() );
 191                 if( $user ) {
 192                         $userId = intval( $user->getId() );
 193                         $userText = $user->getName();
 194                 } else {
 195                         $userId = 0;
 196                         $userText = $this->getUser();
 197                 }
 198
 199                 // avoid memory leak...?
 200                 $linkCache =& LinkCache::singleton();
 201                 $linkCache->clear();
 202
 203                 $article = new Article( $this->title );
 204                 $pageId = $article->getId();
 205                 if( $pageId == 0 ) {
 206                         # must create the page...
 207                         $pageId = $article->insertOn( $dbw );
 208                 }
 209
 210                 # FIXME: Check for exact conflicts
 211                 # FIXME: Use original rev_id optionally
 212                 # FIXME: blah blah blah
 213
 214                 #if( $numrows > 0 ) {
 215                 #       return wfMsg( "importhistoryconflict" );
 216                 #}
 217
 218                 # Insert the row
 219                 $revision = new Revision( array(
 220                         'page'       => $pageId,
 221                         'text'       => $this->getText(),
 222                         'comment'    => $this->getComment(),
 223                         'user'       => $userId,
 224                         'user_text'  => $userText,
 225                         'timestamp'  => $this->timestamp,
 226                         'minor_edit' => $this->minor,
 227                         ) );
 228                 $revId = $revision->insertOn( $dbw );
 229                 $article->updateIfNewerOn( $dbw, $revision );
 230
 231                 return true;
 232         }
 233
 234 }
 235
 236 /**
 237  *
 238  * @package MediaWiki
 239  * @subpackage SpecialPage
 240  */
 241 class WikiImporter {
 242         var $mSource = null;
 243         var $mPageCallback = null;
 244         var $mRevisionCallback = null;
 245         var $lastfield;
 246
 247         function WikiImporter( $source ) {
 248                 $this->setRevisionCallback( array( &$this, "importRevision" ) );
 249                 $this->mSource = $source;
 250         }
 251
 252         function throwXmlError( $err ) {
 253                 $this->debug( "FAILURE: $err" );
 254                 wfDebug( "WikiImporter XML error: $err\n" );
 255         }
 256
 257         # --------------
 258
 259         function doImport() {
 260                 if( empty( $this->mSource ) ) {
 261                         return new WikiErrorMsg( "importnotext" );
 262                 }
 263
 264                 $parser = xml_parser_create( "UTF-8" );
 265
 266                 # case folding violates XML standard, turn it off
 267                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 268
 269                 xml_set_object( $parser, $this );
 270                 xml_set_element_handler( $parser, "in_start", "" );
 271
 272                 $offset = 0; // for context extraction on error reporting
 273                 do {
 274                         $chunk = $this->mSource->readChunk();
 275                         if( !xml_parse( $parser, $chunk, $this->mSource->atEnd() ) ) {
 276                                 wfDebug( "WikiImporter::doImport encountered XML parsing error\n" );
 277                                 return new WikiXmlError( $parser, 'XML import parse failure', $chunk, $offset );
 278                         }
 279                         $offset += strlen( $chunk );
 280                 } while( $chunk !== false && !$this->mSource->atEnd() );
 281                 xml_parser_free( $parser );
 282
 283                 return true;
 284         }
 285
 286         function debug( $data ) {
 287                 #wfDebug( "IMPORT: $data\n" );
 288         }
 289
 290         function notice( $data ) {
 291                 global $wgCommandLineMode;
 292                 if( $wgCommandLineMode ) {
 293                         print "$data\n";
 294                 } else {
 295                         global $wgOut;
 296                         $wgOut->addHTML( "<li>$data</li>\n" );
 297                 }
 298         }
 299
 300         /**
 301          * Sets the action to perform as each new page in the stream is reached.
 302          * @param callable $callback
 303          * @return callable
 304          */
 305         function setPageCallback( $callback ) {
 306                 $previous = $this->mPageCallback;
 307                 $this->mPageCallback = $callback;
 308                 return $previous;
 309         }
 310
 311         /**
 312          * Sets the action to perform as each page revision is reached.
 313          * @param callable $callback
 314          * @return callable
 315          */
 316         function setRevisionCallback( $callback ) {
 317                 $previous = $this->mRevisionCallback;
 318                 $this->mRevisionCallback = $callback;
 319                 return $previous;
 320         }
 321
 322         /**
 323          * Default per-revision callback, performs the import.
 324          * @param WikiRevision $revision
 325          * @access private
 326          */
 327         function importRevision( &$revision ) {
 328                 $dbw =& wfGetDB( DB_MASTER );
 329                 $dbw->deadlockLoop( array( &$revision, 'importOldRevision' ) );
 330         }
 331
 332         /**
 333          * Alternate per-revision callback, for debugging.
 334          * @param WikiRevision $revision
 335          * @access private
 336          */
 337         function debugRevisionHandler( &$revision ) {
 338                 $this->debug( "Got revision:" );
 339                 if( is_object( $revision->title ) ) {
 340                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 341                 } else {
 342                         $this->debug( "-- Title: <invalid>" );
 343                 }
 344                 $this->debug( "-- User: " . $revision->user_text );
 345                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 346                 $this->debug( "-- Comment: " . $revision->comment );
 347                 $this->debug( "-- Text: " . $revision->text );
 348         }
 349
 350         /**
 351          * Notify the callback function when a new <page> is reached.
 352          * @param Title $title
 353          * @access private
 354          */
 355         function pageCallback( $title ) {
 356                 if( is_callable( $this->mPageCallback ) ) {
 357                         call_user_func( $this->mPageCallback, $title );
 358                 }
 359         }
 360
 361
 362         # XML parser callbacks from here out -- beware!
 363         function donothing( $parser, $x, $y="" ) {
 364                 #$this->debug( "donothing" );
 365         }
 366
 367         function in_start( $parser, $name, $attribs ) {
 368                 $this->debug( "in_start $name" );
 369                 if( $name != "mediawiki" ) {
 370                         return $this->throwXMLerror( "Expected <mediawiki>, got <$name>" );
 371                 }
 372                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 373         }
 374
 375         function in_mediawiki( $parser, $name, $attribs ) {
 376                 $this->debug( "in_mediawiki $name" );
 377                 if( $name == 'siteinfo' ) {
 378                         xml_set_element_handler( $parser, "in_siteinfo", "out_siteinfo" );
 379                 } elseif( $name == 'page' ) {
 380                         xml_set_element_handler( $parser, "in_page", "out_page" );
 381                 } else {
 382                         return $this->throwXMLerror( "Expected <page>, got <$name>" );
 383                 }
 384         }
 385         function out_mediawiki( $parser, $name ) {
 386                 $this->debug( "out_mediawiki $name" );
 387                 if( $name != "mediawiki" ) {
 388                         return $this->throwXMLerror( "Expected </mediawiki>, got </$name>" );
 389                 }
 390                 xml_set_element_handler( $parser, "donothing", "donothing" );
 391         }
 392
 393
 394         function in_siteinfo( $parser, $name, $attribs ) {
 395                 // no-ops for now
 396                 $this->debug( "in_siteinfo $name" );
 397                 switch( $name ) {
 398                 case "sitename":
 399                 case "base":
 400                 case "generator":
 401                 case "case":
 402                 case "namespaces":
 403                 case "namespace":
 404                         break;
 405                 default:
 406                         return $this->throwXMLerror( "Element <$name> not allowed in <siteinfo>." );
 407                 }
 408         }
 409
 410         function out_siteinfo( $parser, $name ) {
 411                 if( $name == "siteinfo" ) {
 412                         xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 413                 }
 414         }
 415
 416
 417         function in_page( $parser, $name, $attribs ) {
 418                 $this->debug( "in_page $name" );
 419                 switch( $name ) {
 420                 case "id":
 421                 case "title":
 422                 case "restrictions":
 423                         $this->appendfield = $name;
 424                         $this->appenddata = "";
 425                         $this->parenttag = "page";
 426                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 427                         xml_set_character_data_handler( $parser, "char_append" );
 428                         break;
 429                 case "revision":
 430                         $this->workRevision = new WikiRevision;
 431                         $this->workRevision->setTitle( $this->workTitle );
 432                         xml_set_element_handler( $parser, "in_revision", "out_revision" );
 433                         break;
 434                 default:
 435                         return $this->throwXMLerror( "Element <$name> not allowed in a <page>." );
 436                 }
 437         }
 438
 439         function out_page( $parser, $name ) {
 440                 $this->debug( "out_page $name" );
 441                 if( $name != "page" ) {
 442                         return $this->throwXMLerror( "Expected </page>, got </$name>" );
 443                 }
 444                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 445
 446                 $this->workTitle = NULL;
 447                 $this->workRevision = NULL;
 448         }
 449
 450         function in_nothing( $parser, $name, $attribs ) {
 451                 $this->debug( "in_nothing $name" );
 452                 return $this->throwXMLerror( "No child elements allowed here; got <$name>" );
 453         }
 454         function char_append( $parser, $data ) {
 455                 $this->debug( "char_append '$data'" );
 456                 $this->appenddata .= $data;
 457         }
 458         function out_append( $parser, $name ) {
 459                 $this->debug( "out_append $name" );
 460                 if( $name != $this->appendfield ) {
 461                         return $this->throwXMLerror( "Expected </{$this->appendfield}>, got </$name>" );
 462                 }
 463                 xml_set_element_handler( $parser, "in_$this->parenttag", "out_$this->parenttag" );
 464                 xml_set_character_data_handler( $parser, "donothing" );
 465
 466                 switch( $this->appendfield ) {
 467                 case "title":
 468                         $this->workTitle = $this->appenddata;
 469                         $this->pageCallback( $this->workTitle );
 470                         break;
 471                 case "text":
 472                         $this->workRevision->setText( $this->appenddata );
 473                         break;
 474                 case "username":
 475                         $this->workRevision->setUsername( $this->appenddata );
 476                         break;
 477                 case "ip":
 478                         $this->workRevision->setUserIP( $this->appenddata );
 479                         break;
 480                 case "timestamp":
 481                         $this->workRevision->setTimestamp( $this->appenddata );
 482                         break;
 483                 case "comment":
 484                         $this->workRevision->setComment( $this->appenddata );
 485                         break;
 486                 case "minor":
 487                         $this->workRevision->setMinor( true );
 488                         break;
 489                 default:
 490                         $this->debug( "Bad append: {$this->appendfield}" );
 491                 }
 492                 $this->appendfield = "";
 493                 $this->appenddata = "";
 494         }
 495
 496         function in_revision( $parser, $name, $attribs ) {
 497                 $this->debug( "in_revision $name" );
 498                 switch( $name ) {
 499                 case "id":
 500                 case "timestamp":
 501                 case "comment":
 502                 case "minor":
 503                 case "text":
 504                         $this->parenttag = "revision";
 505                         $this->appendfield = $name;
 506                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 507                         xml_set_character_data_handler( $parser, "char_append" );
 508                         break;
 509                 case "contributor":
 510                         xml_set_element_handler( $parser, "in_contributor", "out_contributor" );
 511                         break;
 512                 default:
 513                         return $this->throwXMLerror( "Element <$name> not allowed in a <revision>." );
 514                 }
 515         }
 516
 517         function out_revision( $parser, $name ) {
 518                 $this->debug( "out_revision $name" );
 519                 if( $name != "revision" ) {
 520                         return $this->throwXMLerror( "Expected </revision>, got </$name>" );
 521                 }
 522                 xml_set_element_handler( $parser, "in_page", "out_page" );
 523
 524                 $out = call_user_func_array( $this->mRevisionCallback,
 525                         array( &$this->workRevision, &$this ) );
 526                 if( !empty( $out ) ) {
 527                         global $wgOut;
 528                         $wgOut->addHTML( "<li>" . $out . "</li>\n" );
 529                 }
 530         }
 531
 532         function in_contributor( $parser, $name, $attribs ) {
 533                 $this->debug( "in_contributor $name" );
 534                 switch( $name ) {
 535                 case "username":
 536                 case "ip":
 537                 case "id":
 538                         $this->parenttag = "contributor";
 539                         $this->appendfield = $name;
 540                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 541                         xml_set_character_data_handler( $parser, "char_append" );
 542                         break;
 543                 default:
 544                         $this->throwXMLerror( "Invalid tag <$name> in <contributor>" );
 545                 }
 546         }
 547
 548         function out_contributor( $parser, $name ) {
 549                 $this->debug( "out_contributor $name" );
 550                 if( $name != "contributor" ) {
 551                         return $this->throwXMLerror( "Expected </contributor>, got </$name>" );
 552                 }
 553                 xml_set_element_handler( $parser, "in_revision", "out_revision" );
 554         }
 555
 556 }
 557
 558 /** @package MediaWiki */
 559 class ImportStringSource {
 560         function ImportStringSource( $string ) {
 561                 $this->mString = $string;
 562                 $this->mRead = false;
 563         }
 564
 565         function atEnd() {
 566                 return $this->mRead;
 567         }
 568
 569         function readChunk() {
 570                 if( $this->atEnd() ) {
 571                         return false;
 572                 } else {
 573                         $this->mRead = true;
 574                         return $this->mString;
 575                 }
 576         }
 577 }
 578
 579 /** @package MediaWiki */
 580 class ImportStreamSource {
 581         function ImportStreamSource( $handle ) {
 582                 $this->mHandle = $handle;
 583         }
 584
 585         function atEnd() {
 586                 return feof( $this->mHandle );
 587         }
 588
 589         function readChunk() {
 590                 return fread( $this->mHandle, 32768 );
 591         }
 592
 593         function newFromFile( $filename ) {
 594                 $file = @fopen( $filename, 'rt' );
 595                 if( !$file ) {
 596                         return new WikiError( "Couldn't open import file" );
 597                 }
 598                 return new ImportStreamSource( $file );
 599         }
 600
 601         function newFromUpload( $fieldname = "xmlimport" ) {
 602                 $upload =& $_FILES[$fieldname];
 603
 604                 if( !isset( $upload ) ) {
 605                         return new WikiErrorMsg( 'importnofile' );
 606                 }
 607                 if( !empty( $upload['error'] ) ) {
 608                         return new WikiErrorMsg( 'importuploaderror', $upload['error'] );
 609                 }
 610                 $fname = $upload['tmp_name'];
 611                 if( is_uploaded_file( $fname ) ) {
 612                         return ImportStreamSource::newFromFile( $fname );
 613                 } else {
 614                         return new WikiErrorMsg( 'importnofile' );
 615                 }
 616         }
 617
 618         function newFromURL( $url ) {
 619                 # fopen-wrappers are normally turned off for security.
 620                 ini_set( "allow_url_fopen", true );
 621                 $ret = ImportStreamSource::newFromFile( $url );
 622                 ini_set( "allow_url_fopen", false );
 623                 return $ret;
 624         }
 625
 626         function newFromInterwiki( $interwiki, $page ) {
 627                 $base = Title::getInterwikiLink( $interwiki );
 628                 if( empty( $base ) ) {
 629                         return new WikiError( 'Bad interwiki link' );
 630                 } else {
 631                         $import = wfUrlencode( "Special:Export/$page" );
 632                         $url = str_replace( "$1", $import, $base );
 633                         return ImportStreamSource::newFromURL( $url );
 634                 }
 635         }
 636 }
 637
 638
 639 ?>