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 require_once( 'WikiError.php' );
  27
  28 /**
  29  * Constructor
  30  */
  31 function wfSpecialImport( $page = '' ) {
  32         global $wgUser, $wgOut, $wgLang, $wgRequest, $wgTitle;
  33         global $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
 131         function setTitle( $text ) {
 132                 $this->title = Title::newFromText( $text );
 133         }
 134
 135         function setTimestamp( $ts ) {
 136                 # 2003-08-05T18:30:02Z
 137                 $this->timestamp = preg_replace( '/^(....)-(..)-(..)T(..):(..):(..)Z$/', '$1$2$3$4$5$6', $ts );
 138         }
 139
 140         function setUsername( $user ) {
 141                 $this->user_text = $user;
 142         }
 143
 144         function setUserIP( $ip ) {
 145                 $this->user_text = $ip;
 146         }
 147
 148         function setText( $text ) {
 149                 $this->text = $text;
 150         }
 151
 152         function setComment( $text ) {
 153                 $this->comment = $text;
 154         }
 155
 156         function getTitle() {
 157                 return $this->title;
 158         }
 159
 160         function getTimestamp() {
 161                 return $this->timestamp;
 162         }
 163
 164         function getUser() {
 165                 return $this->user_text;
 166         }
 167
 168         function getText() {
 169                 return $this->text;
 170         }
 171
 172         function getComment() {
 173                 return $this->comment;
 174         }
 175
 176         function importOldRevision() {
 177                 $fname = "WikiImporter::importOldRevision";
 178                 $dbw =& wfGetDB( DB_MASTER );
 179
 180                 # Sneak a single revision into place
 181                 $user = User::newFromName( $this->getUser() );
 182                 if( $user ) {
 183                         $userId = IntVal( $user->getId() );
 184                         $userText = $user->getName();
 185                 } else {
 186                         $userId = 0;
 187                         $userText = $this->getUser();
 188                 }
 189
 190                 $article = new Article( $this->title );
 191                 $pageId = $article->getId();
 192                 if( $pageId == 0 ) {
 193                         # must create the page...
 194                         $pageId = $article->insertOn( $dbw );
 195                 }
 196
 197                 # FIXME: Check for exact conflicts
 198                 # FIXME: Use original rev_id optionally
 199                 # FIXME: blah blah blah
 200
 201                 #if( $numrows > 0 ) {
 202                 #       return wfMsg( "importhistoryconflict" );
 203                 #}
 204
 205                 # Insert the row
 206                 $revision = new Revision( array(
 207                         'page'       => $pageId,
 208                         'text'       => $this->getText(),
 209                         'comment'    => $this->getComment(),
 210                         'user'       => $userId,
 211                         'user_text'  => $userText,
 212                         'timestamp'  => $this->timestamp,
 213                         'minor_edit' => 0
 214                         ) );
 215                 $revId = $revision->insertOn( $dbw );
 216                 $article->updateIfNewerOn( $dbw, $revision );
 217
 218                 return true;
 219         }
 220
 221 }
 222
 223 /**
 224  *
 225  * @package MediaWiki
 226  * @subpackage SpecialPage
 227  */
 228 class WikiImporter {
 229         var $mSource = null;
 230         var $mPageCallback = null;
 231         var $mRevisionCallback = null;
 232         var $lastfield;
 233
 234         function WikiImporter( $source ) {
 235                 $this->setRevisionCallback( array( &$this, "importRevision" ) );
 236                 $this->mSource = $source;
 237         }
 238
 239         function throwXmlError( $err ) {
 240                 $this->debug( "FAILURE: $err" );
 241         }
 242
 243         # --------------
 244
 245         function doImport() {
 246                 if( empty( $this->mSource ) ) {
 247                         return new WikiErrorMsg( "importnotext" );
 248                 }
 249
 250                 $parser = xml_parser_create( "UTF-8" );
 251
 252                 # case folding violates XML standard, turn it off
 253                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 254
 255                 xml_set_object( $parser, &$this );
 256                 xml_set_element_handler( $parser, "in_start", "" );
 257
 258                 do {
 259                         $chunk = $this->mSource->readChunk();
 260                         if( !xml_parse( $parser, $chunk, $this->mSource->atEnd() ) ) {
 261                                 return new WikiXmlError( $parser );
 262                         }
 263                 } while( $chunk !== false && !$this->mSource->atEnd() );
 264                 xml_parser_free( $parser );
 265
 266                 return true;
 267         }
 268
 269         function debug( $data ) {
 270                 #wfDebug( "IMPORT: $data\n" );
 271         }
 272
 273         function notice( $data ) {
 274                 global $wgCommandLineMode;
 275                 if( $wgCommandLineMode ) {
 276                         print "$data\n";
 277                 } else {
 278                         global $wgOut;
 279                         $wgOut->addHTML( "<li>$data</li>\n" );
 280                 }
 281         }
 282
 283         /**
 284          * Sets the action to perform as each new page in the stream is reached.
 285          * @param callable $callback
 286          * @return callable
 287          */
 288         function setPageCallback( $callback ) {
 289                 $previous = $this->mPageCallback;
 290                 $this->mPageCallback = $callback;
 291                 return $previous;
 292         }
 293
 294         /**
 295          * Sets the action to perform as each page revision is reached.
 296          * @param callable $callback
 297          * @return callable
 298          */
 299         function setRevisionCallback( $callback ) {
 300                 $previous = $this->mRevisionCallback;
 301                 $this->mRevisionCallback = $callback;
 302                 return $previous;
 303         }
 304
 305         /**
 306          * Default per-revision callback, performs the import.
 307          * @param WikiRevision $revision
 308          * @access private
 309          */
 310         function importRevision( &$revision ) {
 311                 $dbw =& wfGetDB( DB_MASTER );
 312                 $dbw->deadlockLoop( array( &$revision, 'importOldRevision' ) );
 313         }
 314
 315         /**
 316          * Alternate per-revision callback, for debugging.
 317          * @param WikiRevision $revision
 318          * @access private
 319          */
 320         function debugRevisionHandler( &$revision ) {
 321                 $this->debug( "Got revision:" );
 322                 if( is_object( $revision->title ) ) {
 323                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 324                 } else {
 325                         $this->debug( "-- Title: <invalid>" );
 326                 }
 327                 $this->debug( "-- User: " . $revision->user_text );
 328                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 329                 $this->debug( "-- Comment: " . $revision->comment );
 330                 $this->debug( "-- Text: " . $revision->text );
 331         }
 332
 333         /**
 334          * Notify the callback function when a new <page> is reached.
 335          * @param Title $title
 336          * @access private
 337          */
 338         function pageCallback( $title ) {
 339                 if( is_callable( $this->mPageCallback ) ) {
 340                         call_user_func( $this->mPageCallback, $title );
 341                 }
 342         }
 343
 344
 345         # XML parser callbacks from here out -- beware!
 346         function donothing( $parser, $x, $y="" ) {
 347                 #$this->debug( "donothing" );
 348         }
 349
 350         function in_start( $parser, $name, $attribs ) {
 351                 $this->debug( "in_start $name" );
 352                 if( $name != "mediawiki" ) {
 353                         return $this->throwXMLerror( "Expected <mediawiki>, got <$name>" );
 354                 }
 355                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 356         }
 357
 358         function in_mediawiki( $parser, $name, $attribs ) {
 359                 $this->debug( "in_mediawiki $name" );
 360                 if( $name == 'siteinfo' ) {
 361                         xml_set_element_handler( $parser, "in_siteinfo", "out_siteinfo" );
 362                 } elseif( $name == 'page' ) {
 363                         xml_set_element_handler( $parser, "in_page", "out_page" );
 364                 } else {
 365                         return $this->throwXMLerror( "Expected <page>, got <$name>" );
 366                 }
 367         }
 368         function out_mediawiki( $parser, $name ) {
 369                 $this->debug( "out_mediawiki $name" );
 370                 if( $name != "mediawiki" ) {
 371                         return $this->throwXMLerror( "Expected </mediawiki>, got </$name>" );
 372                 }
 373                 xml_set_element_handler( $parser, "donothing", "donothing" );
 374         }
 375
 376
 377         function in_siteinfo( $parser, $name, $attribs ) {
 378                 // no-ops for now
 379                 $this->debug( "in_siteinfo $name" );
 380                 switch( $name ) {
 381                 case "sitename":
 382                 case "generator":
 383                 case "case":
 384                 case "namespaces":
 385                 case "namespace":
 386                         break;
 387                 default:
 388                         return $this->throwXMLerror( "Element <$name> not allowed in <siteinfo>." );
 389                 }
 390         }
 391
 392         function out_siteinfo( $parser, $name ) {
 393                 if( $name == "siteinfo" ) {
 394                         xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 395                 }
 396         }
 397
 398
 399         function in_page( $parser, $name, $attribs ) {
 400                 $this->debug( "in_page $name" );
 401                 switch( $name ) {
 402                 case "id":
 403                 case "title":
 404                 case "restrictions":
 405                         $this->appendfield = $name;
 406                         $this->appenddata = "";
 407                         $this->parenttag = "page";
 408                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 409                         xml_set_character_data_handler( $parser, "char_append" );
 410                         break;
 411                 case "revision":
 412                         $this->workRevision = new WikiRevision;
 413                         $this->workRevision->setTitle( $this->workTitle );
 414                         xml_set_element_handler( $parser, "in_revision", "out_revision" );
 415                         break;
 416                 default:
 417                         return $this->throwXMLerror( "Element <$name> not allowed in a <page>." );
 418                 }
 419         }
 420
 421         function out_page( $parser, $name ) {
 422                 $this->debug( "out_page $name" );
 423                 if( $name != "page" ) {
 424                         return $this->throwXMLerror( "Expected </page>, got </$name>" );
 425                 }
 426                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 427
 428                 $this->workTitle = NULL;
 429                 $this->workRevision = NULL;
 430         }
 431
 432         function in_nothing( $parser, $name, $attribs ) {
 433                 $this->debug( "in_nothing $name" );
 434                 return $this->throwXMLerror( "No child elements allowed here; got <$name>" );
 435         }
 436         function char_append( $parser, $data ) {
 437                 $this->debug( "char_append '$data'" );
 438                 $this->appenddata .= $data;
 439         }
 440         function out_append( $parser, $name ) {
 441                 $this->debug( "out_append $name" );
 442                 if( $name != $this->appendfield ) {
 443                         return $this->throwXMLerror( "Expected </{$this->appendfield}>, got </$name>" );
 444                 }
 445                 xml_set_element_handler( $parser, "in_$this->parenttag", "out_$this->parenttag" );
 446                 xml_set_character_data_handler( $parser, "donothing" );
 447
 448                 switch( $this->appendfield ) {
 449                 case "title":
 450                         $this->workTitle = $this->appenddata;
 451                         $this->pageCallback( $this->workTitle );
 452                         break;
 453                 case "text":
 454                         $this->workRevision->setText( $this->appenddata );
 455                         break;
 456                 case "username":
 457                         $this->workRevision->setUsername( $this->appenddata );
 458                         break;
 459                 case "ip":
 460                         $this->workRevision->setUserIP( $this->appenddata );
 461                         break;
 462                 case "timestamp":
 463                         $this->workRevision->setTimestamp( $this->appenddata );
 464                         break;
 465                 case "comment":
 466                         $this->workRevision->setComment( $this->appenddata );
 467                         break;
 468                 default:
 469                         $this->debug( "Bad append: {$this->appendfield}" );
 470                 }
 471                 $this->appendfield = "";
 472                 $this->appenddata = "";
 473         }
 474
 475         function in_revision( $parser, $name, $attribs ) {
 476                 $this->debug( "in_revision $name" );
 477                 switch( $name ) {
 478                 case "id":
 479                 case "timestamp":
 480                 case "comment":
 481                 case "text":
 482                         $this->parenttag = "revision";
 483                         $this->appendfield = $name;
 484                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 485                         xml_set_character_data_handler( $parser, "char_append" );
 486                         break;
 487                 case "contributor":
 488                         xml_set_element_handler( $parser, "in_contributor", "out_contributor" );
 489                         break;
 490                 default:
 491                         return $this->throwXMLerror( "Element <$name> not allowed in a <revision>." );
 492                 }
 493         }
 494
 495         function out_revision( $parser, $name ) {
 496                 $this->debug( "out_revision $name" );
 497                 if( $name != "revision" ) {
 498                         return $this->throwXMLerror( "Expected </revision>, got </$name>" );
 499                 }
 500                 xml_set_element_handler( $parser, "in_page", "out_page" );
 501
 502                 $out = call_user_func( $this->mRevisionCallback,
 503                         &$this->workRevision,
 504                         &$this );
 505                 if( !empty( $out ) ) {
 506                         global $wgOut;
 507                         $wgOut->addHTML( "<li>" . $out . "</li>\n" );
 508                 }
 509         }
 510
 511         function in_contributor( $parser, $name, $attribs ) {
 512                 $this->debug( "in_contributor $name" );
 513                 switch( $name ) {
 514                 case "username":
 515                 case "ip":
 516                         $this->parenttag = "contributor";
 517                         $this->appendfield = $name;
 518                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 519                         xml_set_character_data_handler( $parser, "char_append" );
 520                         break;
 521                 default:
 522                         $this->throwXMLerror( "Invalid tag <$name> in <contributor>" );
 523                 }
 524         }
 525
 526         function out_contributor( $parser, $name ) {
 527                 $this->debug( "out_contributor $name" );
 528                 if( $name != "contributor" ) {
 529                         return $this->throwXMLerror( "Expected </contributor>, got </$name>" );
 530                 }
 531                 xml_set_element_handler( $parser, "in_revision", "out_revision" );
 532         }
 533
 534 }
 535
 536 class ImportStringSource {
 537         function ImportStringSource( $string ) {
 538                 $this->mString = $string;
 539                 $this->mRead = false;
 540         }
 541
 542         function atEnd() {
 543                 return $this->mRead;
 544         }
 545
 546         function readChunk() {
 547                 if( $this->atEnd() ) {
 548                         return false;
 549                 } else {
 550                         $this->mRead = true;
 551                         return $this->mString;
 552                 }
 553         }
 554 }
 555
 556 class ImportStreamSource {
 557         function ImportStreamSource( $handle ) {
 558                 $this->mHandle = $handle;
 559         }
 560
 561         function atEnd() {
 562                 return feof( $this->mHandle );
 563         }
 564
 565         function readChunk() {
 566                 return fread( $this->mHandle, 32768 );
 567         }
 568
 569         function newFromFile( $filename ) {
 570                 $file = @fopen( $filename, 'rt' );
 571                 if( !$file ) {
 572                         return new WikiError( "Couldn't open import file" );
 573                 }
 574                 return new ImportStreamSource( $file );
 575         }
 576
 577         function newFromUpload( $fieldname = "xmlimport" ) {
 578                 global $wgOut;
 579
 580                 $upload =& $_FILES[$fieldname];
 581
 582                 if( !isset( $upload ) ) {
 583                         return new WikiErrorMsg( 'importnofile' );
 584                 }
 585                 if( !empty( $upload['error'] ) ) {
 586                         return new WikiErrorMsg( 'importuploaderror', $upload['error'] );
 587                 }
 588                 $fname = $upload['tmp_name'];
 589                 if( is_uploaded_file( $fname ) ) {
 590                         return ImportStreamSource::newFromFile( $fname );
 591                 } else {
 592                         return new WikiErrorMsg( 'importnofile' );
 593                 }
 594         }
 595
 596         function newFromURL( $url ) {
 597                 # fopen-wrappers are normally turned off for security.
 598                 ini_set( "allow_url_fopen", true );
 599                 $ret = ImportStreamSource::newFromFile( $url );
 600                 ini_set( "allow_url_fopen", false );
 601                 return $ret;
 602         }
 603
 604         function newFromInterwiki( $interwiki, $page ) {
 605                 $base = Title::getInterwikiLink( $interwiki );
 606                 if( empty( $base ) ) {
 607                         return new WikiError( 'Bad interwiki link' );
 608                 } else {
 609                         $import = wfUrlencode( "Special:Export/$page" );
 610                         $url = str_replace( "$1", $import, $base );
 611                         return ImportStreamSource::newFromURL( $url );
 612                 }
 613         }
 614 }
 615
 616
 617 ?>