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, $wgLang, $wgRequest, $wgTitle;
  34         global $wgImportSources;
  35
  36         ###
  37 #       $wgOut->addWikiText( "Special:Import is not ready for this beta release, sorry." );
  38 #       return;
  39         ###
  40
  41         if( $wgRequest->wasPosted() && $wgRequest->getVal( 'action' ) == 'submit') {
  42                 switch( $wgRequest->getVal( "source" ) ) {
  43                 case "upload":
  44                         if( $wgUser->isAllowed( 'importupload' ) ) {
  45                                 $source = ImportStreamSource::newFromUpload( "xmlimport" );
  46                         } else {
  47                                 return $wgOut->permissionRequired( 'importupload' );
  48                         }
  49                         break;
  50                 case "interwiki":
  51                         $source = ImportStreamSource::newFromInterwiki(
  52                                 $wgRequest->getVal( "interwiki" ),
  53                                 $wgRequest->getText( "frompage" ) );
  54                         break;
  55                 default:
  56                         $source = new WikiError( "Unknown import source type" );
  57                 }
  58
  59                 if( WikiError::isError( $source ) ) {
  60                         $wgOut->addWikiText( wfEscapeWikiText( $source->getMessage() ) );
  61                 } else {
  62                         $importer = new WikiImporter( $source );
  63                         $result = $importer->doImport();
  64                         if( WikiError::isError( $result ) ) {
  65                                 $wgOut->addWikiText( wfMsg( "importfailed",
  66                                         wfEscapeWikiText( $result->getMessage() ) ) );
  67                         } else {
  68                                 # Success!
  69                                 $wgOut->addWikiText( wfMsg( "importsuccess" ) );
  70                         }
  71                 }
  72         }
  73
  74         $action = $wgTitle->escapeLocalUrl( 'action=submit' );
  75
  76         if( $wgUser->isAllowed( 'importupload' ) ) {
  77                 $wgOut->addWikiText( wfMsg( "importtext" ) );
  78                 $wgOut->addHTML( "
  79 <fieldset>
  80         <legend>" . wfMsgHtml('upload') . "</legend>
  81         <form enctype='multipart/form-data' method='post' action=\"$action\">
  82                 <input type='hidden' name='action' value='submit' />
  83                 <input type='hidden' name='source' value='upload' />
  84                 <input type='hidden' name='MAX_FILE_SIZE' value='2000000' />
  85                 <input type='file' name='xmlimport' value='' size='30' />
  86                 <input type='submit' value='" . wfMsgHtml( "uploadbtn" ) . "'/>
  87         </form>
  88 </fieldset>
  89 " );
  90         } else {
  91                 if( empty( $wgImportSources ) ) {
  92                         $wgOut->addWikiText( wfMsg( 'importnosources' ) );
  93                 }
  94         }
  95
  96         if( !empty( $wgImportSources ) ) {
  97                 $wgOut->addHTML( "
  98 <fieldset>
  99         <legend>" . wfMsgHtml('importinterwiki') . "</legend>
 100         <form method='post' action=\"$action\">
 101                 <input type='hidden' name='action' value='submit' />
 102                 <input type='hidden' name='source' value='interwiki' />
 103                 <select name='interwiki'>
 104 " );
 105                 foreach( $wgImportSources as $interwiki ) {
 106                         $iw = htmlspecialchars( $interwiki );
 107                         $wgOut->addHTML( "<option value=\"$iw\">$iw</option>\n" );
 108                 }
 109                 $wgOut->addHTML( "
 110                 </select>
 111                 <input name='frompage' />
 112                 <input type='submit' />
 113         </form>
 114 </fieldset>
 115 " );
 116         }
 117 }
 118
 119 /**
 120  *
 121  * @package MediaWiki
 122  * @subpackage SpecialPage
 123  */
 124 class WikiRevision {
 125         var $title = NULL;
 126         var $timestamp = "20010115000000";
 127         var $user = 0;
 128         var $user_text = "";
 129         var $text = "";
 130         var $comment = "";
 131         var $minor = false;
 132
 133         function setTitle( $text ) {
 134                 $this->title = Title::newFromText( $text );
 135         }
 136
 137         function setTimestamp( $ts ) {
 138                 # 2003-08-05T18:30:02Z
 139                 $this->timestamp = preg_replace( '/^(....)-(..)-(..)T(..):(..):(..)Z$/', '$1$2$3$4$5$6', $ts );
 140         }
 141
 142         function setUsername( $user ) {
 143                 $this->user_text = $user;
 144         }
 145
 146         function setUserIP( $ip ) {
 147                 $this->user_text = $ip;
 148         }
 149
 150         function setText( $text ) {
 151                 $this->text = $text;
 152         }
 153
 154         function setComment( $text ) {
 155                 $this->comment = $text;
 156         }
 157
 158         function setMinor( $minor ) {
 159                 $this->minor = (bool)$minor;
 160         }
 161
 162         function getTitle() {
 163                 return $this->title;
 164         }
 165
 166         function getTimestamp() {
 167                 return $this->timestamp;
 168         }
 169
 170         function getUser() {
 171                 return $this->user_text;
 172         }
 173
 174         function getText() {
 175                 return $this->text;
 176         }
 177
 178         function getComment() {
 179                 return $this->comment;
 180         }
 181
 182         function getMinor() {
 183                 return $this->minor;
 184         }
 185
 186         function importOldRevision() {
 187                 $fname = "WikiImporter::importOldRevision";
 188                 $dbw =& wfGetDB( DB_MASTER );
 189
 190                 # Sneak a single revision into place
 191                 $user = User::newFromName( $this->getUser() );
 192                 if( $user ) {
 193                         $userId = intval( $user->getId() );
 194                         $userText = $user->getName();
 195                 } else {
 196                         $userId = 0;
 197                         $userText = $this->getUser();
 198                 }
 199
 200                 $article = new Article( $this->title );
 201                 $pageId = $article->getId();
 202                 if( $pageId == 0 ) {
 203                         # must create the page...
 204                         $pageId = $article->insertOn( $dbw );
 205                 }
 206
 207                 # FIXME: Check for exact conflicts
 208                 # FIXME: Use original rev_id optionally
 209                 # FIXME: blah blah blah
 210
 211                 #if( $numrows > 0 ) {
 212                 #       return wfMsg( "importhistoryconflict" );
 213                 #}
 214
 215                 # Insert the row
 216                 $revision = new Revision( array(
 217                         'page'       => $pageId,
 218                         'text'       => $this->getText(),
 219                         'comment'    => $this->getComment(),
 220                         'user'       => $userId,
 221                         'user_text'  => $userText,
 222                         'timestamp'  => $this->timestamp,
 223                         'minor_edit' => $this->minor,
 224                         ) );
 225                 $revId = $revision->insertOn( $dbw );
 226                 $article->updateIfNewerOn( $dbw, $revision );
 227
 228                 return true;
 229         }
 230
 231 }
 232
 233 /**
 234  *
 235  * @package MediaWiki
 236  * @subpackage SpecialPage
 237  */
 238 class WikiImporter {
 239         var $mSource = null;
 240         var $mPageCallback = null;
 241         var $mRevisionCallback = null;
 242         var $lastfield;
 243
 244         function WikiImporter( $source ) {
 245                 $this->setRevisionCallback( array( &$this, "importRevision" ) );
 246                 $this->mSource = $source;
 247         }
 248
 249         function throwXmlError( $err ) {
 250                 $this->debug( "FAILURE: $err" );
 251                 wfDebug( "WikiImporter XML error: $err\n" );
 252         }
 253
 254         # --------------
 255
 256         function doImport() {
 257                 if( empty( $this->mSource ) ) {
 258                         return new WikiErrorMsg( "importnotext" );
 259                 }
 260
 261                 $parser = xml_parser_create( "UTF-8" );
 262
 263                 # case folding violates XML standard, turn it off
 264                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 265
 266                 xml_set_object( $parser, &$this );
 267                 xml_set_element_handler( $parser, "in_start", "" );
 268
 269                 $offset = 0; // for context extraction on error reporting
 270                 do {
 271                         $chunk = $this->mSource->readChunk();
 272                         if( !xml_parse( $parser, $chunk, $this->mSource->atEnd() ) ) {
 273                                 wfDebug( "WikiImporter::doImport encountered XML parsing error\n" );
 274                                 return new WikiXmlError( $parser, 'XML import parse failure', $chunk, $offset );
 275                         }
 276                         $offset += strlen( $chunk );
 277                 } while( $chunk !== false && !$this->mSource->atEnd() );
 278                 xml_parser_free( $parser );
 279
 280                 return true;
 281         }
 282
 283         function debug( $data ) {
 284                 #wfDebug( "IMPORT: $data\n" );
 285         }
 286
 287         function notice( $data ) {
 288                 global $wgCommandLineMode;
 289                 if( $wgCommandLineMode ) {
 290                         print "$data\n";
 291                 } else {
 292                         global $wgOut;
 293                         $wgOut->addHTML( "<li>$data</li>\n" );
 294                 }
 295         }
 296
 297         /**
 298          * Sets the action to perform as each new page in the stream is reached.
 299          * @param callable $callback
 300          * @return callable
 301          */
 302         function setPageCallback( $callback ) {
 303                 $previous = $this->mPageCallback;
 304                 $this->mPageCallback = $callback;
 305                 return $previous;
 306         }
 307
 308         /**
 309          * Sets the action to perform as each page revision is reached.
 310          * @param callable $callback
 311          * @return callable
 312          */
 313         function setRevisionCallback( $callback ) {
 314                 $previous = $this->mRevisionCallback;
 315                 $this->mRevisionCallback = $callback;
 316                 return $previous;
 317         }
 318
 319         /**
 320          * Default per-revision callback, performs the import.
 321          * @param WikiRevision $revision
 322          * @access private
 323          */
 324         function importRevision( &$revision ) {
 325                 $dbw =& wfGetDB( DB_MASTER );
 326                 $dbw->deadlockLoop( array( &$revision, 'importOldRevision' ) );
 327         }
 328
 329         /**
 330          * Alternate per-revision callback, for debugging.
 331          * @param WikiRevision $revision
 332          * @access private
 333          */
 334         function debugRevisionHandler( &$revision ) {
 335                 $this->debug( "Got revision:" );
 336                 if( is_object( $revision->title ) ) {
 337                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 338                 } else {
 339                         $this->debug( "-- Title: <invalid>" );
 340                 }
 341                 $this->debug( "-- User: " . $revision->user_text );
 342                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 343                 $this->debug( "-- Comment: " . $revision->comment );
 344                 $this->debug( "-- Text: " . $revision->text );
 345         }
 346
 347         /**
 348          * Notify the callback function when a new <page> is reached.
 349          * @param Title $title
 350          * @access private
 351          */
 352         function pageCallback( $title ) {
 353                 if( is_callable( $this->mPageCallback ) ) {
 354                         call_user_func( $this->mPageCallback, $title );
 355                 }
 356         }
 357
 358
 359         # XML parser callbacks from here out -- beware!
 360         function donothing( $parser, $x, $y="" ) {
 361                 #$this->debug( "donothing" );
 362         }
 363
 364         function in_start( $parser, $name, $attribs ) {
 365                 $this->debug( "in_start $name" );
 366                 if( $name != "mediawiki" ) {
 367                         return $this->throwXMLerror( "Expected <mediawiki>, got <$name>" );
 368                 }
 369                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 370         }
 371
 372         function in_mediawiki( $parser, $name, $attribs ) {
 373                 $this->debug( "in_mediawiki $name" );
 374                 if( $name == 'siteinfo' ) {
 375                         xml_set_element_handler( $parser, "in_siteinfo", "out_siteinfo" );
 376                 } elseif( $name == 'page' ) {
 377                         xml_set_element_handler( $parser, "in_page", "out_page" );
 378                 } else {
 379                         return $this->throwXMLerror( "Expected <page>, got <$name>" );
 380                 }
 381         }
 382         function out_mediawiki( $parser, $name ) {
 383                 $this->debug( "out_mediawiki $name" );
 384                 if( $name != "mediawiki" ) {
 385                         return $this->throwXMLerror( "Expected </mediawiki>, got </$name>" );
 386                 }
 387                 xml_set_element_handler( $parser, "donothing", "donothing" );
 388         }
 389
 390
 391         function in_siteinfo( $parser, $name, $attribs ) {
 392                 // no-ops for now
 393                 $this->debug( "in_siteinfo $name" );
 394                 switch( $name ) {
 395                 case "sitename":
 396                 case "base":
 397                 case "generator":
 398                 case "case":
 399                 case "namespaces":
 400                 case "namespace":
 401                         break;
 402                 default:
 403                         return $this->throwXMLerror( "Element <$name> not allowed in <siteinfo>." );
 404                 }
 405         }
 406
 407         function out_siteinfo( $parser, $name ) {
 408                 if( $name == "siteinfo" ) {
 409                         xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 410                 }
 411         }
 412
 413
 414         function in_page( $parser, $name, $attribs ) {
 415                 $this->debug( "in_page $name" );
 416                 switch( $name ) {
 417                 case "id":
 418                 case "title":
 419                 case "restrictions":
 420                         $this->appendfield = $name;
 421                         $this->appenddata = "";
 422                         $this->parenttag = "page";
 423                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 424                         xml_set_character_data_handler( $parser, "char_append" );
 425                         break;
 426                 case "revision":
 427                         $this->workRevision = new WikiRevision;
 428                         $this->workRevision->setTitle( $this->workTitle );
 429                         xml_set_element_handler( $parser, "in_revision", "out_revision" );
 430                         break;
 431                 default:
 432                         return $this->throwXMLerror( "Element <$name> not allowed in a <page>." );
 433                 }
 434         }
 435
 436         function out_page( $parser, $name ) {
 437                 $this->debug( "out_page $name" );
 438                 if( $name != "page" ) {
 439                         return $this->throwXMLerror( "Expected </page>, got </$name>" );
 440                 }
 441                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 442
 443                 $this->workTitle = NULL;
 444                 $this->workRevision = NULL;
 445         }
 446
 447         function in_nothing( $parser, $name, $attribs ) {
 448                 $this->debug( "in_nothing $name" );
 449                 return $this->throwXMLerror( "No child elements allowed here; got <$name>" );
 450         }
 451         function char_append( $parser, $data ) {
 452                 $this->debug( "char_append '$data'" );
 453                 $this->appenddata .= $data;
 454         }
 455         function out_append( $parser, $name ) {
 456                 $this->debug( "out_append $name" );
 457                 if( $name != $this->appendfield ) {
 458                         return $this->throwXMLerror( "Expected </{$this->appendfield}>, got </$name>" );
 459                 }
 460                 xml_set_element_handler( $parser, "in_$this->parenttag", "out_$this->parenttag" );
 461                 xml_set_character_data_handler( $parser, "donothing" );
 462
 463                 switch( $this->appendfield ) {
 464                 case "title":
 465                         $this->workTitle = $this->appenddata;
 466                         $this->pageCallback( $this->workTitle );
 467                         break;
 468                 case "text":
 469                         $this->workRevision->setText( $this->appenddata );
 470                         break;
 471                 case "username":
 472                         $this->workRevision->setUsername( $this->appenddata );
 473                         break;
 474                 case "ip":
 475                         $this->workRevision->setUserIP( $this->appenddata );
 476                         break;
 477                 case "timestamp":
 478                         $this->workRevision->setTimestamp( $this->appenddata );
 479                         break;
 480                 case "comment":
 481                         $this->workRevision->setComment( $this->appenddata );
 482                         break;
 483                 case "minor":
 484                         $this->workRevision->setMinor( true );
 485                         break;
 486                 default:
 487                         $this->debug( "Bad append: {$this->appendfield}" );
 488                 }
 489                 $this->appendfield = "";
 490                 $this->appenddata = "";
 491         }
 492
 493         function in_revision( $parser, $name, $attribs ) {
 494                 $this->debug( "in_revision $name" );
 495                 switch( $name ) {
 496                 case "id":
 497                 case "timestamp":
 498                 case "comment":
 499                 case "minor":
 500                 case "text":
 501                         $this->parenttag = "revision";
 502                         $this->appendfield = $name;
 503                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 504                         xml_set_character_data_handler( $parser, "char_append" );
 505                         break;
 506                 case "contributor":
 507                         xml_set_element_handler( $parser, "in_contributor", "out_contributor" );
 508                         break;
 509                 default:
 510                         return $this->throwXMLerror( "Element <$name> not allowed in a <revision>." );
 511                 }
 512         }
 513
 514         function out_revision( $parser, $name ) {
 515                 $this->debug( "out_revision $name" );
 516                 if( $name != "revision" ) {
 517                         return $this->throwXMLerror( "Expected </revision>, got </$name>" );
 518                 }
 519                 xml_set_element_handler( $parser, "in_page", "out_page" );
 520
 521                 $out = call_user_func( $this->mRevisionCallback,
 522                         &$this->workRevision,
 523                         &$this );
 524                 if( !empty( $out ) ) {
 525                         global $wgOut;
 526                         $wgOut->addHTML( "<li>" . $out . "</li>\n" );
 527                 }
 528         }
 529
 530         function in_contributor( $parser, $name, $attribs ) {
 531                 $this->debug( "in_contributor $name" );
 532                 switch( $name ) {
 533                 case "username":
 534                 case "ip":
 535                 case "id":
 536                         $this->parenttag = "contributor";
 537                         $this->appendfield = $name;
 538                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 539                         xml_set_character_data_handler( $parser, "char_append" );
 540                         break;
 541                 default:
 542                         $this->throwXMLerror( "Invalid tag <$name> in <contributor>" );
 543                 }
 544         }
 545
 546         function out_contributor( $parser, $name ) {
 547                 $this->debug( "out_contributor $name" );
 548                 if( $name != "contributor" ) {
 549                         return $this->throwXMLerror( "Expected </contributor>, got </$name>" );
 550                 }
 551                 xml_set_element_handler( $parser, "in_revision", "out_revision" );
 552         }
 553
 554 }
 555
 556 /** @package MediaWiki */
 557 class ImportStringSource {
 558         function ImportStringSource( $string ) {
 559                 $this->mString = $string;
 560                 $this->mRead = false;
 561         }
 562
 563         function atEnd() {
 564                 return $this->mRead;
 565         }
 566
 567         function readChunk() {
 568                 if( $this->atEnd() ) {
 569                         return false;
 570                 } else {
 571                         $this->mRead = true;
 572                         return $this->mString;
 573                 }
 574         }
 575 }
 576
 577 /** @package MediaWiki */
 578 class ImportStreamSource {
 579         function ImportStreamSource( $handle ) {
 580                 $this->mHandle = $handle;
 581         }
 582
 583         function atEnd() {
 584                 return feof( $this->mHandle );
 585         }
 586
 587         function readChunk() {
 588                 return fread( $this->mHandle, 32768 );
 589         }
 590
 591         function newFromFile( $filename ) {
 592                 $file = @fopen( $filename, 'rt' );
 593                 if( !$file ) {
 594                         return new WikiError( "Couldn't open import file" );
 595                 }
 596                 return new ImportStreamSource( $file );
 597         }
 598
 599         function newFromUpload( $fieldname = "xmlimport" ) {
 600                 global $wgOut;
 601
 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 ?>