includes/SpecialImport.php

   1 <?php
   2 # Copyright (C) 2003 Brion Vibber <brion@pobox.com>
   3 # http://www.mediawiki.org/
   4 #
   5 # This program is free software; you can redistribute it and/or modify
   6 # it under the terms of the GNU General Public License as published by
   7 # the Free Software Foundation; either version 2 of the License, or
   8 # (at your option) any later version.
   9 #
  10 # This program is distributed in the hope that it will be useful,
  11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13 # GNU General Public License for more details.
  14 #
  15 # You should have received a copy of the GNU General Public License along
  16 # with this program; if not, write to the Free Software Foundation, Inc.,
  17 # 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
  18 # http://www.gnu.org/copyleft/gpl.html
  19
  20 function wfSpecialImport( $page = "" ) {
  21         global $wgOut, $wgLang, $wgRequest, $wgTitle;
  22         global $wgImportSources;
  23
  24         ###
  25         $wgOut->addWikiText( "Special:Import is not ready for this beta release, sorry." );
  26         return;
  27         ###
  28
  29         if( $wgRequest->wasPosted() && $wgRequest->getVal( 'action' ) == 'submit') {
  30                 $importer = new WikiImporter();
  31
  32                 switch( $wgRequest->getVal( "source" ) ) {
  33                 case "upload":
  34                         $ok = $importer->setupFromUpload( "xmlimport" );
  35                         break;
  36                 case "interwiki":
  37                         $ok = $importer->setupFromInterwiki(
  38                                 $wgRequest->getVal( "interwiki" ),
  39                                 $wgRequest->getText( "frompage" ) );
  40                         break;
  41                 default:
  42                         $ok = false;
  43                 }
  44
  45                 if( $ok ) {
  46                         $importer->setRevisionHandler( "wfImportOldRevision" );
  47                         if( $importer->doImport() ) {
  48                                 # Success!
  49                                 $wgOut->addHTML( "<p>" . wfMsg( "importsuccess" ) . "</p>" );
  50                         } else {
  51                                 $wgOut->addHTML( "<p>" . wfMsg( "importfailed",
  52                                         htmlspecialchars( $importer->getError() ) ) . "</p>" );
  53                         }
  54                 } else {
  55                         $wgOut->addWikiText( htmlspecialchars( $importer->getError() ) );
  56                 }
  57         }
  58
  59         $wgOut->addWikiText( "<p>" . wfMsg( "importtext" ) . "</p>" );
  60         $action = $wgTitle->escapeLocalUrl();
  61         $wgOut->addHTML( "
  62 <fieldset>
  63         <legend>Upload XML</legend>
  64         <form enctype='multipart/form-data' method='post' action=\"$action\">
  65                 <input type='hidden' name='action' value='submit' />
  66                 <input type='hidden' name='source' value='upload' />
  67                 <input type='hidden' name='MAX_FILE_SIZE' value='200000' />
  68                 <input type='file' name='xmlimport' value='' size='30' />
  69                 <input type='submit' value='" . htmlspecialchars( wfMsg( "uploadbtn" ) ) . "'/>
  70         </form>
  71 </fieldset>
  72 " );
  73
  74         if( !empty( $wgImportSources ) ) {
  75                 $wgOut->addHTML( "
  76 <fieldset>
  77         <legend>Interwiki import</legend>
  78         <form method='post' action=\"$action\">
  79                 <input type='hidden' name='action' value='submit' />
  80                 <input type='hidden' name='source' value='interwiki' />
  81                 <select name='interwiki'>
  82 " );
  83                 foreach( $wgImportSources as $interwiki ) {
  84                         $iw = htmlspecialchars( $interwiki );
  85                         $wgOut->addHTML( "<option value=\"$iw\">$iw</option>\n" );
  86                 }
  87                 $wgOut->addHTML( "
  88                 </select>
  89                 <input name='frompage' />
  90                 <input type='submit' />
  91         </form>
  92 </fieldset>
  93 " );
  94         }
  95 }
  96
  97 function wfImportOldRevision( &$revision ) {
  98         global $wgOut;
  99         $fname = "wfImportOldRevision";
 100
 101         # Sneak a single revision into place
 102         $ns = IntVal( $revision->title->getNamespace() );
 103         $t = wfStrencode( $revision->title->getDBkey() );
 104         $text = wfStrencode( $revision->getText() );
 105         $ts = wfStrencode( $revision->timestamp );
 106         $its = wfStrencode( wfInvertTimestamp( $revision->timestamp ) ) ;
 107         $comment = wfStrencode( $revision->getComment() );
 108
 109         $user = User::newFromName( $revision->getUser() );
 110         $user_id = IntVal( $user->getId() );
 111         $user_text = wfStrencode( $user->getName() );
 112
 113         $minor = 0; # ??
 114         $flags = "";
 115
 116         # Make sure it doesn't already exist
 117         $sql = "SELECT 1 FROM old WHERE old_namespace=$ns AND old_title='$t' AND old_timestamp='$ts'";
 118         $res = wfQuery( $sql, DB_WRITE, $fname );
 119         $numrows = wfNumRows( $res );
 120         wfFreeResult( $res );
 121         if( $numrows > 0 ) {
 122                 return wfMsg( "importhistoryconflict" );
 123         }
 124
 125         $res = wfQuery( "INSERT INTO old " .
 126           "(old_namespace,old_title,old_text,old_comment,old_user,old_user_text," .
 127           "old_timestamp,inverse_timestamp,old_minor_edit,old_flags) " .
 128           "VALUES ($ns,'$t','$text','$comment',$user_id,'$user_text','$ts','$its',$minor,'$flags')",
 129           DB_WRITE, $fname );
 130
 131         return wfMsg( "ok" );
 132 }
 133
 134 class WikiRevision {
 135         var $title = NULL;
 136         var $timestamp = "20010115000000";
 137         var $user = 0;
 138         var $user_text = "";
 139         var $text = "";
 140         var $comment = "";
 141
 142         function setTitle( $text ) {
 143                 $text = $this->fixEncoding( $text );
 144                 $this->title = Title::newFromText( $text );
 145         }
 146
 147         function setTimestamp( $ts ) {
 148                 # 2003-08-05T18:30:02Z
 149                 $this->timestamp = preg_replace( '/^(....)-(..)-(..)T(..):(..):(..)Z$/', '$1$2$3$4$5$6', $ts );
 150         }
 151
 152         function setUsername( $user ) {
 153                 $this->user_text = $this->fixEncoding( $user );
 154         }
 155
 156         function setUserIP( $ip ) {
 157                 $this->user_text = $this->fixEncoding( $ip );
 158         }
 159
 160         function setText( $text ) {
 161                 $this->text = $this->fixEncoding( $text );
 162         }
 163
 164         function setComment( $text ) {
 165                 $this->comment = $this->fixEncoding( $text );
 166         }
 167
 168         function fixEncoding( $data ) {
 169                 global $wgLang, $wgInputEncoding;
 170
 171                 if( strcasecmp( $wgInputEncoding, "utf-8" ) == 0 ) {
 172                         return $data;
 173                 } else {
 174                         return $wgLang->iconv( "utf-8", $wgInputEncoding, $data );
 175                 }
 176         }
 177
 178         function getTitle() {
 179                 return $this->title;
 180         }
 181
 182         function getTimestamp() {
 183                 return $this->timestamp;
 184         }
 185
 186         function getUser() {
 187                 return $this->user_text;
 188         }
 189
 190         function getText() {
 191                 return $this->text;
 192         }
 193
 194         function getComment() {
 195                 return $this->comment;
 196         }
 197 }
 198
 199 class WikiImporter {
 200         var $mSource = NULL;
 201         var $mError = "";
 202         var $mXmlError = XML_ERROR_NONE;
 203         var $mRevisionHandler = NULL;
 204         var $lastfield;
 205
 206         function WikiImporter() {
 207                 $this->setRevisionHandler( array( &$this, "defaultRevisionHandler" ) );
 208         }
 209
 210         function setError( $err ) {
 211                 $this->mError = $err;
 212                 return false;
 213         }
 214
 215         function getError() {
 216                 if( $this->mXmlError == XML_ERROR_NONE ) {
 217                         return $this->mError;
 218                 } else {
 219                         return xml_error_string( $this->mXmlError );
 220                 }
 221         }
 222
 223         function throwXmlError( $err ) {
 224                 $this->debug( "FAILURE: $err" );
 225         }
 226
 227         function setupFromFile( $filename ) {
 228                 $this->mSource = file_get_contents( $filename );
 229                 return true;
 230         }
 231
 232         function setupFromUpload( $fieldname = "xmlimport" ) {
 233                 global $wgOut;
 234
 235                 $upload =& $_FILES[$fieldname];
 236
 237                 if( !isset( $upload ) ) {
 238                         return $this->setError( wfMsg( "importnofile" ) );
 239                 }
 240                 if( !empty( $upload['error'] ) ) {
 241                         return $this->setError( wfMsg( "importuploaderror", $upload['error'] ) );
 242                 }
 243                 $fname = $upload['tmp_name'];
 244                 if( is_uploaded_file( $fname ) ) {
 245                         return $this->setupFromFile( $fname );
 246                 } else {
 247                         return $this->setError( wfMsg( "importnofile" ) );
 248                 }
 249         }
 250
 251         function setupFromURL( $url ) {
 252                 # fopen-wrappers are normally turned off for security.
 253                 ini_set( "allow_url_fopen", true );
 254                 $ret = $this->setupFromFile( $url );
 255                 ini_set( "allow_url_fopen", false );
 256                 return $ret;
 257         }
 258
 259         function setupFromInterwiki( $interwiki, $page ) {
 260                 $base = Title::getInterwikiLink( $interwiki );
 261                 if( empty( $base ) ) {
 262                         return false;
 263                 } else {
 264                         $import = wfUrlencode( "Special:Export/$page" );
 265                         $url = str_replace( "$1", $import, $base );
 266                         $this->notice( "Importing from $url" );
 267                         return $this->setupFromURL( $url );
 268                 }
 269         }
 270
 271         # --------------
 272
 273         function doImport() {
 274                 if( empty( $this->mSource ) ) {
 275                         return $this->setError( wfMsg( "importnotext" ) );
 276                 }
 277
 278                 $parser = xml_parser_create( "UTF-8" );
 279
 280                 # case folding violates XML standard, turn it off
 281                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
 282
 283                 xml_set_object( $parser, &$this );
 284                 xml_set_element_handler( $parser, "in_start", "" );
 285
 286                 if( !xml_parse( $parser, $this->mSource, true ) ) {
 287                         # return error message
 288                         $this->mXmlError = xml_get_error_code( $parser );
 289                         xml_parser_free( $parser );
 290                         return false;
 291                 }
 292                 xml_parser_free( $parser );
 293
 294                 return true;
 295         }
 296
 297         function debug( $data ) {
 298                 global $wgOut;
 299                 # $this->notice( "DEBUG: $data\n" );
 300         }
 301
 302         function notice( $data ) {
 303                 global $wgCommandLineMode;
 304                 if( $wgCommandLineMode ) {
 305                         print "$data\n";
 306                 } else {
 307                         global $wgOut;
 308                         $wgOut->addHTML( "<li>$data</li>\n" );
 309                 }
 310         }
 311
 312         function setRevisionHandler( $functionref ) {
 313                 $this->mRevisionHandler = $functionref;
 314         }
 315
 316         function defaultRevisionHandler( &$revision ) {
 317                 $this->debug( "Got revision:" );
 318                 if( is_object( $revision->title ) ) {
 319                         $this->debug( "-- Title: " . $revision->title->getPrefixedText() );
 320                 } else {
 321                         $this->debug( "-- Title: <invalid>" );
 322                 }
 323                 $this->debug( "-- User: " . $revision->user_text );
 324                 $this->debug( "-- Timestamp: " . $revision->timestamp );
 325                 $this->debug( "-- Comment: " . $revision->comment );
 326                 $this->debug( "-- Text: " . $revision->text );
 327         }
 328
 329
 330
 331         # XML parser callbacks from here out -- beware!
 332         function donothing( $parser, $x, $y="" ) {
 333                 #$this->debug( "donothing" );
 334         }
 335
 336         function in_start( $parser, $name, $attribs ) {
 337                 $this->debug( "in_start $name" );
 338                 if( $name != "mediawiki" ) {
 339                         return $this->throwXMLerror( "Expected <mediawiki>, got <$name>" );
 340                 }
 341                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 342         }
 343
 344         function in_mediawiki( $parser, $name, $attribs ) {
 345                 $this->debug( "in_mediawiki $name" );
 346                 if( $name != "page" ) {
 347                         return $this->throwXMLerror( "Expected <page>, got <$name>" );
 348                 }
 349                 xml_set_element_handler( $parser, "in_page", "out_page" );
 350         }
 351         function out_mediawiki( $parser, $name ) {
 352                 $this->debug( "out_mediawiki $name" );
 353                 if( $name != "mediawiki" ) {
 354                         return $this->throwXMLerror( "Expected </mediawiki>, got </$name>" );
 355                 }
 356                 xml_set_element_handler( $parser, "donothing", "donothing" );
 357         }
 358
 359         function in_page( $parser, $name, $attribs ) {
 360                 $this->debug( "in_page $name" );
 361                 switch( $name ) {
 362                 case "id":
 363                 case "title":
 364                 case "restrictions":
 365                         $this->appendfield = $name;
 366                         $this->appenddata = "";
 367                         $this->parenttag = "page";
 368                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 369                         xml_set_character_data_handler( $parser, "char_append" );
 370                         break;
 371                 case "revision":
 372                         $this->workRevision = new WikiRevision;
 373                         $this->workRevision->setTitle( $this->workTitle );
 374                         xml_set_element_handler( $parser, "in_revision", "out_revision" );
 375                         break;
 376                 default:
 377                         return $this->throwXMLerror( "Element <$name> not allowed in a <page>." );
 378                 }
 379         }
 380
 381         function out_page( $parser, $name ) {
 382                 $this->debug( "out_page $name" );
 383                 if( $name != "page" ) {
 384                         return $this->throwXMLerror( "Expected </page>, got </$name>" );
 385                 }
 386                 xml_set_element_handler( $parser, "in_mediawiki", "out_mediawiki" );
 387
 388                 $this->workTitle = NULL;
 389                 $this->workRevision = NULL;
 390         }
 391
 392         function in_nothing( $parser, $name, $attribs ) {
 393                 $this->debug( "in_nothing $name" );
 394                 return $this->throwXMLerror( "No child elements allowed here; got <$name>" );
 395         }
 396         function char_append( $parser, $data ) {
 397                 $this->debug( "char_append '$data'" );
 398                 $this->appenddata .= $data;
 399         }
 400         function out_append( $parser, $name ) {
 401                 $this->debug( "out_append $name" );
 402                 if( $name != $this->appendfield ) {
 403                         return $this->throwXMLerror( "Expected </{$this->appendfield}>, got </$name>" );
 404                 }
 405                 xml_set_element_handler( $parser, "in_$this->parenttag", "out_$this->parenttag" );
 406                 xml_set_character_data_handler( $parser, "donothing" );
 407                 switch( $this->appendfield ) {
 408                 case "title":
 409                         $this->workTitle = $this->appenddata;
 410                         break;
 411                 case "text":
 412                         $this->workRevision->setText( $this->appenddata );
 413                         break;
 414                 case "username":
 415                         $this->workRevision->setUsername( $this->appenddata );
 416                         break;
 417                 case "ip":
 418                         $this->workRevision->setUserIP( $this->appenddata );
 419                         break;
 420                 case "timestamp":
 421                         $this->workRevision->setTimestamp( $this->appenddata );
 422                         break;
 423                 case "comment":
 424                         $this->workRevision->setComment( $this->appenddata );
 425                         break;
 426                 default;
 427                         $this->debug( "Bad append: {$this->appendfield}" );
 428                 }
 429                 $this->appendfield = "";
 430                 $this->appenddata = "";
 431         }
 432
 433         function in_revision( $parser, $name, $attribs ) {
 434                 $this->debug( "in_revision $name" );
 435                 switch( $name ) {
 436                 case "id":
 437                 case "timestamp":
 438                 case "comment":
 439                 case "text":
 440                         $this->parenttag = "revision";
 441                         $this->appendfield = $name;
 442                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 443                         xml_set_character_data_handler( $parser, "char_append" );
 444                         break;
 445                 case "contributor":
 446                         xml_set_element_handler( $parser, "in_contributor", "out_contributor" );
 447                         break;
 448                 default:
 449                         return $this->throwXMLerror( "Element <$name> not allowed in a <revision>." );
 450                 }
 451         }
 452
 453         function out_revision( $parser, $name ) {
 454                 $this->debug( "out_revision $name" );
 455                 if( $name != "revision" ) {
 456                         return $this->throwXMLerror( "Expected </revision>, got </$name>" );
 457                 }
 458                 xml_set_element_handler( $parser, "in_page", "out_page" );
 459
 460                 $out = call_user_func( $this->mRevisionHandler, &$this->workRevision, &$this );
 461                 if( !empty( $out ) ) {
 462                         global $wgOut;
 463                         $wgOut->addHTML( "<li>" . $out . "</li>\n" );
 464                 }
 465         }
 466
 467         function in_contributor( $parser, $name, $attribs ) {
 468                 $this->debug( "in_contributor $name" );
 469                 switch( $name ) {
 470                 case "username":
 471                 case "ip":
 472                         $this->parenttag = "contributor";
 473                         $this->appendfield = $name;
 474                         xml_set_element_handler( $parser, "in_nothing", "out_append" );
 475                         xml_set_character_data_handler( $parser, "char_append" );
 476                         break;
 477                 default:
 478                         $this->throwXMLerror( "Invalid tag <$name> in <contributor>" );
 479                 }
 480         }
 481
 482         function out_contributor( $parser, $name ) {
 483                 $this->debug( "out_contributor $name" );
 484                 if( $name != "contributor" ) {
 485                         return $this->throwXMLerror( "Expected </contributor>, got </$name>" );
 486                 }
 487                 xml_set_element_handler( $parser, "in_revision", "out_revision" );
 488         }
 489 }
 490
 491
 492 ?>