includes/Export.php

   1 <?php
   2 # Copyright (C) 2003, 2005, 2006 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 # 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18 # http://www.gnu.org/copyleft/gpl.html
  19
  20 /**
  21  * @defgroup Dump Dump
  22  */
  23
  24 /**
  25  * @ingroup SpecialPage Dump
  26  */
  27 class WikiExporter {
  28         var $list_authors = false ; # Return distinct author list (when not returning full history)
  29         var $author_list = "" ;
  30
  31         var $dumpUploads = false;
  32
  33         const FULL = 1;
  34         const CURRENT = 2;
  35         const STABLE = 4; // extension defined
  36         const LOGS = 8;
  37
  38         const BUFFER = 0;
  39         const STREAM = 1;
  40
  41         const TEXT = 0;
  42         const STUB = 1;
  43
  44         /**
  45          * If using WikiExporter::STREAM to stream a large amount of data,
  46          * provide a database connection which is not managed by
  47          * LoadBalancer to read from: some history blob types will
  48          * make additional queries to pull source data while the
  49          * main query is still running.
  50          *
  51          * @param $db Database
  52          * @param $history Mixed: one of WikiExporter::FULL or WikiExporter::CURRENT,
  53          *                 or an associative array:
  54          *                   offset: non-inclusive offset at which to start the query
  55          *                   limit: maximum number of rows to return
  56          *                   dir: "asc" or "desc" timestamp order
  57          * @param $buffer Int: one of WikiExporter::BUFFER or WikiExporter::STREAM
  58          * @param $text Int: one of WikiExporter::TEXT or WikiExporter::STUB
  59          */
  60         function __construct( &$db, $history = WikiExporter::CURRENT,
  61                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  62                 $this->db =& $db;
  63                 $this->history = $history;
  64                 $this->buffer  = $buffer;
  65                 $this->writer  = new XmlDumpWriter();
  66                 $this->sink    = new DumpOutput();
  67                 $this->text    = $text;
  68         }
  69
  70         /**
  71          * Set the DumpOutput or DumpFilter object which will receive
  72          * various row objects and XML output for filtering. Filters
  73          * can be chained or used as callbacks.
  74          *
  75          * @param $sink mixed
  76          */
  77         public function setOutputSink( &$sink ) {
  78                 $this->sink =& $sink;
  79         }
  80
  81         public function openStream() {
  82                 $output = $this->writer->openStream();
  83                 $this->sink->writeOpenStream( $output );
  84         }
  85
  86         public function closeStream() {
  87                 $output = $this->writer->closeStream();
  88                 $this->sink->writeCloseStream( $output );
  89         }
  90
  91         /**
  92          * Dumps a series of page and revision records for all pages
  93          * in the database, either including complete history or only
  94          * the most recent version.
  95          */
  96         public function allPages() {
  97                 return $this->dumpFrom( '' );
  98         }
  99
 100         /**
 101          * Dumps a series of page and revision records for those pages
 102          * in the database falling within the page_id range given.
 103          * @param $start Int: inclusive lower limit (this id is included)
 104          * @param $end   Int: Exclusive upper limit (this id is not included)
 105          *                   If 0, no upper limit.
 106          */
 107         public function pagesByRange( $start, $end ) {
 108                 $condition = 'page_id >= ' . intval( $start );
 109                 if( $end ) {
 110                         $condition .= ' AND page_id < ' . intval( $end );
 111                 }
 112                 return $this->dumpFrom( $condition );
 113         }
 114
 115         /**
 116          * @param $title Title
 117          */
 118         public function pageByTitle( $title ) {
 119                 return $this->dumpFrom(
 120                         'page_namespace=' . $title->getNamespace() .
 121                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 122         }
 123
 124         public function pageByName( $name ) {
 125                 $title = Title::newFromText( $name );
 126                 if( is_null( $title ) ) {
 127                         return new WikiError( "Can't export invalid title" );
 128                 } else {
 129                         return $this->pageByTitle( $title );
 130                 }
 131         }
 132
 133         public function pagesByName( $names ) {
 134                 foreach( $names as $name ) {
 135                         $this->pageByName( $name );
 136                 }
 137         }
 138
 139         public function allLogs() {
 140                 return $this->dumpFrom( '' );
 141         }
 142
 143         public function logsByRange( $start, $end ) {
 144                 $condition = 'log_id >= ' . intval( $start );
 145                 if( $end ) {
 146                         $condition .= ' AND log_id < ' . intval( $end );
 147                 }
 148                 return $this->dumpFrom( $condition );
 149         }
 150
 151         # Generates the distinct list of authors of an article
 152         # Not called by default (depends on $this->list_authors)
 153         # Can be set by Special:Export when not exporting whole history
 154         protected function do_list_authors( $page , $revision , $cond ) {
 155                 $fname = "do_list_authors" ;
 156                 wfProfileIn( $fname );
 157                 $this->author_list = "<contributors>";
 158                 //rev_deleted
 159                 $nothidden = '('.$this->db->bitAnd('rev_deleted', Revision::DELETED_USER) . ') = 0';
 160
 161                 $sql = "SELECT DISTINCT rev_user_text,rev_user FROM {$page},{$revision}
 162                 WHERE page_id=rev_page AND $nothidden AND " . $cond ;
 163                 $result = $this->db->query( $sql, $fname );
 164                 $resultset = $this->db->resultObject( $result );
 165                 while( $row = $resultset->fetchObject() ) {
 166                         $this->author_list .= "<contributor>" .
 167                                 "<username>" .
 168                                 htmlentities( $row->rev_user_text )  .
 169                                 "</username>" .
 170                                 "<id>" .
 171                                 $row->rev_user .
 172                                 "</id>" .
 173                                 "</contributor>";
 174                 }
 175                 wfProfileOut( $fname );
 176                 $this->author_list .= "</contributors>";
 177         }
 178
 179         protected function dumpFrom( $cond = '' ) {
 180                 wfProfileIn( __METHOD__ );
 181                 # For logging dumps...
 182                 if( $this->history & self::LOGS ) {
 183                         if( $this->buffer == WikiExporter::STREAM ) {
 184                                 $prev = $this->db->bufferResults( false );
 185                         }
 186                         $where = array( 'user_id = log_user' );
 187                         # Hide private logs
 188                         $hideLogs = LogEventsList::getExcludeClause( $this->db );
 189                         if( $hideLogs ) $where[] = $hideLogs;
 190                         # Add on any caller specified conditions
 191                         if( $cond ) $where[] = $cond;
 192                         # Get logging table name for logging.* clause
 193                         $logging = $this->db->tableName('logging');
 194                         $result = $this->db->select( array('logging','user'),
 195                                 array( "{$logging}.*", 'user_name' ), // grab the user name
 196                                 $where,
 197                                 __METHOD__,
 198                                 array( 'ORDER BY' => 'log_id', 'USE INDEX' => array('logging' => 'PRIMARY') )
 199                         );
 200                         $wrapper = $this->db->resultObject( $result );
 201                         $this->outputLogStream( $wrapper );
 202                         if( $this->buffer == WikiExporter::STREAM ) {
 203                                 $this->db->bufferResults( $prev );
 204                         }
 205                 # For page dumps...
 206                 } else {
 207                         $tables = array( 'page', 'revision' );
 208                         $opts = array( 'ORDER BY' => 'page_id ASC' );
 209                         $opts['USE INDEX'] = array();
 210                         $join = array();
 211                         if( is_array( $this->history ) ) {
 212                                 # Time offset/limit for all pages/history...
 213                                 $revJoin = 'page_id=rev_page';
 214                                 # Set time order
 215                                 if( $this->history['dir'] == 'asc' ) {
 216                                         $op = '>';
 217                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 218                                 } else {
 219                                         $op = '<';
 220                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 221                                 }
 222                                 # Set offset
 223                                 if( !empty( $this->history['offset'] ) ) {
 224                                         $revJoin .= " AND rev_timestamp $op " .
 225                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 226                                 }
 227                                 $join['revision'] = array('INNER JOIN',$revJoin);
 228                                 # Set query limit
 229                                 if( !empty( $this->history['limit'] ) ) {
 230                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 231                                 }
 232                         } elseif( $this->history & WikiExporter::FULL ) {
 233                                 # Full history dumps...
 234                                 $join['revision'] = array('INNER JOIN','page_id=rev_page');
 235                         } elseif( $this->history & WikiExporter::CURRENT ) {
 236                                 # Latest revision dumps...
 237                                 if( $this->list_authors && $cond != '' )  { // List authors, if so desired
 238                                         list($page,$revision) = $this->db->tableNamesN('page','revision');
 239                                         $this->do_list_authors( $page, $revision, $cond );
 240                                 }
 241                                 $join['revision'] = array('INNER JOIN','page_id=rev_page AND page_latest=rev_id');
 242                         } elseif( $this->history & WikiExporter::STABLE ) {
 243                                 # "Stable" revision dumps...
 244                                 # Default JOIN, to be overridden...
 245                                 $join['revision'] = array('INNER JOIN','page_id=rev_page AND page_latest=rev_id');
 246                                 # One, and only one hook should set this, and return false
 247                                 if( wfRunHooks( 'WikiExporter::dumpStableQuery', array(&$tables,&$opts,&$join) ) ) {
 248                                         wfProfileOut( __METHOD__ );
 249                                         return new WikiError( __METHOD__." given invalid history dump type." );
 250                                 }
 251                         } else {
 252                                 # Uknown history specification parameter?
 253                                 wfProfileOut( __METHOD__ );
 254                                 return new WikiError( __METHOD__." given invalid history dump type." );
 255                         }
 256                         # Query optimization hacks
 257                         if( $cond == '' ) {
 258                                 $opts[] = 'STRAIGHT_JOIN';
 259                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 260                         }
 261                         # Build text join options
 262                         if( $this->text != WikiExporter::STUB ) { // 1-pass
 263                                 $tables[] = 'text';
 264                                 $join['text'] = array('INNER JOIN','rev_text_id=old_id');
 265                         }
 266
 267                         if( $this->buffer == WikiExporter::STREAM ) {
 268                                 $prev = $this->db->bufferResults( false );
 269                         }
 270
 271                         wfRunHooks( 'ModifyExportQuery',
 272                                                 array( $this->db, &$tables, &$cond, &$opts, &$join ) );
 273
 274                         # Do the query!
 275                         $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 276                         $wrapper = $this->db->resultObject( $result );
 277                         # Output dump results
 278                         $this->outputPageStream( $wrapper );
 279                         if( $this->list_authors ) {
 280                                 $this->outputPageStream( $wrapper );
 281                         }
 282
 283                         if( $this->buffer == WikiExporter::STREAM ) {
 284                                 $this->db->bufferResults( $prev );
 285                         }
 286                 }
 287                 wfProfileOut( __METHOD__ );
 288         }
 289
 290         /**
 291          * Runs through a query result set dumping page and revision records.
 292          * The result set should be sorted/grouped by page to avoid duplicate
 293          * page records in the output.
 294          *
 295          * The result set will be freed once complete. Should be safe for
 296          * streaming (non-buffered) queries, as long as it was made on a
 297          * separate database connection not managed by LoadBalancer; some
 298          * blob storage types will make queries to pull source data.
 299          *
 300          * @param $resultset ResultWrapper
 301          */
 302         protected function outputPageStream( $resultset ) {
 303                 $last = null;
 304                 while( $row = $resultset->fetchObject() ) {
 305                         if( is_null( $last ) ||
 306                                 $last->page_namespace != $row->page_namespace ||
 307                                 $last->page_title     != $row->page_title ) {
 308                                 if( isset( $last ) ) {
 309                                         $output = '';
 310                                         if( $this->dumpUploads ) {
 311                                                 $output .= $this->writer->writeUploads( $last );
 312                                         }
 313                                         $output .= $this->writer->closePage();
 314                                         $this->sink->writeClosePage( $output );
 315                                 }
 316                                 $output = $this->writer->openPage( $row );
 317                                 $this->sink->writeOpenPage( $row, $output );
 318                                 $last = $row;
 319                         }
 320                         $output = $this->writer->writeRevision( $row );
 321                         $this->sink->writeRevision( $row, $output );
 322                 }
 323                 if( isset( $last ) ) {
 324                         $output = '';
 325                         if( $this->dumpUploads ) {
 326                                 $output .= $this->writer->writeUploads( $last );
 327                         }
 328                         $output .= $this->author_list;
 329                         $output .= $this->writer->closePage();
 330                         $this->sink->writeClosePage( $output );
 331                 }
 332         }
 333
 334         protected function outputLogStream( $resultset ) {
 335                 while( $row = $resultset->fetchObject() ) {
 336                         $output = $this->writer->writeLogItem( $row );
 337                         $this->sink->writeLogItem( $row, $output );
 338                 }
 339         }
 340 }
 341
 342 /**
 343  * @ingroup Dump
 344  */
 345 class XmlDumpWriter {
 346
 347         /**
 348          * Returns the export schema version.
 349          * @return string
 350          */
 351         function schemaVersion() {
 352                 return "0.4";
 353         }
 354
 355         /**
 356          * Opens the XML output stream's root <mediawiki> element.
 357          * This does not include an xml directive, so is safe to include
 358          * as a subelement in a larger XML stream. Namespace and XML Schema
 359          * references are included.
 360          *
 361          * Output will be encoded in UTF-8.
 362          *
 363          * @return string
 364          */
 365         function openStream() {
 366                 global $wgContLanguageCode;
 367                 $ver = $this->schemaVersion();
 368                 return Xml::element( 'mediawiki', array(
 369                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 370                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 371                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 372                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 373                         'version'            => $ver,
 374                         'xml:lang'           => $wgContLanguageCode ),
 375                         null ) .
 376                         "\n" .
 377                         $this->siteInfo();
 378         }
 379
 380         function siteInfo() {
 381                 $info = array(
 382                         $this->sitename(),
 383                         $this->homelink(),
 384                         $this->generator(),
 385                         $this->caseSetting(),
 386                         $this->namespaces() );
 387                 return "  <siteinfo>\n    " .
 388                         implode( "\n    ", $info ) .
 389                         "\n  </siteinfo>\n";
 390         }
 391
 392         function sitename() {
 393                 global $wgSitename;
 394                 return Xml::element( 'sitename', array(), $wgSitename );
 395         }
 396
 397         function generator() {
 398                 global $wgVersion;
 399                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 400         }
 401
 402         function homelink() {
 403                 return Xml::element( 'base', array(), Title::newMainPage()->getFullUrl() );
 404         }
 405
 406         function caseSetting() {
 407                 global $wgCapitalLinks;
 408                 // "case-insensitive" option is reserved for future
 409                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 410                 return Xml::element( 'case', array(), $sensitivity );
 411         }
 412
 413         function namespaces() {
 414                 global $wgContLang;
 415                 $spaces = "<namespaces>\n";
 416                 foreach( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 417                         $spaces .= '      ' .
 418                                 Xml::element( 'namespace',
 419                                         array(  'key' => $ns,
 420                                                         'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 421                                         ), $title ) . "\n";
 422                 }
 423                 $spaces .= "    </namespaces>";
 424                 return $spaces;
 425         }
 426
 427         /**
 428          * Closes the output stream with the closing root element.
 429          * Call when finished dumping things.
 430          */
 431         function closeStream() {
 432                 return "</mediawiki>\n";
 433         }
 434
 435
 436         /**
 437          * Opens a <page> section on the output stream, with data
 438          * from the given database row.
 439          *
 440          * @param $row object
 441          * @return string
 442          * @access private
 443          */
 444         function openPage( $row ) {
 445                 $out = "  <page>\n";
 446                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 447                 $out .= '    ' . Xml::elementClean( 'title', array(), $title->getPrefixedText() ) . "\n";
 448                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 449                 if( $row->page_is_redirect ) {
 450                         $out .= '    ' . Xml::element( 'redirect', array() ) . "\n";
 451                 }
 452                 if( '' != $row->page_restrictions ) {
 453                         $out .= '    ' . Xml::element( 'restrictions', array(),
 454                                 strval( $row->page_restrictions ) ) . "\n";
 455                 }
 456
 457                 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 458
 459                 return $out;
 460         }
 461
 462         /**
 463          * Closes a <page> section on the output stream.
 464          *
 465          * @access private
 466          */
 467         function closePage() {
 468                 return "  </page>\n";
 469         }
 470
 471         /**
 472          * Dumps a <revision> section on the output stream, with
 473          * data filled in from the given database row.
 474          *
 475          * @param $row object
 476          * @return string
 477          * @access private
 478          */
 479         function writeRevision( $row ) {
 480                 $fname = 'WikiExporter::dumpRev';
 481                 wfProfileIn( $fname );
 482
 483                 $out  = "    <revision>\n";
 484                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 485
 486                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 487
 488                 if( $row->rev_deleted & Revision::DELETED_USER ) {
 489                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 490                 } else {
 491                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 492                 }
 493
 494                 if( $row->rev_minor_edit ) {
 495                         $out .=  "      <minor/>\n";
 496                 }
 497                 if( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 498                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 499                 } elseif( $row->rev_comment != '' ) {
 500                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 501                 }
 502
 503                 if( $row->rev_deleted & Revision::DELETED_TEXT ) {
 504                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 505                 } elseif( isset( $row->old_text ) ) {
 506                         // Raw text from the database may have invalid chars
 507                         $text = strval( Revision::getRevisionText( $row ) );
 508                         $out .= "      " . Xml::elementClean( 'text',
 509                                 array( 'xml:space' => 'preserve' ),
 510                                 strval( $text ) ) . "\n";
 511                 } else {
 512                         // Stub output
 513                         $out .= "      " . Xml::element( 'text',
 514                                 array( 'id' => $row->rev_text_id ),
 515                                 "" ) . "\n";
 516                 }
 517
 518                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 519
 520                 $out .= "    </revision>\n";
 521
 522                 wfProfileOut( $fname );
 523                 return $out;
 524         }
 525
 526         /**
 527          * Dumps a <logitem> section on the output stream, with
 528          * data filled in from the given database row.
 529          *
 530          * @param $row object
 531          * @return string
 532          * @access private
 533          */
 534         function writeLogItem( $row ) {
 535                 $fname = 'WikiExporter::writeLogItem';
 536                 wfProfileIn( $fname );
 537
 538                 $out  = "    <logitem>\n";
 539                 $out .= "      " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 540
 541                 $out .= $this->writeTimestamp( $row->log_timestamp );
 542
 543                 if( $row->log_deleted & LogPage::DELETED_USER ) {
 544                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 545                 } else {
 546                         $out .= $this->writeContributor( $row->log_user, $row->user_name );
 547                 }
 548
 549                 if( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 550                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 551                 } elseif( $row->log_comment != '' ) {
 552                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 553                 }
 554
 555                 $out .= "      " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 556                 $out .= "      " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 557
 558                 if( $row->log_deleted & LogPage::DELETED_ACTION ) {
 559                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 560                 } else {
 561                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 562                         $out .= "      " . Xml::elementClean( 'logtitle', null, $title->getPrefixedText() ) . "\n";
 563                         $out .= "      " . Xml::elementClean( 'params',
 564                                 array( 'xml:space' => 'preserve' ),
 565                                 strval( $row->log_params ) ) . "\n";
 566                 }
 567
 568                 $out .= "    </logitem>\n";
 569
 570                 wfProfileOut( $fname );
 571                 return $out;
 572         }
 573
 574         function writeTimestamp( $timestamp ) {
 575                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 576                 return "      " . Xml::element( 'timestamp', null, $ts ) . "\n";
 577         }
 578
 579         function writeContributor( $id, $text ) {
 580                 $out = "      <contributor>\n";
 581                 if( $id ) {
 582                         $out .= "        " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 583                         $out .= "        " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 584                 } else {
 585                         $out .= "        " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 586                 }
 587                 $out .= "      </contributor>\n";
 588                 return $out;
 589         }
 590
 591         /**
 592          * Warning! This data is potentially inconsistent. :(
 593          */
 594         function writeUploads( $row ) {
 595                 if( $row->page_namespace == NS_IMAGE ) {
 596                         $img = wfFindFile( $row->page_title );
 597                         if( $img ) {
 598                                 $out = '';
 599                                 foreach( array_reverse( $img->getHistory() ) as $ver ) {
 600                                         $out .= $this->writeUpload( $ver );
 601                                 }
 602                                 $out .= $this->writeUpload( $img );
 603                                 return $out;
 604                         }
 605                 }
 606                 return '';
 607         }
 608
 609         function writeUpload( $file ) {
 610                 return "    <upload>\n" .
 611                         $this->writeTimestamp( $file->getTimestamp() ) .
 612                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 613                         "      " . Xml::elementClean( 'comment', null, $file->getDescription() ) . "\n" .
 614                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 615                         "      " . Xml::element( 'src', null, $file->getFullUrl() ) . "\n" .
 616                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 617                         "    </upload>\n";
 618         }
 619
 620 }
 621
 622
 623 /**
 624  * Base class for output stream; prints to stdout or buffer or whereever.
 625  * @ingroup Dump
 626  */
 627 class DumpOutput {
 628         function writeOpenStream( $string ) {
 629                 $this->write( $string );
 630         }
 631
 632         function writeCloseStream( $string ) {
 633                 $this->write( $string );
 634         }
 635
 636         function writeOpenPage( $page, $string ) {
 637                 $this->write( $string );
 638         }
 639
 640         function writeClosePage( $string ) {
 641                 $this->write( $string );
 642         }
 643
 644         function writeRevision( $rev, $string ) {
 645                 $this->write( $string );
 646         }
 647
 648         function writeLogItem( $rev, $string ) {
 649                 $this->write( $string );
 650         }
 651
 652         /**
 653          * Override to write to a different stream type.
 654          * @return bool
 655          */
 656         function write( $string ) {
 657                 print $string;
 658         }
 659 }
 660
 661 /**
 662  * Stream outputter to send data to a file.
 663  * @ingroup Dump
 664  */
 665 class DumpFileOutput extends DumpOutput {
 666         var $handle;
 667
 668         function DumpFileOutput( $file ) {
 669                 $this->handle = fopen( $file, "wt" );
 670         }
 671
 672         function write( $string ) {
 673                 fputs( $this->handle, $string );
 674         }
 675 }
 676
 677 /**
 678  * Stream outputter to send data to a file via some filter program.
 679  * Even if compression is available in a library, using a separate
 680  * program can allow us to make use of a multi-processor system.
 681  * @ingroup Dump
 682  */
 683 class DumpPipeOutput extends DumpFileOutput {
 684         function DumpPipeOutput( $command, $file = null ) {
 685                 if( !is_null( $file ) ) {
 686                         $command .=  " > " . wfEscapeShellArg( $file );
 687                 }
 688                 $this->handle = popen( $command, "w" );
 689         }
 690 }
 691
 692 /**
 693  * Sends dump output via the gzip compressor.
 694  * @ingroup Dump
 695  */
 696 class DumpGZipOutput extends DumpPipeOutput {
 697         function DumpGZipOutput( $file ) {
 698                 parent::DumpPipeOutput( "gzip", $file );
 699         }
 700 }
 701
 702 /**
 703  * Sends dump output via the bgzip2 compressor.
 704  * @ingroup Dump
 705  */
 706 class DumpBZip2Output extends DumpPipeOutput {
 707         function DumpBZip2Output( $file ) {
 708                 parent::DumpPipeOutput( "bzip2", $file );
 709         }
 710 }
 711
 712 /**
 713  * Sends dump output via the p7zip compressor.
 714  * @ingroup Dump
 715  */
 716 class Dump7ZipOutput extends DumpPipeOutput {
 717         function Dump7ZipOutput( $file ) {
 718                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 719                 // Suppress annoying useless crap from p7zip
 720                 // Unfortunately this could suppress real error messages too
 721                 $command .= ' >' . wfGetNull() . ' 2>&1';
 722                 parent::DumpPipeOutput( $command );
 723         }
 724 }
 725
 726
 727
 728 /**
 729  * Dump output filter class.
 730  * This just does output filtering and streaming; XML formatting is done
 731  * higher up, so be careful in what you do.
 732  * @ingroup Dump
 733  */
 734 class DumpFilter {
 735         function DumpFilter( &$sink ) {
 736                 $this->sink =& $sink;
 737         }
 738
 739         function writeOpenStream( $string ) {
 740                 $this->sink->writeOpenStream( $string );
 741         }
 742
 743         function writeCloseStream( $string ) {
 744                 $this->sink->writeCloseStream( $string );
 745         }
 746
 747         function writeOpenPage( $page, $string ) {
 748                 $this->sendingThisPage = $this->pass( $page, $string );
 749                 if( $this->sendingThisPage ) {
 750                         $this->sink->writeOpenPage( $page, $string );
 751                 }
 752         }
 753
 754         function writeClosePage( $string ) {
 755                 if( $this->sendingThisPage ) {
 756                         $this->sink->writeClosePage( $string );
 757                         $this->sendingThisPage = false;
 758                 }
 759         }
 760
 761         function writeRevision( $rev, $string ) {
 762                 if( $this->sendingThisPage ) {
 763                         $this->sink->writeRevision( $rev, $string );
 764                 }
 765         }
 766
 767         function writeLogItem( $rev, $string ) {
 768                 $this->sink->writeRevision( $rev, $string );
 769         }
 770
 771         /**
 772          * Override for page-based filter types.
 773          * @return bool
 774          */
 775         function pass( $page ) {
 776                 return true;
 777         }
 778 }
 779
 780 /**
 781  * Simple dump output filter to exclude all talk pages.
 782  * @ingroup Dump
 783  */
 784 class DumpNotalkFilter extends DumpFilter {
 785         function pass( $page ) {
 786                 return !MWNamespace::isTalk( $page->page_namespace );
 787         }
 788 }
 789
 790 /**
 791  * Dump output filter to include or exclude pages in a given set of namespaces.
 792  * @ingroup Dump
 793  */
 794 class DumpNamespaceFilter extends DumpFilter {
 795         var $invert = false;
 796         var $namespaces = array();
 797
 798         function DumpNamespaceFilter( &$sink, $param ) {
 799                 parent::DumpFilter( $sink );
 800
 801                 $constants = array(
 802                         "NS_MAIN"           => NS_MAIN,
 803                         "NS_TALK"           => NS_TALK,
 804                         "NS_USER"           => NS_USER,
 805                         "NS_USER_TALK"      => NS_USER_TALK,
 806                         "NS_PROJECT"        => NS_PROJECT,
 807                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
 808                         "NS_FILE"           => NS_FILE,
 809                         "NS_FILE_TALK"      => NS_FILE_TALK,
 810                         "NS_IMAGE"          => NS_IMAGE,  // NS_IMAGE is an alias for NS_FILE
 811                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
 812                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
 813                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
 814                         "NS_TEMPLATE"       => NS_TEMPLATE,
 815                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
 816                         "NS_HELP"           => NS_HELP,
 817                         "NS_HELP_TALK"      => NS_HELP_TALK,
 818                         "NS_CATEGORY"       => NS_CATEGORY,
 819                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
 820
 821                 if( $param{0} == '!' ) {
 822                         $this->invert = true;
 823                         $param = substr( $param, 1 );
 824                 }
 825
 826                 foreach( explode( ',', $param ) as $key ) {
 827                         $key = trim( $key );
 828                         if( isset( $constants[$key] ) ) {
 829                                 $ns = $constants[$key];
 830                                 $this->namespaces[$ns] = true;
 831                         } elseif( is_numeric( $key ) ) {
 832                                 $ns = intval( $key );
 833                                 $this->namespaces[$ns] = true;
 834                         } else {
 835                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
 836                         }
 837                 }
 838         }
 839
 840         function pass( $page ) {
 841                 $match = isset( $this->namespaces[$page->page_namespace] );
 842                 return $this->invert xor $match;
 843         }
 844 }
 845
 846
 847 /**
 848  * Dump output filter to include only the last revision in each page sequence.
 849  * @ingroup Dump
 850  */
 851 class DumpLatestFilter extends DumpFilter {
 852         var $page, $pageString, $rev, $revString;
 853
 854         function writeOpenPage( $page, $string ) {
 855                 $this->page = $page;
 856                 $this->pageString = $string;
 857         }
 858
 859         function writeClosePage( $string ) {
 860                 if( $this->rev ) {
 861                         $this->sink->writeOpenPage( $this->page, $this->pageString );
 862                         $this->sink->writeRevision( $this->rev, $this->revString );
 863                         $this->sink->writeClosePage( $string );
 864                 }
 865                 $this->rev = null;
 866                 $this->revString = null;
 867                 $this->page = null;
 868                 $this->pageString = null;
 869         }
 870
 871         function writeRevision( $rev, $string ) {
 872                 if( $rev->rev_id == $this->page->page_latest ) {
 873                         $this->rev = $rev;
 874                         $this->revString = $string;
 875                 }
 876         }
 877 }
 878
 879 /**
 880  * Base class for output stream; prints to stdout or buffer or whereever.
 881  * @ingroup Dump
 882  */
 883 class DumpMultiWriter {
 884         function DumpMultiWriter( $sinks ) {
 885                 $this->sinks = $sinks;
 886                 $this->count = count( $sinks );
 887         }
 888
 889         function writeOpenStream( $string ) {
 890                 for( $i = 0; $i < $this->count; $i++ ) {
 891                         $this->sinks[$i]->writeOpenStream( $string );
 892                 }
 893         }
 894
 895         function writeCloseStream( $string ) {
 896                 for( $i = 0; $i < $this->count; $i++ ) {
 897                         $this->sinks[$i]->writeCloseStream( $string );
 898                 }
 899         }
 900
 901         function writeOpenPage( $page, $string ) {
 902                 for( $i = 0; $i < $this->count; $i++ ) {
 903                         $this->sinks[$i]->writeOpenPage( $page, $string );
 904                 }
 905         }
 906
 907         function writeClosePage( $string ) {
 908                 for( $i = 0; $i < $this->count; $i++ ) {
 909                         $this->sinks[$i]->writeClosePage( $string );
 910                 }
 911         }
 912
 913         function writeRevision( $rev, $string ) {
 914                 for( $i = 0; $i < $this->count; $i++ ) {
 915                         $this->sinks[$i]->writeRevision( $rev, $string );
 916                 }
 917         }
 918 }
 919
 920 function xmlsafe( $string ) {
 921         $fname = 'xmlsafe';
 922         wfProfileIn( $fname );
 923
 924         /**
 925          * The page may contain old data which has not been properly normalized.
 926          * Invalid UTF-8 sequences or forbidden control characters will make our
 927          * XML output invalid, so be sure to strip them out.
 928          */
 929         $string = UtfNormal::cleanUp( $string );
 930
 931         $string = htmlspecialchars( $string );
 932         wfProfileOut( $fname );
 933         return $string;
 934 }