includes/Export.php

   1 <?php
   2 /**
   3  * Base classes for dumps and export
   4  *
   5  * Copyright © 2003, 2005, 2006 Brion Vibber <brion@pobox.com>
   6  * http://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  */
  25
  26 /**
  27  * @defgroup Dump Dump
  28  */
  29
  30 /**
  31  * @ingroup SpecialPage Dump
  32  */
  33 class WikiExporter {
  34         var $list_authors = false ; # Return distinct author list (when not returning full history)
  35         var $author_list = "" ;
  36
  37         var $dumpUploads = false;
  38         var $dumpUploadFileContents = false;
  39
  40         const FULL = 1;
  41         const CURRENT = 2;
  42         const STABLE = 4; // extension defined
  43         const LOGS = 8;
  44         const RANGE = 16;
  45
  46         const BUFFER = 0;
  47         const STREAM = 1;
  48
  49         const TEXT = 0;
  50         const STUB = 1;
  51
  52         var $buffer;
  53
  54         var $text;
  55
  56         /**
  57          * @var DumpOutput
  58          */
  59         var $sink;
  60
  61         /**
  62          * If using WikiExporter::STREAM to stream a large amount of data,
  63          * provide a database connection which is not managed by
  64          * LoadBalancer to read from: some history blob types will
  65          * make additional queries to pull source data while the
  66          * main query is still running.
  67          *
  68          * @param $db DatabaseBase
  69          * @param $history Mixed: one of WikiExporter::FULL, WikiExporter::CURRENT,
  70          *                 WikiExporter::RANGE or WikiExporter::STABLE,
  71          *                 or an associative array:
  72          *                   offset: non-inclusive offset at which to start the query
  73          *                   limit: maximum number of rows to return
  74          *                   dir: "asc" or "desc" timestamp order
  75          * @param $buffer Int: one of WikiExporter::BUFFER or WikiExporter::STREAM
  76          * @param $text Int: one of WikiExporter::TEXT or WikiExporter::STUB
  77          */
  78         function __construct( &$db, $history = WikiExporter::CURRENT,
  79                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  80                 $this->db =& $db;
  81                 $this->history = $history;
  82                 $this->buffer  = $buffer;
  83                 $this->writer  = new XmlDumpWriter();
  84                 $this->sink    = new DumpOutput();
  85                 $this->text    = $text;
  86         }
  87
  88         /**
  89          * Set the DumpOutput or DumpFilter object which will receive
  90          * various row objects and XML output for filtering. Filters
  91          * can be chained or used as callbacks.
  92          *
  93          * @param $sink mixed
  94          */
  95         public function setOutputSink( &$sink ) {
  96                 $this->sink =& $sink;
  97         }
  98
  99         public function openStream() {
 100                 $output = $this->writer->openStream();
 101                 $this->sink->writeOpenStream( $output );
 102         }
 103
 104         public function closeStream() {
 105                 $output = $this->writer->closeStream();
 106                 $this->sink->writeCloseStream( $output );
 107         }
 108
 109         /**
 110          * Dumps a series of page and revision records for all pages
 111          * in the database, either including complete history or only
 112          * the most recent version.
 113          */
 114         public function allPages() {
 115                 $this->dumpFrom( '' );
 116         }
 117
 118         /**
 119          * Dumps a series of page and revision records for those pages
 120          * in the database falling within the page_id range given.
 121          * @param $start Int: inclusive lower limit (this id is included)
 122          * @param $end   Int: Exclusive upper limit (this id is not included)
 123          *                   If 0, no upper limit.
 124          */
 125         public function pagesByRange( $start, $end ) {
 126                 $condition = 'page_id >= ' . intval( $start );
 127                 if ( $end ) {
 128                         $condition .= ' AND page_id < ' . intval( $end );
 129                 }
 130                 $this->dumpFrom( $condition );
 131         }
 132
 133         /**
 134          * Dumps a series of page and revision records for those pages
 135          * in the database with revisions falling within the rev_id range given.
 136          * @param $start Int: inclusive lower limit (this id is included)
 137          * @param $end   Int: Exclusive upper limit (this id is not included)
 138          *                   If 0, no upper limit.
 139          */
 140         public function revsByRange( $start, $end ) {
 141                 $condition = 'rev_id >= ' . intval( $start );
 142                 if ( $end ) {
 143                         $condition .= ' AND rev_id < ' . intval( $end );
 144                 }
 145                 $this->dumpFrom( $condition );
 146         }
 147
 148         /**
 149          * @param $title Title
 150          */
 151         public function pageByTitle( $title ) {
 152                 $this->dumpFrom(
 153                         'page_namespace=' . $title->getNamespace() .
 154                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 155         }
 156
 157         /**
 158          * @param $name string
 159          * @throws MWException
 160          */
 161         public function pageByName( $name ) {
 162                 $title = Title::newFromText( $name );
 163                 if ( is_null( $title ) ) {
 164                         throw new MWException( "Can't export invalid title" );
 165                 } else {
 166                         $this->pageByTitle( $title );
 167                 }
 168         }
 169
 170         /**
 171          * @param $names array
 172          */
 173         public function pagesByName( $names ) {
 174                 foreach ( $names as $name ) {
 175                         $this->pageByName( $name );
 176                 }
 177         }
 178
 179         public function allLogs() {
 180                 $this->dumpFrom( '' );
 181         }
 182
 183         /**
 184          * @param $start int
 185          * @param $end int
 186          */
 187         public function logsByRange( $start, $end ) {
 188                 $condition = 'log_id >= ' . intval( $start );
 189                 if ( $end ) {
 190                         $condition .= ' AND log_id < ' . intval( $end );
 191                 }
 192                 $this->dumpFrom( $condition );
 193         }
 194
 195         /**
 196          * Generates the distinct list of authors of an article
 197          * Not called by default (depends on $this->list_authors)
 198          * Can be set by Special:Export when not exporting whole history
 199          *
 200          * @param $cond
 201          */
 202         protected function do_list_authors( $cond ) {
 203                 wfProfileIn( __METHOD__ );
 204                 $this->author_list = "<contributors>";
 205                 // rev_deleted
 206
 207                 $res = $this->db->select(
 208                         array( 'page', 'revision' ),
 209                         array( 'DISTINCT rev_user_text', 'rev_user' ),
 210                         array(
 211                                 $this->db->bitAnd( 'rev_deleted', Revision::DELETED_USER ) . ' = 0',
 212                                 $cond,
 213                                 'page_id = rev_id',
 214                         ),
 215                         __METHOD__
 216                 );
 217
 218                 foreach ( $res as $row ) {
 219                         $this->author_list .= "<contributor>" .
 220                                 "<username>" .
 221                                 htmlentities( $row->rev_user_text )  .
 222                                 "</username>" .
 223                                 "<id>" .
 224                                 $row->rev_user .
 225                                 "</id>" .
 226                                 "</contributor>";
 227                 }
 228                 $this->author_list .= "</contributors>";
 229                 wfProfileOut( __METHOD__ );
 230         }
 231
 232         /**
 233          * @param $cond string
 234          * @throws MWException
 235          * @throws Exception
 236          */
 237         protected function dumpFrom( $cond = '' ) {
 238                 wfProfileIn( __METHOD__ );
 239                 # For logging dumps...
 240                 if ( $this->history & self::LOGS ) {
 241                         $where = array( 'user_id = log_user' );
 242                         # Hide private logs
 243                         $hideLogs = LogEventsList::getExcludeClause( $this->db );
 244                         if ( $hideLogs ) $where[] = $hideLogs;
 245                         # Add on any caller specified conditions
 246                         if ( $cond ) $where[] = $cond;
 247                         # Get logging table name for logging.* clause
 248                         $logging = $this->db->tableName( 'logging' );
 249
 250                         if ( $this->buffer == WikiExporter::STREAM ) {
 251                                 $prev = $this->db->bufferResults( false );
 252                         }
 253                         $wrapper = null; // Assuring $wrapper is not undefined, if exception occurs early
 254                         try {
 255                                 $result = $this->db->select( array( 'logging', 'user' ),
 256                                         array( "{$logging}.*", 'user_name' ), // grab the user name
 257                                         $where,
 258                                         __METHOD__,
 259                                         array( 'ORDER BY' => 'log_id', 'USE INDEX' => array( 'logging' => 'PRIMARY' ) )
 260                                 );
 261                                 $wrapper = $this->db->resultObject( $result );
 262                                 $this->outputLogStream( $wrapper );
 263                                 if ( $this->buffer == WikiExporter::STREAM ) {
 264                                         $this->db->bufferResults( $prev );
 265                                 }
 266                         } catch ( Exception $e ) {
 267                                 // Throwing the exception does not reliably free the resultset, and
 268                                 // would also leave the connection in unbuffered mode.
 269
 270                                 // Freeing result
 271                                 try {
 272                                         if ( $wrapper ) {
 273                                                 $wrapper->free();
 274                                         }
 275                                 } catch ( Exception $e2 ) {
 276                                         // Already in panic mode -> ignoring $e2 as $e has
 277                                         // higher priority
 278                                 }
 279
 280                                 // Putting database back in previous buffer mode
 281                                 try {
 282                                         if ( $this->buffer == WikiExporter::STREAM ) {
 283                                                 $this->db->bufferResults( $prev );
 284                                         }
 285                                 } catch ( Exception $e2 ) {
 286                                         // Already in panic mode -> ignoring $e2 as $e has
 287                                         // higher priority
 288                                 }
 289
 290                                 // Inform caller about problem
 291                                 throw $e;
 292                         }
 293                 # For page dumps...
 294                 } else {
 295                         $tables = array( 'page', 'revision' );
 296                         $opts = array( 'ORDER BY' => 'page_id ASC' );
 297                         $opts['USE INDEX'] = array();
 298                         $join = array();
 299                         if ( is_array( $this->history ) ) {
 300                                 # Time offset/limit for all pages/history...
 301                                 $revJoin = 'page_id=rev_page';
 302                                 # Set time order
 303                                 if ( $this->history['dir'] == 'asc' ) {
 304                                         $op = '>';
 305                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 306                                 } else {
 307                                         $op = '<';
 308                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 309                                 }
 310                                 # Set offset
 311                                 if ( !empty( $this->history['offset'] ) ) {
 312                                         $revJoin .= " AND rev_timestamp $op " .
 313                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 314                                 }
 315                                 $join['revision'] = array( 'INNER JOIN', $revJoin );
 316                                 # Set query limit
 317                                 if ( !empty( $this->history['limit'] ) ) {
 318                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 319                                 }
 320                         } elseif ( $this->history & WikiExporter::FULL ) {
 321                                 # Full history dumps...
 322                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 323                         } elseif ( $this->history & WikiExporter::CURRENT ) {
 324                                 # Latest revision dumps...
 325                                 if ( $this->list_authors && $cond != '' )  { // List authors, if so desired
 326                                         $this->do_list_authors( $cond );
 327                                 }
 328                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 329                         } elseif ( $this->history & WikiExporter::STABLE ) {
 330                                 # "Stable" revision dumps...
 331                                 # Default JOIN, to be overridden...
 332                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 333                                 # One, and only one hook should set this, and return false
 334                                 if ( wfRunHooks( 'WikiExporter::dumpStableQuery', array( &$tables, &$opts, &$join ) ) ) {
 335                                         wfProfileOut( __METHOD__ );
 336                                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 337                                 }
 338                         } elseif ( $this->history & WikiExporter::RANGE ) {
 339                                 # Dump of revisions within a specified range
 340                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 341                                 $opts['ORDER BY'] = array( 'rev_page ASC', 'rev_id ASC' );
 342                         } else {
 343                                 # Uknown history specification parameter?
 344                                 wfProfileOut( __METHOD__ );
 345                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 346                         }
 347                         # Query optimization hacks
 348                         if ( $cond == '' ) {
 349                                 $opts[] = 'STRAIGHT_JOIN';
 350                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 351                         }
 352                         # Build text join options
 353                         if ( $this->text != WikiExporter::STUB ) { // 1-pass
 354                                 $tables[] = 'text';
 355                                 $join['text'] = array( 'INNER JOIN', 'rev_text_id=old_id' );
 356                         }
 357
 358                         if ( $this->buffer == WikiExporter::STREAM ) {
 359                                 $prev = $this->db->bufferResults( false );
 360                         }
 361
 362                         $wrapper = null; // Assuring $wrapper is not undefined, if exception occurs early
 363                         try {
 364                                 wfRunHooks( 'ModifyExportQuery',
 365                                                 array( $this->db, &$tables, &$cond, &$opts, &$join ) );
 366
 367                                 # Do the query!
 368                                 $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 369                                 $wrapper = $this->db->resultObject( $result );
 370                                 # Output dump results
 371                                 $this->outputPageStream( $wrapper );
 372
 373                                 if ( $this->buffer == WikiExporter::STREAM ) {
 374                                         $this->db->bufferResults( $prev );
 375                                 }
 376                         } catch ( Exception $e ) {
 377                                 // Throwing the exception does not reliably free the resultset, and
 378                                 // would also leave the connection in unbuffered mode.
 379
 380                                 // Freeing result
 381                                 try {
 382                                         if ( $wrapper ) {
 383                                                 $wrapper->free();
 384                                         }
 385                                 } catch ( Exception $e2 ) {
 386                                         // Already in panic mode -> ignoring $e2 as $e has
 387                                         // higher priority
 388                                 }
 389
 390                                 // Putting database back in previous buffer mode
 391                                 try {
 392                                         if ( $this->buffer == WikiExporter::STREAM ) {
 393                                                 $this->db->bufferResults( $prev );
 394                                         }
 395                                 } catch ( Exception $e2 ) {
 396                                         // Already in panic mode -> ignoring $e2 as $e has
 397                                         // higher priority
 398                                 }
 399
 400                                 // Inform caller about problem
 401                                 throw $e;
 402                         }
 403                 }
 404                 wfProfileOut( __METHOD__ );
 405         }
 406
 407         /**
 408          * Runs through a query result set dumping page and revision records.
 409          * The result set should be sorted/grouped by page to avoid duplicate
 410          * page records in the output.
 411          *
 412          * Should be safe for
 413          * streaming (non-buffered) queries, as long as it was made on a
 414          * separate database connection not managed by LoadBalancer; some
 415          * blob storage types will make queries to pull source data.
 416          *
 417          * @param $resultset ResultWrapper
 418          */
 419         protected function outputPageStream( $resultset ) {
 420                 $last = null;
 421                 foreach ( $resultset as $row ) {
 422                         if ( is_null( $last ) ||
 423                                 $last->page_namespace != $row->page_namespace ||
 424                                 $last->page_title     != $row->page_title ) {
 425                                 if ( isset( $last ) ) {
 426                                         $output = '';
 427                                         if ( $this->dumpUploads ) {
 428                                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 429                                         }
 430                                         $output .= $this->writer->closePage();
 431                                         $this->sink->writeClosePage( $output );
 432                                 }
 433                                 $output = $this->writer->openPage( $row );
 434                                 $this->sink->writeOpenPage( $row, $output );
 435                                 $last = $row;
 436                         }
 437                         $output = $this->writer->writeRevision( $row );
 438                         $this->sink->writeRevision( $row, $output );
 439                 }
 440                 if ( isset( $last ) ) {
 441                         $output = '';
 442                         if ( $this->dumpUploads ) {
 443                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 444                         }
 445                         $output .= $this->author_list;
 446                         $output .= $this->writer->closePage();
 447                         $this->sink->writeClosePage( $output );
 448                 }
 449         }
 450
 451         /**
 452          * @param $resultset array
 453          */
 454         protected function outputLogStream( $resultset ) {
 455                 foreach ( $resultset as $row ) {
 456                         $output = $this->writer->writeLogItem( $row );
 457                         $this->sink->writeLogItem( $row, $output );
 458                 }
 459         }
 460 }
 461
 462 /**
 463  * @ingroup Dump
 464  */
 465 class XmlDumpWriter {
 466         /**
 467          * Returns the export schema version.
 468          * @return string
 469          */
 470         function schemaVersion() {
 471                 return "0.8"; #FIXME: Make sure to bump this to > 0.7 when merging Wikidata branch!
 472         }
 473
 474         /**
 475          * Opens the XML output stream's root <mediawiki> element.
 476          * This does not include an xml directive, so is safe to include
 477          * as a subelement in a larger XML stream. Namespace and XML Schema
 478          * references are included.
 479          *
 480          * Output will be encoded in UTF-8.
 481          *
 482          * @return string
 483          */
 484         function openStream() {
 485                 global $wgLanguageCode;
 486                 $ver = $this->schemaVersion();
 487                 return Xml::element( 'mediawiki', array(
 488                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 489                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 490                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 491                                                 "http://www.mediawiki.org/xml/export-$ver.xsd", #TODO: how do we get a new version up there?
 492                         'version'            => $ver,
 493                         'xml:lang'           => $wgLanguageCode ),
 494                         null ) .
 495                         "\n" .
 496                         $this->siteInfo();
 497         }
 498
 499         /**
 500          * @return string
 501          */
 502         function siteInfo() {
 503                 $info = array(
 504                         $this->sitename(),
 505                         $this->homelink(),
 506                         $this->generator(),
 507                         $this->caseSetting(),
 508                         $this->namespaces() );
 509                 return "  <siteinfo>\n    " .
 510                         implode( "\n    ", $info ) .
 511                         "\n  </siteinfo>\n";
 512         }
 513
 514         /**
 515          * @return string
 516          */
 517         function sitename() {
 518                 global $wgSitename;
 519                 return Xml::element( 'sitename', array(), $wgSitename );
 520         }
 521
 522         /**
 523          * @return string
 524          */
 525         function generator() {
 526                 global $wgVersion;
 527                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 528         }
 529
 530         /**
 531          * @return string
 532          */
 533         function homelink() {
 534                 return Xml::element( 'base', array(), Title::newMainPage()->getCanonicalUrl() );
 535         }
 536
 537         /**
 538          * @return string
 539          */
 540         function caseSetting() {
 541                 global $wgCapitalLinks;
 542                 // "case-insensitive" option is reserved for future
 543                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 544                 return Xml::element( 'case', array(), $sensitivity );
 545         }
 546
 547         /**
 548          * @return string
 549          */
 550         function namespaces() {
 551                 global $wgContLang;
 552                 $spaces = "<namespaces>\n";
 553                 foreach ( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 554                         $spaces .= '      ' .
 555                                 Xml::element( 'namespace',
 556                                         array(  'key' => $ns,
 557                                                         'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 558                                         ), $title ) . "\n";
 559                 }
 560                 $spaces .= "    </namespaces>";
 561                 return $spaces;
 562         }
 563
 564         /**
 565          * Closes the output stream with the closing root element.
 566          * Call when finished dumping things.
 567          *
 568          * @return string
 569          */
 570         function closeStream() {
 571                 return "</mediawiki>\n";
 572         }
 573
 574         /**
 575          * Opens a <page> section on the output stream, with data
 576          * from the given database row.
 577          *
 578          * @param $row object
 579          * @return string
 580          * @access private
 581          */
 582         function openPage( $row ) {
 583                 $out = "  <page>\n";
 584                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 585                 $out .= '    ' . Xml::elementClean( 'title', array(), self::canonicalTitle( $title ) ) . "\n";
 586                 $out .= '    ' . Xml::element( 'ns', array(), strval( $row->page_namespace) ) . "\n";
 587                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 588                 if ( $row->page_is_redirect ) {
 589                         $page = WikiPage::factory( $title );
 590                         $redirect = $page->getRedirectTarget();
 591                         if ( $redirect instanceOf Title && $redirect->isValidRedirectTarget() ) {
 592                                 $out .= '    ' . Xml::element( 'redirect', array( 'title' => self::canonicalTitle( $redirect ) ) ) . "\n";
 593                         }
 594                 }
 595
 596                 if ( $row->page_restrictions != '' ) {
 597                         $out .= '    ' . Xml::element( 'restrictions', array(),
 598                                 strval( $row->page_restrictions ) ) . "\n";
 599                 }
 600
 601                 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 602
 603                 return $out;
 604         }
 605
 606         /**
 607          * Closes a <page> section on the output stream.
 608          *
 609          * @access private
 610          * @return string
 611          */
 612         function closePage() {
 613                 return "  </page>\n";
 614         }
 615
 616         /**
 617          * Dumps a <revision> section on the output stream, with
 618          * data filled in from the given database row.
 619          *
 620          * @param $row object
 621          * @return string
 622          * @access private
 623          */
 624         function writeRevision( $row ) {
 625                 wfProfileIn( __METHOD__ );
 626
 627                 $out  = "    <revision>\n";
 628                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 629                 if( $row->rev_parent_id ) {
 630                         $out .= "      " . Xml::element( 'parentid', null, strval( $row->rev_parent_id ) ) . "\n";
 631                 }
 632
 633                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 634
 635                 if ( $row->rev_deleted & Revision::DELETED_USER ) {
 636                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 637                 } else {
 638                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 639                 }
 640
 641                 if ( $row->rev_minor_edit ) {
 642                         $out .=  "      <minor/>\n";
 643                 }
 644                 if ( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 645                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 646                 } elseif ( $row->rev_comment != '' ) {
 647                         $out .= "      " . Xml::elementClean( 'comment', array(), strval( $row->rev_comment ) ) . "\n";
 648                 }
 649
 650                 $text = '';
 651                 if ( $row->rev_deleted & Revision::DELETED_TEXT ) {
 652                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 653                 } elseif ( isset( $row->old_text ) ) {
 654                         // Raw text from the database may have invalid chars
 655                         $text = strval( Revision::getRevisionText( $row ) );
 656                         $out .= "      " . Xml::elementClean( 'text',
 657                                 array( 'xml:space' => 'preserve', 'bytes' => intval( $row->rev_len ) ),
 658                                 strval( $text ) ) . "\n";
 659                 } else {
 660                         // Stub output
 661                         $out .= "      " . Xml::element( 'text',
 662                                 array( 'id' => $row->rev_text_id, 'bytes' => intval( $row->rev_len ) ),
 663                                 "" ) . "\n";
 664                 }
 665
 666                 if ( $row->rev_sha1 && !( $row->rev_deleted & Revision::DELETED_TEXT ) ) {
 667                         $out .= "      " . Xml::element('sha1', null, strval( $row->rev_sha1 ) ) . "\n";
 668                 } else {
 669                         $out .= "      <sha1/>\n";
 670                 }
 671
 672                 if ( isset( $row->rev_content_model ) && !is_null( $row->rev_content_model )  ) {
 673                         $content_model = strval( $row->rev_content_model );
 674                 } else {
 675                         // probably using $wgContentHandlerUseDB = false;
 676                         // @todo: test!
 677                         $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 678                         $content_model = ContentHandler::getDefaultModelFor( $title );
 679                 }
 680
 681                 $out .= "      " . Xml::element('model', null, strval( $content_model ) ) . "\n";
 682
 683                 if ( isset( $row->rev_content_format ) && !is_null( $row->rev_content_format ) ) {
 684                         $content_format = strval( $row->rev_content_format );
 685                 } else {
 686                         // probably using $wgContentHandlerUseDB = false;
 687                         // @todo: test!
 688                         $content_handler = ContentHandler::getForModelID( $content_model );
 689                         $content_format = $content_handler->getDefaultFormat();
 690                 }
 691
 692                 $out .= "      " . Xml::element('format', null, strval( $content_format ) ) . "\n";
 693
 694                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 695
 696                 $out .= "    </revision>\n";
 697
 698                 wfProfileOut( __METHOD__ );
 699                 return $out;
 700         }
 701
 702         /**
 703          * Dumps a <logitem> section on the output stream, with
 704          * data filled in from the given database row.
 705          *
 706          * @param $row object
 707          * @return string
 708          * @access private
 709          */
 710         function writeLogItem( $row ) {
 711                 wfProfileIn( __METHOD__ );
 712
 713                 $out  = "  <logitem>\n";
 714                 $out .= "    " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 715
 716                 $out .= $this->writeTimestamp( $row->log_timestamp, "    " );
 717
 718                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 719                         $out .= "    " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 720                 } else {
 721                         $out .= $this->writeContributor( $row->log_user, $row->user_name, "    " );
 722                 }
 723
 724                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 725                         $out .= "    " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 726                 } elseif ( $row->log_comment != '' ) {
 727                         $out .= "    " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 728                 }
 729
 730                 $out .= "    " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 731                 $out .= "    " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 732
 733                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 734                         $out .= "    " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 735                 } else {
 736                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 737                         $out .= "    " . Xml::elementClean( 'logtitle', null, self::canonicalTitle( $title ) ) . "\n";
 738                         $out .= "    " . Xml::elementClean( 'params',
 739                                 array( 'xml:space' => 'preserve' ),
 740                                 strval( $row->log_params ) ) . "\n";
 741                 }
 742
 743                 $out .= "  </logitem>\n";
 744
 745                 wfProfileOut( __METHOD__ );
 746                 return $out;
 747         }
 748
 749         /**
 750          * @param $timestamp string
 751          * @return string
 752          */
 753         function writeTimestamp( $timestamp, $indent = "      " ) {
 754                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 755                 return $indent . Xml::element( 'timestamp', null, $ts ) . "\n";
 756         }
 757
 758         /**
 759          * @param $id
 760          * @param $text string
 761          * @return string
 762          */
 763         function writeContributor( $id, $text, $indent = "      " ) {
 764                 $out = $indent . "<contributor>\n";
 765                 if ( $id || !IP::isValid( $text ) ) {
 766                         $out .= $indent . "  " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 767                         $out .= $indent . "  " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 768                 } else {
 769                         $out .= $indent . "  " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 770                 }
 771                 $out .= $indent . "</contributor>\n";
 772                 return $out;
 773         }
 774
 775         /**
 776          * Warning! This data is potentially inconsistent. :(
 777          * @param $row
 778          * @param $dumpContents bool
 779          * @return string
 780          */
 781         function writeUploads( $row, $dumpContents = false ) {
 782                 if ( $row->page_namespace == NS_IMAGE ) {
 783                         $img = wfLocalFile( $row->page_title );
 784                         if ( $img && $img->exists() ) {
 785                                 $out = '';
 786                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 787                                         $out .= $this->writeUpload( $ver, $dumpContents );
 788                                 }
 789                                 $out .= $this->writeUpload( $img, $dumpContents );
 790                                 return $out;
 791                         }
 792                 }
 793                 return '';
 794         }
 795
 796         /**
 797          * @param $file File
 798          * @param $dumpContents bool
 799          * @return string
 800          */
 801         function writeUpload( $file, $dumpContents = false ) {
 802                 if ( $file->isOld() ) {
 803                         $archiveName = "      " .
 804                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 805                 } else {
 806                         $archiveName = '';
 807                 }
 808                 if ( $dumpContents ) {
 809                         # Dump file as base64
 810                         # Uses only XML-safe characters, so does not need escaping
 811                         $contents = '      <contents encoding="base64">' .
 812                                 chunk_split( base64_encode( file_get_contents( $file->getPath() ) ) ) .
 813                                 "      </contents>\n";
 814                 } else {
 815                         $contents = '';
 816                 }
 817                 if ( $file->isDeleted( File::DELETED_COMMENT ) ) {
 818                         $comment = Xml::element( 'comment', array( 'deleted' => 'deleted' ) );
 819                 } else {
 820                         $comment = Xml::elementClean( 'comment', null, $file->getDescription() );
 821                 }
 822                 return "    <upload>\n" .
 823                         $this->writeTimestamp( $file->getTimestamp() ) .
 824                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 825                         "      " . $comment . "\n" .
 826                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 827                         $archiveName .
 828                         "      " . Xml::element( 'src', null, $file->getCanonicalUrl() ) . "\n" .
 829                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 830                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 831                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 832                         $contents .
 833                         "    </upload>\n";
 834         }
 835
 836         /**
 837          * Return prefixed text form of title, but using the content language's
 838          * canonical namespace. This skips any special-casing such as gendered
 839          * user namespaces -- which while useful, are not yet listed in the
 840          * XML <siteinfo> data so are unsafe in export.
 841          *
 842          * @param Title $title
 843          * @return string
 844          * @since 1.18
 845          */
 846         public static function canonicalTitle( Title $title ) {
 847                 if ( $title->getInterwiki() ) {
 848                         return $title->getPrefixedText();
 849                 }
 850
 851                 global $wgContLang;
 852                 $prefix = str_replace( '_', ' ', $wgContLang->getNsText( $title->getNamespace() ) );
 853
 854                 if ( $prefix !== '' ) {
 855                         $prefix .= ':';
 856                 }
 857
 858                 return $prefix . $title->getText();
 859         }
 860 }
 861
 862
 863 /**
 864  * Base class for output stream; prints to stdout or buffer or whereever.
 865  * @ingroup Dump
 866  */
 867 class DumpOutput {
 868
 869         /**
 870          * @param $string string
 871          */
 872         function writeOpenStream( $string ) {
 873                 $this->write( $string );
 874         }
 875
 876         /**
 877          * @param $string string
 878          */
 879         function writeCloseStream( $string ) {
 880                 $this->write( $string );
 881         }
 882
 883         /**
 884          * @param $page
 885          * @param $string string
 886          */
 887         function writeOpenPage( $page, $string ) {
 888                 $this->write( $string );
 889         }
 890
 891         /**
 892          * @param $string string
 893          */
 894         function writeClosePage( $string ) {
 895                 $this->write( $string );
 896         }
 897
 898         /**
 899          * @param $rev
 900          * @param $string string
 901          */
 902         function writeRevision( $rev, $string ) {
 903                 $this->write( $string );
 904         }
 905
 906         /**
 907          * @param $rev
 908          * @param $string string
 909          */
 910         function writeLogItem( $rev, $string ) {
 911                 $this->write( $string );
 912         }
 913
 914         /**
 915          * Override to write to a different stream type.
 916          * @param $string string
 917          * @return bool
 918          */
 919         function write( $string ) {
 920                 print $string;
 921         }
 922
 923         /**
 924          * Close the old file, move it to a specified name,
 925          * and reopen new file with the old name. Use this
 926          * for writing out a file in multiple pieces
 927          * at specified checkpoints (e.g. every n hours).
 928          * @param $newname mixed File name. May be a string or an array with one element
 929          */
 930         function closeRenameAndReopen( $newname ) {
 931                 return;
 932         }
 933
 934         /**
 935          * Close the old file, and move it to a specified name.
 936          * Use this for the last piece of a file written out
 937          * at specified checkpoints (e.g. every n hours).
 938          * @param $newname mixed File name. May be a string or an array with one element
 939          * @param $open bool If true, a new file with the old filename will be opened again for writing (default: false)
 940          */
 941         function closeAndRename( $newname, $open = false ) {
 942                 return;
 943         }
 944
 945         /**
 946          * Returns the name of the file or files which are
 947          * being written to, if there are any.
 948          * @return null
 949          */
 950         function getFilenames() {
 951                 return NULL;
 952         }
 953 }
 954
 955 /**
 956  * Stream outputter to send data to a file.
 957  * @ingroup Dump
 958  */
 959 class DumpFileOutput extends DumpOutput {
 960         protected $handle = false, $filename;
 961
 962         /**
 963          * @param $file
 964          */
 965         function __construct( $file ) {
 966                 $this->handle = fopen( $file, "wt" );
 967                 $this->filename = $file;
 968         }
 969
 970         /**
 971          * @param $string string
 972          */
 973         function writeCloseStream( $string ) {
 974                 parent::writeCloseStream( $string );
 975                 if ( $this->handle ) {
 976                         fclose( $this->handle );
 977                         $this->handle = false;
 978                 }
 979         }
 980
 981         /**
 982          * @param $string string
 983          */
 984         function write( $string ) {
 985                 fputs( $this->handle, $string );
 986         }
 987
 988         /**
 989          * @param $newname
 990          */
 991         function closeRenameAndReopen( $newname ) {
 992                 $this->closeAndRename( $newname, true );
 993         }
 994
 995         /**
 996          * @param $newname
 997          * @throws MWException
 998          */
 999         function renameOrException( $newname ) {
1000                         if (! rename( $this->filename, $newname ) ) {
1001                                 throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" );
1002                         }
1003         }
1004
1005         /**
1006          * @param $newname array
1007          * @return mixed
1008          * @throws MWException
1009          */
1010         function checkRenameArgCount( $newname ) {
1011                 if ( is_array( $newname ) ) {
1012                         if ( count( $newname ) > 1 ) {
1013                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
1014                         } else {
1015                                 $newname = $newname[0];
1016                         }
1017                 }
1018                 return $newname;
1019         }
1020
1021         /**
1022          * @param $newname mixed
1023          * @param $open bool
1024          */
1025         function closeAndRename( $newname, $open = false ) {
1026                 $newname = $this->checkRenameArgCount( $newname );
1027                 if ( $newname ) {
1028                         if ( $this->handle ) {
1029                                 fclose( $this->handle );
1030                                 $this->handle = false;
1031                         }
1032                         $this->renameOrException( $newname );
1033                         if ( $open ) {
1034                                 $this->handle = fopen( $this->filename, "wt" );
1035                         }
1036                 }
1037         }
1038
1039         /**
1040          * @return string|null
1041          */
1042         function getFilenames() {
1043                 return $this->filename;
1044         }
1045 }
1046
1047 /**
1048  * Stream outputter to send data to a file via some filter program.
1049  * Even if compression is available in a library, using a separate
1050  * program can allow us to make use of a multi-processor system.
1051  * @ingroup Dump
1052  */
1053 class DumpPipeOutput extends DumpFileOutput {
1054         protected $command, $filename;
1055         protected $procOpenResource = false;
1056
1057         /**
1058          * @param $command
1059          * @param $file null
1060          */
1061         function __construct( $command, $file = null ) {
1062                 if ( !is_null( $file ) ) {
1063                         $command .=  " > " . wfEscapeShellArg( $file );
1064                 }
1065
1066                 $this->startCommand( $command );
1067                 $this->command = $command;
1068                 $this->filename = $file;
1069         }
1070
1071         /**
1072          * @param $string string
1073          */
1074         function writeCloseStream( $string ) {
1075                 parent::writeCloseStream( $string );
1076                 if ( $this->procOpenResource ) {
1077                         proc_close( $this->procOpenResource );
1078                         $this->procOpenResource = false;
1079                 }
1080         }
1081
1082         /**
1083          * @param $command
1084          */
1085         function startCommand( $command ) {
1086                 $spec = array(
1087                         0 => array( "pipe", "r" ),
1088                 );
1089                 $pipes = array();
1090                 $this->procOpenResource = proc_open( $command, $spec, $pipes );
1091                 $this->handle = $pipes[0];
1092         }
1093
1094         /**
1095          * @param mixed $newname
1096          */
1097         function closeRenameAndReopen( $newname ) {
1098                 $this->closeAndRename( $newname, true );
1099         }
1100
1101         /**
1102          * @param $newname mixed
1103          * @param $open bool
1104          */
1105         function closeAndRename( $newname, $open = false ) {
1106                 $newname = $this->checkRenameArgCount( $newname );
1107                 if ( $newname ) {
1108                         if ( $this->handle ) {
1109                                 fclose( $this->handle );
1110                                 $this->handle = false;
1111                         }
1112                         if ( $this->procOpenResource ) {
1113                                 proc_close( $this->procOpenResource );
1114                                 $this->procOpenResource = false;
1115                         }
1116                         $this->renameOrException( $newname );
1117                         if ( $open ) {
1118                                 $command = $this->command;
1119                                 $command .=  " > " . wfEscapeShellArg( $this->filename );
1120                                 $this->startCommand( $command );
1121                         }
1122                 }
1123         }
1124
1125 }
1126
1127 /**
1128  * Sends dump output via the gzip compressor.
1129  * @ingroup Dump
1130  */
1131 class DumpGZipOutput extends DumpPipeOutput {
1132
1133         /**
1134          * @param $file string
1135          */
1136         function __construct( $file ) {
1137                 parent::__construct( "gzip", $file );
1138         }
1139 }
1140
1141 /**
1142  * Sends dump output via the bgzip2 compressor.
1143  * @ingroup Dump
1144  */
1145 class DumpBZip2Output extends DumpPipeOutput {
1146
1147         /**
1148          * @param $file string
1149          */
1150         function __construct( $file ) {
1151                 parent::__construct( "bzip2", $file );
1152         }
1153 }
1154
1155 /**
1156  * Sends dump output via the p7zip compressor.
1157  * @ingroup Dump
1158  */
1159 class Dump7ZipOutput extends DumpPipeOutput {
1160
1161         /**
1162          * @param $file string
1163          */
1164         function __construct( $file ) {
1165                 $command = $this->setup7zCommand( $file );
1166                 parent::__construct( $command );
1167                 $this->filename = $file;
1168         }
1169
1170         /**
1171          * @param $file string
1172          * @return string
1173          */
1174         function setup7zCommand( $file ) {
1175                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
1176                 // Suppress annoying useless crap from p7zip
1177                 // Unfortunately this could suppress real error messages too
1178                 $command .= ' >' . wfGetNull() . ' 2>&1';
1179                 return( $command );
1180         }
1181
1182         /**
1183          * @param $newname string
1184          * @param $open bool
1185          */
1186         function closeAndRename( $newname, $open = false ) {
1187                 $newname = $this->checkRenameArgCount( $newname );
1188                 if ( $newname ) {
1189                         fclose( $this->handle );
1190                         proc_close( $this->procOpenResource );
1191                         $this->renameOrException( $newname );
1192                         if ( $open ) {
1193                                 $command = $this->setup7zCommand( $this->filename );
1194                                 $this->startCommand( $command );
1195                         }
1196                 }
1197         }
1198 }
1199
1200 /**
1201  * Dump output filter class.
1202  * This just does output filtering and streaming; XML formatting is done
1203  * higher up, so be careful in what you do.
1204  * @ingroup Dump
1205  */
1206 class DumpFilter {
1207
1208         /**
1209          * @var DumpOutput
1210          * FIXME will need to be made protected whenever legacy code
1211          * is updated.
1212          */
1213         public $sink;
1214
1215         /**
1216          * @var bool
1217          */
1218         protected $sendingThisPage;
1219
1220         /**
1221          * @param $sink DumpOutput
1222          */
1223         function __construct( &$sink ) {
1224                 $this->sink =& $sink;
1225         }
1226
1227         /**
1228          * @param $string string
1229          */
1230         function writeOpenStream( $string ) {
1231                 $this->sink->writeOpenStream( $string );
1232         }
1233
1234         /**
1235          * @param $string string
1236          */
1237         function writeCloseStream( $string ) {
1238                 $this->sink->writeCloseStream( $string );
1239         }
1240
1241         /**
1242          * @param $page
1243          * @param $string string
1244          */
1245         function writeOpenPage( $page, $string ) {
1246                 $this->sendingThisPage = $this->pass( $page, $string );
1247                 if ( $this->sendingThisPage ) {
1248                         $this->sink->writeOpenPage( $page, $string );
1249                 }
1250         }
1251
1252         /**
1253          * @param $string string
1254          */
1255         function writeClosePage( $string ) {
1256                 if ( $this->sendingThisPage ) {
1257                         $this->sink->writeClosePage( $string );
1258                         $this->sendingThisPage = false;
1259                 }
1260         }
1261
1262         /**
1263          * @param $rev
1264          * @param $string string
1265          */
1266         function writeRevision( $rev, $string ) {
1267                 if ( $this->sendingThisPage ) {
1268                         $this->sink->writeRevision( $rev, $string );
1269                 }
1270         }
1271
1272         /**
1273          * @param $rev
1274          * @param $string string
1275          */
1276         function writeLogItem( $rev, $string ) {
1277                 $this->sink->writeRevision( $rev, $string );
1278         }
1279
1280         /**
1281          * @param $newname string
1282          */
1283         function closeRenameAndReopen( $newname ) {
1284                 $this->sink->closeRenameAndReopen( $newname );
1285         }
1286
1287         /**
1288          * @param $newname string
1289          * @param $open bool
1290          */
1291         function closeAndRename( $newname, $open = false ) {
1292                 $this->sink->closeAndRename( $newname, $open );
1293         }
1294
1295         /**
1296          * @return array
1297          */
1298         function getFilenames() {
1299                 return $this->sink->getFilenames();
1300         }
1301
1302         /**
1303          * Override for page-based filter types.
1304          * @param $page
1305          * @return bool
1306          */
1307         function pass( $page ) {
1308                 return true;
1309         }
1310 }
1311
1312 /**
1313  * Simple dump output filter to exclude all talk pages.
1314  * @ingroup Dump
1315  */
1316 class DumpNotalkFilter extends DumpFilter {
1317
1318         /**
1319          * @param $page
1320          * @return bool
1321          */
1322         function pass( $page ) {
1323                 return !MWNamespace::isTalk( $page->page_namespace );
1324         }
1325 }
1326
1327 /**
1328  * Dump output filter to include or exclude pages in a given set of namespaces.
1329  * @ingroup Dump
1330  */
1331 class DumpNamespaceFilter extends DumpFilter {
1332         var $invert = false;
1333         var $namespaces = array();
1334
1335         /**
1336          * @param $sink DumpOutput
1337          * @param $param
1338          */
1339         function __construct( &$sink, $param ) {
1340                 parent::__construct( $sink );
1341
1342                 $constants = array(
1343                         "NS_MAIN"           => NS_MAIN,
1344                         "NS_TALK"           => NS_TALK,
1345                         "NS_USER"           => NS_USER,
1346                         "NS_USER_TALK"      => NS_USER_TALK,
1347                         "NS_PROJECT"        => NS_PROJECT,
1348                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
1349                         "NS_FILE"           => NS_FILE,
1350                         "NS_FILE_TALK"      => NS_FILE_TALK,
1351                         "NS_IMAGE"          => NS_IMAGE,  // NS_IMAGE is an alias for NS_FILE
1352                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
1353                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
1354                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
1355                         "NS_TEMPLATE"       => NS_TEMPLATE,
1356                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
1357                         "NS_HELP"           => NS_HELP,
1358                         "NS_HELP_TALK"      => NS_HELP_TALK,
1359                         "NS_CATEGORY"       => NS_CATEGORY,
1360                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
1361
1362                 if ( $param { 0 } == '!' ) {
1363                         $this->invert = true;
1364                         $param = substr( $param, 1 );
1365                 }
1366
1367                 foreach ( explode( ',', $param ) as $key ) {
1368                         $key = trim( $key );
1369                         if ( isset( $constants[$key] ) ) {
1370                                 $ns = $constants[$key];
1371                                 $this->namespaces[$ns] = true;
1372                         } elseif ( is_numeric( $key ) ) {
1373                                 $ns = intval( $key );
1374                                 $this->namespaces[$ns] = true;
1375                         } else {
1376                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
1377                         }
1378                 }
1379         }
1380
1381         /**
1382          * @param $page
1383          * @return bool
1384          */
1385         function pass( $page ) {
1386                 $match = isset( $this->namespaces[$page->page_namespace] );
1387                 return $this->invert xor $match;
1388         }
1389 }
1390
1391
1392 /**
1393  * Dump output filter to include only the last revision in each page sequence.
1394  * @ingroup Dump
1395  */
1396 class DumpLatestFilter extends DumpFilter {
1397         var $page, $pageString, $rev, $revString;
1398
1399         /**
1400          * @param $page
1401          * @param $string string
1402          */
1403         function writeOpenPage( $page, $string ) {
1404                 $this->page = $page;
1405                 $this->pageString = $string;
1406         }
1407
1408         /**
1409          * @param $string string
1410          */
1411         function writeClosePage( $string ) {
1412                 if ( $this->rev ) {
1413                         $this->sink->writeOpenPage( $this->page, $this->pageString );
1414                         $this->sink->writeRevision( $this->rev, $this->revString );
1415                         $this->sink->writeClosePage( $string );
1416                 }
1417                 $this->rev = null;
1418                 $this->revString = null;
1419                 $this->page = null;
1420                 $this->pageString = null;
1421         }
1422
1423         /**
1424          * @param $rev
1425          * @param $string string
1426          */
1427         function writeRevision( $rev, $string ) {
1428                 if ( $rev->rev_id == $this->page->page_latest ) {
1429                         $this->rev = $rev;
1430                         $this->revString = $string;
1431                 }
1432         }
1433 }
1434
1435 /**
1436  * Base class for output stream; prints to stdout or buffer or whereever.
1437  * @ingroup Dump
1438  */
1439 class DumpMultiWriter {
1440
1441         /**
1442          * @param $sinks
1443          */
1444         function __construct( $sinks ) {
1445                 $this->sinks = $sinks;
1446                 $this->count = count( $sinks );
1447         }
1448
1449         /**
1450          * @param $string string
1451          */
1452         function writeOpenStream( $string ) {
1453                 for ( $i = 0; $i < $this->count; $i++ ) {
1454                         $this->sinks[$i]->writeOpenStream( $string );
1455                 }
1456         }
1457
1458         /**
1459          * @param $string string
1460          */
1461         function writeCloseStream( $string ) {
1462                 for ( $i = 0; $i < $this->count; $i++ ) {
1463                         $this->sinks[$i]->writeCloseStream( $string );
1464                 }
1465         }
1466
1467         /**
1468          * @param $page
1469          * @param $string string
1470          */
1471         function writeOpenPage( $page, $string ) {
1472                 for ( $i = 0; $i < $this->count; $i++ ) {
1473                         $this->sinks[$i]->writeOpenPage( $page, $string );
1474                 }
1475         }
1476
1477         /**
1478          * @param $string
1479          */
1480         function writeClosePage( $string ) {
1481                 for ( $i = 0; $i < $this->count; $i++ ) {
1482                         $this->sinks[$i]->writeClosePage( $string );
1483                 }
1484         }
1485
1486         /**
1487          * @param $rev
1488          * @param $string
1489          */
1490         function writeRevision( $rev, $string ) {
1491                 for ( $i = 0; $i < $this->count; $i++ ) {
1492                         $this->sinks[$i]->writeRevision( $rev, $string );
1493                 }
1494         }
1495
1496         /**
1497          * @param $newnames
1498          */
1499         function closeRenameAndReopen( $newnames ) {
1500                 $this->closeAndRename( $newnames, true );
1501         }
1502
1503         /**
1504          * @param $newnames array
1505          * @param bool $open
1506          */
1507         function closeAndRename( $newnames, $open = false ) {
1508                 for ( $i = 0; $i < $this->count; $i++ ) {
1509                         $this->sinks[$i]->closeAndRename( $newnames[$i], $open );
1510                 }
1511         }
1512
1513         /**
1514          * @return array
1515          */
1516         function getFilenames() {
1517                 $filenames = array();
1518                 for ( $i = 0; $i < $this->count; $i++ ) {
1519                         $filenames[] =  $this->sinks[$i]->getFilenames();
1520                 }
1521                 return $filenames;
1522         }
1523
1524 }
1525
1526 /**
1527  * @param $string string
1528  * @return string
1529  */
1530 function xmlsafe( $string ) {
1531         wfProfileIn( __FUNCTION__ );
1532
1533         /**
1534          * The page may contain old data which has not been properly normalized.
1535          * Invalid UTF-8 sequences or forbidden control characters will make our
1536          * XML output invalid, so be sure to strip them out.
1537          */
1538         $string = UtfNormal::cleanUp( $string );
1539
1540         $string = htmlspecialchars( $string );
1541         wfProfileOut( __FUNCTION__ );
1542         return $string;
1543 }