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 "0.7";
 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",
 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
 630                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 631
 632                 if ( $row->rev_deleted & Revision::DELETED_USER ) {
 633                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 634                 } else {
 635                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 636                 }
 637
 638                 if ( $row->rev_minor_edit ) {
 639                         $out .=  "      <minor/>\n";
 640                 }
 641                 if ( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 642                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 643                 } elseif ( $row->rev_comment != '' ) {
 644                         $out .= "      " . Xml::elementClean( 'comment', array(), strval( $row->rev_comment ) ) . "\n";
 645                 }
 646
 647                 if ( isset( $row->rev_content_model ) && !is_null( $row->rev_content_model )  ) {
 648                         $name = ContentHandler::getContentModelName( $row->rev_content_model );
 649                         $out .= "      " . Xml::element('model', array( 'name' => $name ), strval( $row->rev_content_model ) ) . "\n";
 650                 }
 651
 652                 if ( isset( $row->rev_content_format ) && !is_null( $row->rev_content_format ) ) {
 653                         $mime = ContentHandler::getContentFormatMimeType( $row->rev_content_format );
 654                         $out .= "      " . Xml::element('format', array( 'mime' => $mime ), strval( $row->rev_content_format ) ) . "\n";
 655                 }
 656
 657                 $text = '';
 658                 if ( $row->rev_deleted & Revision::DELETED_TEXT ) {
 659                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 660                 } elseif ( isset( $row->old_text ) ) {
 661                         // Raw text from the database may have invalid chars
 662                         $text = strval( Revision::getRevisionText( $row ) );
 663                         $out .= "      " . Xml::elementClean( 'text',
 664                                 array( 'xml:space' => 'preserve', 'bytes' => intval( $row->rev_len ) ),
 665                                 strval( $text ) ) . "\n";
 666                 } else {
 667                         // Stub output
 668                         $out .= "      " . Xml::element( 'text',
 669                                 array( 'id' => $row->rev_text_id, 'bytes' => intval( $row->rev_len ) ),
 670                                 "" ) . "\n";
 671                 }
 672
 673                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 674
 675                 $out .= "    </revision>\n";
 676
 677                 wfProfileOut( __METHOD__ );
 678                 return $out;
 679         }
 680
 681         /**
 682          * Dumps a <logitem> section on the output stream, with
 683          * data filled in from the given database row.
 684          *
 685          * @param $row object
 686          * @return string
 687          * @access private
 688          */
 689         function writeLogItem( $row ) {
 690                 wfProfileIn( __METHOD__ );
 691
 692                 $out  = "  <logitem>\n";
 693                 $out .= "    " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 694
 695                 $out .= $this->writeTimestamp( $row->log_timestamp, "    " );
 696
 697                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 698                         $out .= "    " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 699                 } else {
 700                         $out .= $this->writeContributor( $row->log_user, $row->user_name, "    " );
 701                 }
 702
 703                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 704                         $out .= "    " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 705                 } elseif ( $row->log_comment != '' ) {
 706                         $out .= "    " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 707                 }
 708
 709                 $out .= "    " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 710                 $out .= "    " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 711
 712                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 713                         $out .= "    " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 714                 } else {
 715                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 716                         $out .= "    " . Xml::elementClean( 'logtitle', null, self::canonicalTitle( $title ) ) . "\n";
 717                         $out .= "    " . Xml::elementClean( 'params',
 718                                 array( 'xml:space' => 'preserve' ),
 719                                 strval( $row->log_params ) ) . "\n";
 720                 }
 721
 722                 $out .= "  </logitem>\n";
 723
 724                 wfProfileOut( __METHOD__ );
 725                 return $out;
 726         }
 727
 728         /**
 729          * @param $timestamp string
 730          * @return string
 731          */
 732         function writeTimestamp( $timestamp, $indent = "      " ) {
 733                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 734                 return $indent . Xml::element( 'timestamp', null, $ts ) . "\n";
 735         }
 736
 737         /**
 738          * @param $id
 739          * @param $text string
 740          * @return string
 741          */
 742         function writeContributor( $id, $text, $indent = "      " ) {
 743                 $out = $indent . "<contributor>\n";
 744                 if ( $id || !IP::isValid( $text ) ) {
 745                         $out .= $indent . "  " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 746                         $out .= $indent . "  " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 747                 } else {
 748                         $out .= $indent . "  " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 749                 }
 750                 $out .= $indent . "</contributor>\n";
 751                 return $out;
 752         }
 753
 754         /**
 755          * Warning! This data is potentially inconsistent. :(
 756          * @param $row
 757          * @param $dumpContents bool
 758          * @return string
 759          */
 760         function writeUploads( $row, $dumpContents = false ) {
 761                 if ( $row->page_namespace == NS_IMAGE ) {
 762                         $img = wfLocalFile( $row->page_title );
 763                         if ( $img && $img->exists() ) {
 764                                 $out = '';
 765                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 766                                         $out .= $this->writeUpload( $ver, $dumpContents );
 767                                 }
 768                                 $out .= $this->writeUpload( $img, $dumpContents );
 769                                 return $out;
 770                         }
 771                 }
 772                 return '';
 773         }
 774
 775         /**
 776          * @param $file File
 777          * @param $dumpContents bool
 778          * @return string
 779          */
 780         function writeUpload( $file, $dumpContents = false ) {
 781                 if ( $file->isOld() ) {
 782                         $archiveName = "      " .
 783                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 784                 } else {
 785                         $archiveName = '';
 786                 }
 787                 if ( $dumpContents ) {
 788                         # Dump file as base64
 789                         # Uses only XML-safe characters, so does not need escaping
 790                         $contents = '      <contents encoding="base64">' .
 791                                 chunk_split( base64_encode( file_get_contents( $file->getPath() ) ) ) .
 792                                 "      </contents>\n";
 793                 } else {
 794                         $contents = '';
 795                 }
 796                 if ( $file->isDeleted( File::DELETED_COMMENT ) ) {
 797                         $comment = Xml::element( 'comment', array( 'deleted' => 'deleted' ) );
 798                 } else {
 799                         $comment = Xml::elementClean( 'comment', null, $file->getDescription() );
 800                 }
 801                 return "    <upload>\n" .
 802                         $this->writeTimestamp( $file->getTimestamp() ) .
 803                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 804                         "      " . $comment . "\n" .
 805                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 806                         $archiveName .
 807                         "      " . Xml::element( 'src', null, $file->getCanonicalUrl() ) . "\n" .
 808                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 809                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 810                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 811                         $contents .
 812                         "    </upload>\n";
 813         }
 814
 815         /**
 816          * Return prefixed text form of title, but using the content language's
 817          * canonical namespace. This skips any special-casing such as gendered
 818          * user namespaces -- which while useful, are not yet listed in the
 819          * XML <siteinfo> data so are unsafe in export.
 820          *
 821          * @param Title $title
 822          * @return string
 823          * @since 1.18
 824          */
 825         public static function canonicalTitle( Title $title ) {
 826                 if ( $title->getInterwiki() ) {
 827                         return $title->getPrefixedText();
 828                 }
 829
 830                 global $wgContLang;
 831                 $prefix = str_replace( '_', ' ', $wgContLang->getNsText( $title->getNamespace() ) );
 832
 833                 if ( $prefix !== '' ) {
 834                         $prefix .= ':';
 835                 }
 836
 837                 return $prefix . $title->getText();
 838         }
 839 }
 840
 841
 842 /**
 843  * Base class for output stream; prints to stdout or buffer or whereever.
 844  * @ingroup Dump
 845  */
 846 class DumpOutput {
 847
 848         /**
 849          * @param $string string
 850          */
 851         function writeOpenStream( $string ) {
 852                 $this->write( $string );
 853         }
 854
 855         /**
 856          * @param $string string
 857          */
 858         function writeCloseStream( $string ) {
 859                 $this->write( $string );
 860         }
 861
 862         /**
 863          * @param $page
 864          * @param $string string
 865          */
 866         function writeOpenPage( $page, $string ) {
 867                 $this->write( $string );
 868         }
 869
 870         /**
 871          * @param $string string
 872          */
 873         function writeClosePage( $string ) {
 874                 $this->write( $string );
 875         }
 876
 877         /**
 878          * @param $rev
 879          * @param $string string
 880          */
 881         function writeRevision( $rev, $string ) {
 882                 $this->write( $string );
 883         }
 884
 885         /**
 886          * @param $rev
 887          * @param $string string
 888          */
 889         function writeLogItem( $rev, $string ) {
 890                 $this->write( $string );
 891         }
 892
 893         /**
 894          * Override to write to a different stream type.
 895          * @param $string string
 896          * @return bool
 897          */
 898         function write( $string ) {
 899                 print $string;
 900         }
 901
 902         /**
 903          * Close the old file, move it to a specified name,
 904          * and reopen new file with the old name. Use this
 905          * for writing out a file in multiple pieces
 906          * at specified checkpoints (e.g. every n hours).
 907          * @param $newname mixed File name. May be a string or an array with one element
 908          */
 909         function closeRenameAndReopen( $newname ) {
 910                 return;
 911         }
 912
 913         /**
 914          * Close the old file, and move it to a specified name.
 915          * Use this for the last piece of a file written out
 916          * at specified checkpoints (e.g. every n hours).
 917          * @param $newname mixed File name. May be a string or an array with one element
 918          * @param $open bool If true, a new file with the old filename will be opened again for writing (default: false)
 919          */
 920         function closeAndRename( $newname, $open = false ) {
 921                 return;
 922         }
 923
 924         /**
 925          * Returns the name of the file or files which are
 926          * being written to, if there are any.
 927          * @return null
 928          */
 929         function getFilenames() {
 930                 return NULL;
 931         }
 932 }
 933
 934 /**
 935  * Stream outputter to send data to a file.
 936  * @ingroup Dump
 937  */
 938 class DumpFileOutput extends DumpOutput {
 939         protected $handle = false, $filename;
 940
 941         /**
 942          * @param $file
 943          */
 944         function __construct( $file ) {
 945                 $this->handle = fopen( $file, "wt" );
 946                 $this->filename = $file;
 947         }
 948
 949         /**
 950          * @param $string string
 951          */
 952         function writeCloseStream( $string ) {
 953                 parent::writeCloseStream( $string );
 954                 if ( $this->handle ) {
 955                         fclose( $this->handle );
 956                         $this->handle = false;
 957                 }
 958         }
 959
 960         /**
 961          * @param $string string
 962          */
 963         function write( $string ) {
 964                 fputs( $this->handle, $string );
 965         }
 966
 967         /**
 968          * @param $newname
 969          */
 970         function closeRenameAndReopen( $newname ) {
 971                 $this->closeAndRename( $newname, true );
 972         }
 973
 974         /**
 975          * @param $newname
 976          * @throws MWException
 977          */
 978         function renameOrException( $newname ) {
 979                         if (! rename( $this->filename, $newname ) ) {
 980                                 throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" );
 981                         }
 982         }
 983
 984         /**
 985          * @param $newname array
 986          * @return mixed
 987          * @throws MWException
 988          */
 989         function checkRenameArgCount( $newname ) {
 990                 if ( is_array( $newname ) ) {
 991                         if ( count( $newname ) > 1 ) {
 992                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
 993                         } else {
 994                                 $newname = $newname[0];
 995                         }
 996                 }
 997                 return $newname;
 998         }
 999
1000         /**
1001          * @param $newname mixed
1002          * @param $open bool
1003          */
1004         function closeAndRename( $newname, $open = false ) {
1005                 $newname = $this->checkRenameArgCount( $newname );
1006                 if ( $newname ) {
1007                         if ( $this->handle ) {
1008                                 fclose( $this->handle );
1009                                 $this->handle = false;
1010                         }
1011                         $this->renameOrException( $newname );
1012                         if ( $open ) {
1013                                 $this->handle = fopen( $this->filename, "wt" );
1014                         }
1015                 }
1016         }
1017
1018         /**
1019          * @return string|null
1020          */
1021         function getFilenames() {
1022                 return $this->filename;
1023         }
1024 }
1025
1026 /**
1027  * Stream outputter to send data to a file via some filter program.
1028  * Even if compression is available in a library, using a separate
1029  * program can allow us to make use of a multi-processor system.
1030  * @ingroup Dump
1031  */
1032 class DumpPipeOutput extends DumpFileOutput {
1033         protected $command, $filename;
1034         protected $procOpenResource = false;
1035
1036         /**
1037          * @param $command
1038          * @param $file null
1039          */
1040         function __construct( $command, $file = null ) {
1041                 if ( !is_null( $file ) ) {
1042                         $command .=  " > " . wfEscapeShellArg( $file );
1043                 }
1044
1045                 $this->startCommand( $command );
1046                 $this->command = $command;
1047                 $this->filename = $file;
1048         }
1049
1050         /**
1051          * @param $string string
1052          */
1053         function writeCloseStream( $string ) {
1054                 parent::writeCloseStream( $string );
1055                 if ( $this->procOpenResource ) {
1056                         proc_close( $this->procOpenResource );
1057                         $this->procOpenResource = false;
1058                 }
1059         }
1060
1061         /**
1062          * @param $command
1063          */
1064         function startCommand( $command ) {
1065                 $spec = array(
1066                         0 => array( "pipe", "r" ),
1067                 );
1068                 $pipes = array();
1069                 $this->procOpenResource = proc_open( $command, $spec, $pipes );
1070                 $this->handle = $pipes[0];
1071         }
1072
1073         /**
1074          * @param mixed $newname
1075          */
1076         function closeRenameAndReopen( $newname ) {
1077                 $this->closeAndRename( $newname, true );
1078         }
1079
1080         /**
1081          * @param $newname mixed
1082          * @param $open bool
1083          */
1084         function closeAndRename( $newname, $open = false ) {
1085                 $newname = $this->checkRenameArgCount( $newname );
1086                 if ( $newname ) {
1087                         if ( $this->handle ) {
1088                                 fclose( $this->handle );
1089                                 $this->handle = false;
1090                         }
1091                         if ( $this->procOpenResource ) {
1092                                 proc_close( $this->procOpenResource );
1093                                 $this->procOpenResource = false;
1094                         }
1095                         $this->renameOrException( $newname );
1096                         if ( $open ) {
1097                                 $command = $this->command;
1098                                 $command .=  " > " . wfEscapeShellArg( $this->filename );
1099                                 $this->startCommand( $command );
1100                         }
1101                 }
1102         }
1103
1104 }
1105
1106 /**
1107  * Sends dump output via the gzip compressor.
1108  * @ingroup Dump
1109  */
1110 class DumpGZipOutput extends DumpPipeOutput {
1111
1112         /**
1113          * @param $file string
1114          */
1115         function __construct( $file ) {
1116                 parent::__construct( "gzip", $file );
1117         }
1118 }
1119
1120 /**
1121  * Sends dump output via the bgzip2 compressor.
1122  * @ingroup Dump
1123  */
1124 class DumpBZip2Output extends DumpPipeOutput {
1125
1126         /**
1127          * @param $file string
1128          */
1129         function __construct( $file ) {
1130                 parent::__construct( "bzip2", $file );
1131         }
1132 }
1133
1134 /**
1135  * Sends dump output via the p7zip compressor.
1136  * @ingroup Dump
1137  */
1138 class Dump7ZipOutput extends DumpPipeOutput {
1139
1140         /**
1141          * @param $file string
1142          */
1143         function __construct( $file ) {
1144                 $command = $this->setup7zCommand( $file );
1145                 parent::__construct( $command );
1146                 $this->filename = $file;
1147         }
1148
1149         /**
1150          * @param $file string
1151          * @return string
1152          */
1153         function setup7zCommand( $file ) {
1154                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
1155                 // Suppress annoying useless crap from p7zip
1156                 // Unfortunately this could suppress real error messages too
1157                 $command .= ' >' . wfGetNull() . ' 2>&1';
1158                 return( $command );
1159         }
1160
1161         /**
1162          * @param $newname string
1163          * @param $open bool
1164          */
1165         function closeAndRename( $newname, $open = false ) {
1166                 $newname = $this->checkRenameArgCount( $newname );
1167                 if ( $newname ) {
1168                         fclose( $this->handle );
1169                         proc_close( $this->procOpenResource );
1170                         $this->renameOrException( $newname );
1171                         if ( $open ) {
1172                                 $command = $this->setup7zCommand( $this->filename );
1173                                 $this->startCommand( $command );
1174                         }
1175                 }
1176         }
1177 }
1178
1179 /**
1180  * Dump output filter class.
1181  * This just does output filtering and streaming; XML formatting is done
1182  * higher up, so be careful in what you do.
1183  * @ingroup Dump
1184  */
1185 class DumpFilter {
1186
1187         /**
1188          * @var DumpOutput
1189          * FIXME will need to be made protected whenever legacy code
1190          * is updated.
1191          */
1192         public $sink;
1193
1194         /**
1195          * @var bool
1196          */
1197         protected $sendingThisPage;
1198
1199         /**
1200          * @param $sink DumpOutput
1201          */
1202         function __construct( &$sink ) {
1203                 $this->sink =& $sink;
1204         }
1205
1206         /**
1207          * @param $string string
1208          */
1209         function writeOpenStream( $string ) {
1210                 $this->sink->writeOpenStream( $string );
1211         }
1212
1213         /**
1214          * @param $string string
1215          */
1216         function writeCloseStream( $string ) {
1217                 $this->sink->writeCloseStream( $string );
1218         }
1219
1220         /**
1221          * @param $page
1222          * @param $string string
1223          */
1224         function writeOpenPage( $page, $string ) {
1225                 $this->sendingThisPage = $this->pass( $page, $string );
1226                 if ( $this->sendingThisPage ) {
1227                         $this->sink->writeOpenPage( $page, $string );
1228                 }
1229         }
1230
1231         /**
1232          * @param $string string
1233          */
1234         function writeClosePage( $string ) {
1235                 if ( $this->sendingThisPage ) {
1236                         $this->sink->writeClosePage( $string );
1237                         $this->sendingThisPage = false;
1238                 }
1239         }
1240
1241         /**
1242          * @param $rev
1243          * @param $string string
1244          */
1245         function writeRevision( $rev, $string ) {
1246                 if ( $this->sendingThisPage ) {
1247                         $this->sink->writeRevision( $rev, $string );
1248                 }
1249         }
1250
1251         /**
1252          * @param $rev
1253          * @param $string string
1254          */
1255         function writeLogItem( $rev, $string ) {
1256                 $this->sink->writeRevision( $rev, $string );
1257         }
1258
1259         /**
1260          * @param $newname string
1261          */
1262         function closeRenameAndReopen( $newname ) {
1263                 $this->sink->closeRenameAndReopen( $newname );
1264         }
1265
1266         /**
1267          * @param $newname string
1268          * @param $open bool
1269          */
1270         function closeAndRename( $newname, $open = false ) {
1271                 $this->sink->closeAndRename( $newname, $open );
1272         }
1273
1274         /**
1275          * @return array
1276          */
1277         function getFilenames() {
1278                 return $this->sink->getFilenames();
1279         }
1280
1281         /**
1282          * Override for page-based filter types.
1283          * @param $page
1284          * @return bool
1285          */
1286         function pass( $page ) {
1287                 return true;
1288         }
1289 }
1290
1291 /**
1292  * Simple dump output filter to exclude all talk pages.
1293  * @ingroup Dump
1294  */
1295 class DumpNotalkFilter extends DumpFilter {
1296
1297         /**
1298          * @param $page
1299          * @return bool
1300          */
1301         function pass( $page ) {
1302                 return !MWNamespace::isTalk( $page->page_namespace );
1303         }
1304 }
1305
1306 /**
1307  * Dump output filter to include or exclude pages in a given set of namespaces.
1308  * @ingroup Dump
1309  */
1310 class DumpNamespaceFilter extends DumpFilter {
1311         var $invert = false;
1312         var $namespaces = array();
1313
1314         /**
1315          * @param $sink DumpOutput
1316          * @param $param
1317          */
1318         function __construct( &$sink, $param ) {
1319                 parent::__construct( $sink );
1320
1321                 $constants = array(
1322                         "NS_MAIN"           => NS_MAIN,
1323                         "NS_TALK"           => NS_TALK,
1324                         "NS_USER"           => NS_USER,
1325                         "NS_USER_TALK"      => NS_USER_TALK,
1326                         "NS_PROJECT"        => NS_PROJECT,
1327                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
1328                         "NS_FILE"           => NS_FILE,
1329                         "NS_FILE_TALK"      => NS_FILE_TALK,
1330                         "NS_IMAGE"          => NS_IMAGE,  // NS_IMAGE is an alias for NS_FILE
1331                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
1332                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
1333                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
1334                         "NS_TEMPLATE"       => NS_TEMPLATE,
1335                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
1336                         "NS_HELP"           => NS_HELP,
1337                         "NS_HELP_TALK"      => NS_HELP_TALK,
1338                         "NS_CATEGORY"       => NS_CATEGORY,
1339                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
1340
1341                 if ( $param { 0 } == '!' ) {
1342                         $this->invert = true;
1343                         $param = substr( $param, 1 );
1344                 }
1345
1346                 foreach ( explode( ',', $param ) as $key ) {
1347                         $key = trim( $key );
1348                         if ( isset( $constants[$key] ) ) {
1349                                 $ns = $constants[$key];
1350                                 $this->namespaces[$ns] = true;
1351                         } elseif ( is_numeric( $key ) ) {
1352                                 $ns = intval( $key );
1353                                 $this->namespaces[$ns] = true;
1354                         } else {
1355                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
1356                         }
1357                 }
1358         }
1359
1360         /**
1361          * @param $page
1362          * @return bool
1363          */
1364         function pass( $page ) {
1365                 $match = isset( $this->namespaces[$page->page_namespace] );
1366                 return $this->invert xor $match;
1367         }
1368 }
1369
1370
1371 /**
1372  * Dump output filter to include only the last revision in each page sequence.
1373  * @ingroup Dump
1374  */
1375 class DumpLatestFilter extends DumpFilter {
1376         var $page, $pageString, $rev, $revString;
1377
1378         /**
1379          * @param $page
1380          * @param $string string
1381          */
1382         function writeOpenPage( $page, $string ) {
1383                 $this->page = $page;
1384                 $this->pageString = $string;
1385         }
1386
1387         /**
1388          * @param $string string
1389          */
1390         function writeClosePage( $string ) {
1391                 if ( $this->rev ) {
1392                         $this->sink->writeOpenPage( $this->page, $this->pageString );
1393                         $this->sink->writeRevision( $this->rev, $this->revString );
1394                         $this->sink->writeClosePage( $string );
1395                 }
1396                 $this->rev = null;
1397                 $this->revString = null;
1398                 $this->page = null;
1399                 $this->pageString = null;
1400         }
1401
1402         /**
1403          * @param $rev
1404          * @param $string string
1405          */
1406         function writeRevision( $rev, $string ) {
1407                 if ( $rev->rev_id == $this->page->page_latest ) {
1408                         $this->rev = $rev;
1409                         $this->revString = $string;
1410                 }
1411         }
1412 }
1413
1414 /**
1415  * Base class for output stream; prints to stdout or buffer or whereever.
1416  * @ingroup Dump
1417  */
1418 class DumpMultiWriter {
1419
1420         /**
1421          * @param $sinks
1422          */
1423         function __construct( $sinks ) {
1424                 $this->sinks = $sinks;
1425                 $this->count = count( $sinks );
1426         }
1427
1428         /**
1429          * @param $string string
1430          */
1431         function writeOpenStream( $string ) {
1432                 for ( $i = 0; $i < $this->count; $i++ ) {
1433                         $this->sinks[$i]->writeOpenStream( $string );
1434                 }
1435         }
1436
1437         /**
1438          * @param $string string
1439          */
1440         function writeCloseStream( $string ) {
1441                 for ( $i = 0; $i < $this->count; $i++ ) {
1442                         $this->sinks[$i]->writeCloseStream( $string );
1443                 }
1444         }
1445
1446         /**
1447          * @param $page
1448          * @param $string string
1449          */
1450         function writeOpenPage( $page, $string ) {
1451                 for ( $i = 0; $i < $this->count; $i++ ) {
1452                         $this->sinks[$i]->writeOpenPage( $page, $string );
1453                 }
1454         }
1455
1456         /**
1457          * @param $string
1458          */
1459         function writeClosePage( $string ) {
1460                 for ( $i = 0; $i < $this->count; $i++ ) {
1461                         $this->sinks[$i]->writeClosePage( $string );
1462                 }
1463         }
1464
1465         /**
1466          * @param $rev
1467          * @param $string
1468          */
1469         function writeRevision( $rev, $string ) {
1470                 for ( $i = 0; $i < $this->count; $i++ ) {
1471                         $this->sinks[$i]->writeRevision( $rev, $string );
1472                 }
1473         }
1474
1475         /**
1476          * @param $newnames
1477          */
1478         function closeRenameAndReopen( $newnames ) {
1479                 $this->closeAndRename( $newnames, true );
1480         }
1481
1482         /**
1483          * @param $newnames array
1484          * @param bool $open
1485          */
1486         function closeAndRename( $newnames, $open = false ) {
1487                 for ( $i = 0; $i < $this->count; $i++ ) {
1488                         $this->sinks[$i]->closeAndRename( $newnames[$i], $open );
1489                 }
1490         }
1491
1492         /**
1493          * @return array
1494          */
1495         function getFilenames() {
1496                 $filenames = array();
1497                 for ( $i = 0; $i < $this->count; $i++ ) {
1498                         $filenames[] =  $this->sinks[$i]->getFilenames();
1499                 }
1500                 return $filenames;
1501         }
1502
1503 }
1504
1505 /**
1506  * @param $string string
1507  * @return string
1508  */
1509 function xmlsafe( $string ) {
1510         wfProfileIn( __FUNCTION__ );
1511
1512         /**
1513          * The page may contain old data which has not been properly normalized.
1514          * Invalid UTF-8 sequences or forbidden control characters will make our
1515          * XML output invalid, so be sure to strip them out.
1516          */
1517         $string = UtfNormal::cleanUp( $string );
1518
1519         $string = htmlspecialchars( $string );
1520         wfProfileOut( __FUNCTION__ );
1521         return $string;
1522 }