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          * Returns the export schema version.
  63          * @return string
  64          */
  65         public static function schemaVersion() {
  66                 return "0.8";
  67         }
  68
  69         /**
  70          * If using WikiExporter::STREAM to stream a large amount of data,
  71          * provide a database connection which is not managed by
  72          * LoadBalancer to read from: some history blob types will
  73          * make additional queries to pull source data while the
  74          * main query is still running.
  75          *
  76          * @param $db DatabaseBase
  77          * @param $history Mixed: one of WikiExporter::FULL, WikiExporter::CURRENT,
  78          *                 WikiExporter::RANGE or WikiExporter::STABLE,
  79          *                 or an associative array:
  80          *                   offset: non-inclusive offset at which to start the query
  81          *                   limit: maximum number of rows to return
  82          *                   dir: "asc" or "desc" timestamp order
  83          * @param int $buffer one of WikiExporter::BUFFER or WikiExporter::STREAM
  84          * @param int $text one of WikiExporter::TEXT or WikiExporter::STUB
  85          */
  86         function __construct( $db, $history = WikiExporter::CURRENT,
  87                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  88                 $this->db = $db;
  89                 $this->history = $history;
  90                 $this->buffer = $buffer;
  91                 $this->writer = new XmlDumpWriter();
  92                 $this->sink = new DumpOutput();
  93                 $this->text = $text;
  94         }
  95
  96         /**
  97          * Set the DumpOutput or DumpFilter object which will receive
  98          * various row objects and XML output for filtering. Filters
  99          * can be chained or used as callbacks.
 100          *
 101          * @param $sink mixed
 102          */
 103         public function setOutputSink( &$sink ) {
 104                 $this->sink =& $sink;
 105         }
 106
 107         public function openStream() {
 108                 $output = $this->writer->openStream();
 109                 $this->sink->writeOpenStream( $output );
 110         }
 111
 112         public function closeStream() {
 113                 $output = $this->writer->closeStream();
 114                 $this->sink->writeCloseStream( $output );
 115         }
 116
 117         /**
 118          * Dumps a series of page and revision records for all pages
 119          * in the database, either including complete history or only
 120          * the most recent version.
 121          */
 122         public function allPages() {
 123                 $this->dumpFrom( '' );
 124         }
 125
 126         /**
 127          * Dumps a series of page and revision records for those pages
 128          * in the database falling within the page_id range given.
 129          * @param int $start inclusive lower limit (this id is included)
 130          * @param $end   Int: Exclusive upper limit (this id is not included)
 131          *                   If 0, no upper limit.
 132          */
 133         public function pagesByRange( $start, $end ) {
 134                 $condition = 'page_id >= ' . intval( $start );
 135                 if ( $end ) {
 136                         $condition .= ' AND page_id < ' . intval( $end );
 137                 }
 138                 $this->dumpFrom( $condition );
 139         }
 140
 141         /**
 142          * Dumps a series of page and revision records for those pages
 143          * in the database with revisions falling within the rev_id range given.
 144          * @param int $start inclusive lower limit (this id is included)
 145          * @param $end   Int: Exclusive upper limit (this id is not included)
 146          *                   If 0, no upper limit.
 147          */
 148         public function revsByRange( $start, $end ) {
 149                 $condition = 'rev_id >= ' . intval( $start );
 150                 if ( $end ) {
 151                         $condition .= ' AND rev_id < ' . intval( $end );
 152                 }
 153                 $this->dumpFrom( $condition );
 154         }
 155
 156         /**
 157          * @param $title Title
 158          */
 159         public function pageByTitle( $title ) {
 160                 $this->dumpFrom(
 161                         'page_namespace=' . $title->getNamespace() .
 162                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 163         }
 164
 165         /**
 166          * @param $name string
 167          * @throws MWException
 168          */
 169         public function pageByName( $name ) {
 170                 $title = Title::newFromText( $name );
 171                 if ( is_null( $title ) ) {
 172                         throw new MWException( "Can't export invalid title" );
 173                 } else {
 174                         $this->pageByTitle( $title );
 175                 }
 176         }
 177
 178         /**
 179          * @param $names array
 180          */
 181         public function pagesByName( $names ) {
 182                 foreach ( $names as $name ) {
 183                         $this->pageByName( $name );
 184                 }
 185         }
 186
 187         public function allLogs() {
 188                 $this->dumpFrom( '' );
 189         }
 190
 191         /**
 192          * @param $start int
 193          * @param $end int
 194          */
 195         public function logsByRange( $start, $end ) {
 196                 $condition = 'log_id >= ' . intval( $start );
 197                 if ( $end ) {
 198                         $condition .= ' AND log_id < ' . intval( $end );
 199                 }
 200                 $this->dumpFrom( $condition );
 201         }
 202
 203         /**
 204          * Generates the distinct list of authors of an article
 205          * Not called by default (depends on $this->list_authors)
 206          * Can be set by Special:Export when not exporting whole history
 207          *
 208          * @param $cond
 209          */
 210         protected function do_list_authors( $cond ) {
 211                 wfProfileIn( __METHOD__ );
 212                 $this->author_list = "<contributors>";
 213                 // rev_deleted
 214
 215                 $res = $this->db->select(
 216                         array( 'page', 'revision' ),
 217                         array( 'DISTINCT rev_user_text', 'rev_user' ),
 218                         array(
 219                                 $this->db->bitAnd( 'rev_deleted', Revision::DELETED_USER ) . ' = 0',
 220                                 $cond,
 221                                 'page_id = rev_id',
 222                         ),
 223                         __METHOD__
 224                 );
 225
 226                 foreach ( $res as $row ) {
 227                         $this->author_list .= "<contributor>" .
 228                                 "<username>" .
 229                                 htmlentities( $row->rev_user_text ) .
 230                                 "</username>" .
 231                                 "<id>" .
 232                                 $row->rev_user .
 233                                 "</id>" .
 234                                 "</contributor>";
 235                 }
 236                 $this->author_list .= "</contributors>";
 237                 wfProfileOut( __METHOD__ );
 238         }
 239
 240         /**
 241          * @param $cond string
 242          * @throws MWException
 243          * @throws Exception
 244          */
 245         protected function dumpFrom( $cond = '' ) {
 246                 wfProfileIn( __METHOD__ );
 247                 # For logging dumps...
 248                 if ( $this->history & self::LOGS ) {
 249                         $where = array( 'user_id = log_user' );
 250                         # Hide private logs
 251                         $hideLogs = LogEventsList::getExcludeClause( $this->db );
 252                         if ( $hideLogs ) $where[] = $hideLogs;
 253                         # Add on any caller specified conditions
 254                         if ( $cond ) $where[] = $cond;
 255                         # Get logging table name for logging.* clause
 256                         $logging = $this->db->tableName( 'logging' );
 257
 258                         if ( $this->buffer == WikiExporter::STREAM ) {
 259                                 $prev = $this->db->bufferResults( false );
 260                         }
 261                         $wrapper = null; // Assuring $wrapper is not undefined, if exception occurs early
 262                         try {
 263                                 $result = $this->db->select( array( 'logging', 'user' ),
 264                                         array( "{$logging}.*", 'user_name' ), // grab the user name
 265                                         $where,
 266                                         __METHOD__,
 267                                         array( 'ORDER BY' => 'log_id', 'USE INDEX' => array( 'logging' => 'PRIMARY' ) )
 268                                 );
 269                                 $wrapper = $this->db->resultObject( $result );
 270                                 $this->outputLogStream( $wrapper );
 271                                 if ( $this->buffer == WikiExporter::STREAM ) {
 272                                         $this->db->bufferResults( $prev );
 273                                 }
 274                         } catch ( Exception $e ) {
 275                                 // Throwing the exception does not reliably free the resultset, and
 276                                 // would also leave the connection in unbuffered mode.
 277
 278                                 // Freeing result
 279                                 try {
 280                                         if ( $wrapper ) {
 281                                                 $wrapper->free();
 282                                         }
 283                                 } catch ( Exception $e2 ) {
 284                                         // Already in panic mode -> ignoring $e2 as $e has
 285                                         // higher priority
 286                                 }
 287
 288                                 // Putting database back in previous buffer mode
 289                                 try {
 290                                         if ( $this->buffer == WikiExporter::STREAM ) {
 291                                                 $this->db->bufferResults( $prev );
 292                                         }
 293                                 } catch ( Exception $e2 ) {
 294                                         // Already in panic mode -> ignoring $e2 as $e has
 295                                         // higher priority
 296                                 }
 297
 298                                 // Inform caller about problem
 299                                 wfProfileOut( __METHOD__ );
 300                                 throw $e;
 301                         }
 302                 # For page dumps...
 303                 } else {
 304                         $tables = array( 'page', 'revision' );
 305                         $opts = array( 'ORDER BY' => 'page_id ASC' );
 306                         $opts['USE INDEX'] = array();
 307                         $join = array();
 308                         if ( is_array( $this->history ) ) {
 309                                 # Time offset/limit for all pages/history...
 310                                 $revJoin = 'page_id=rev_page';
 311                                 # Set time order
 312                                 if ( $this->history['dir'] == 'asc' ) {
 313                                         $op = '>';
 314                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 315                                 } else {
 316                                         $op = '<';
 317                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 318                                 }
 319                                 # Set offset
 320                                 if ( !empty( $this->history['offset'] ) ) {
 321                                         $revJoin .= " AND rev_timestamp $op " .
 322                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 323                                 }
 324                                 $join['revision'] = array( 'INNER JOIN', $revJoin );
 325                                 # Set query limit
 326                                 if ( !empty( $this->history['limit'] ) ) {
 327                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 328                                 }
 329                         } elseif ( $this->history & WikiExporter::FULL ) {
 330                                 # Full history dumps...
 331                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 332                         } elseif ( $this->history & WikiExporter::CURRENT ) {
 333                                 # Latest revision dumps...
 334                                 if ( $this->list_authors && $cond != '' ) { // List authors, if so desired
 335                                         $this->do_list_authors( $cond );
 336                                 }
 337                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 338                         } elseif ( $this->history & WikiExporter::STABLE ) {
 339                                 # "Stable" revision dumps...
 340                                 # Default JOIN, to be overridden...
 341                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 342                                 # One, and only one hook should set this, and return false
 343                                 if ( wfRunHooks( 'WikiExporter::dumpStableQuery', array( &$tables, &$opts, &$join ) ) ) {
 344                                         wfProfileOut( __METHOD__ );
 345                                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 346                                 }
 347                         } elseif ( $this->history & WikiExporter::RANGE ) {
 348                                 # Dump of revisions within a specified range
 349                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 350                                 $opts['ORDER BY'] = array( 'rev_page ASC', 'rev_id ASC' );
 351                         } else {
 352                                 # Unknown history specification parameter?
 353                                 wfProfileOut( __METHOD__ );
 354                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 355                         }
 356                         # Query optimization hacks
 357                         if ( $cond == '' ) {
 358                                 $opts[] = 'STRAIGHT_JOIN';
 359                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 360                         }
 361                         # Build text join options
 362                         if ( $this->text != WikiExporter::STUB ) { // 1-pass
 363                                 $tables[] = 'text';
 364                                 $join['text'] = array( 'INNER JOIN', 'rev_text_id=old_id' );
 365                         }
 366
 367                         if ( $this->buffer == WikiExporter::STREAM ) {
 368                                 $prev = $this->db->bufferResults( false );
 369                         }
 370
 371                         $wrapper = null; // Assuring $wrapper is not undefined, if exception occurs early
 372                         try {
 373                                 wfRunHooks( 'ModifyExportQuery',
 374                                                 array( $this->db, &$tables, &$cond, &$opts, &$join ) );
 375
 376                                 # Do the query!
 377                                 $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 378                                 $wrapper = $this->db->resultObject( $result );
 379                                 # Output dump results
 380                                 $this->outputPageStream( $wrapper );
 381
 382                                 if ( $this->buffer == WikiExporter::STREAM ) {
 383                                         $this->db->bufferResults( $prev );
 384                                 }
 385                         } catch ( Exception $e ) {
 386                                 // Throwing the exception does not reliably free the resultset, and
 387                                 // would also leave the connection in unbuffered mode.
 388
 389                                 // Freeing result
 390                                 try {
 391                                         if ( $wrapper ) {
 392                                                 $wrapper->free();
 393                                         }
 394                                 } catch ( Exception $e2 ) {
 395                                         // Already in panic mode -> ignoring $e2 as $e has
 396                                         // higher priority
 397                                 }
 398
 399                                 // Putting database back in previous buffer mode
 400                                 try {
 401                                         if ( $this->buffer == WikiExporter::STREAM ) {
 402                                                 $this->db->bufferResults( $prev );
 403                                         }
 404                                 } catch ( Exception $e2 ) {
 405                                         // Already in panic mode -> ignoring $e2 as $e has
 406                                         // higher priority
 407                                 }
 408
 409                                 // Inform caller about problem
 410                                 throw $e;
 411                         }
 412                 }
 413                 wfProfileOut( __METHOD__ );
 414         }
 415
 416         /**
 417          * Runs through a query result set dumping page and revision records.
 418          * The result set should be sorted/grouped by page to avoid duplicate
 419          * page records in the output.
 420          *
 421          * Should be safe for
 422          * streaming (non-buffered) queries, as long as it was made on a
 423          * separate database connection not managed by LoadBalancer; some
 424          * blob storage types will make queries to pull source data.
 425          *
 426          * @param $resultset ResultWrapper
 427          */
 428         protected function outputPageStream( $resultset ) {
 429                 $last = null;
 430                 foreach ( $resultset as $row ) {
 431                         if ( $last === null ||
 432                                 $last->page_namespace != $row->page_namespace ||
 433                                 $last->page_title != $row->page_title ) {
 434                                 if ( $last !== null ) {
 435                                         $output = '';
 436                                         if ( $this->dumpUploads ) {
 437                                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 438                                         }
 439                                         $output .= $this->writer->closePage();
 440                                         $this->sink->writeClosePage( $output );
 441                                 }
 442                                 $output = $this->writer->openPage( $row );
 443                                 $this->sink->writeOpenPage( $row, $output );
 444                                 $last = $row;
 445                         }
 446                         $output = $this->writer->writeRevision( $row );
 447                         $this->sink->writeRevision( $row, $output );
 448                 }
 449                 if ( $last !== null ) {
 450                         $output = '';
 451                         if ( $this->dumpUploads ) {
 452                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 453                         }
 454                         $output .= $this->author_list;
 455                         $output .= $this->writer->closePage();
 456                         $this->sink->writeClosePage( $output );
 457                 }
 458         }
 459
 460         /**
 461          * @param $resultset array
 462          */
 463         protected function outputLogStream( $resultset ) {
 464                 foreach ( $resultset as $row ) {
 465                         $output = $this->writer->writeLogItem( $row );
 466                         $this->sink->writeLogItem( $row, $output );
 467                 }
 468         }
 469 }
 470
 471 /**
 472  * @ingroup Dump
 473  */
 474 class XmlDumpWriter {
 475         /**
 476          * Returns the export schema version.
 477          * @deprecated in 1.20; use WikiExporter::schemaVersion() instead
 478          * @return string
 479          */
 480         function schemaVersion() {
 481                 wfDeprecated( __METHOD__, '1.20' );
 482                 return WikiExporter::schemaVersion();
 483         }
 484
 485         /**
 486          * Opens the XML output stream's root "<mediawiki>" element.
 487          * This does not include an xml directive, so is safe to include
 488          * as a subelement in a larger XML stream. Namespace and XML Schema
 489          * references are included.
 490          *
 491          * Output will be encoded in UTF-8.
 492          *
 493          * @return string
 494          */
 495         function openStream() {
 496                 global $wgLanguageCode;
 497                 $ver = WikiExporter::schemaVersion();
 498                 return Xml::element( 'mediawiki', array(
 499                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 500                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 501                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 502                                                 "http://www.mediawiki.org/xml/export-$ver.xsd", #TODO: how do we get a new version up there?
 503                         'version'            => $ver,
 504                         'xml:lang'           => $wgLanguageCode ),
 505                         null ) .
 506                         "\n" .
 507                         $this->siteInfo();
 508         }
 509
 510         /**
 511          * @return string
 512          */
 513         function siteInfo() {
 514                 $info = array(
 515                         $this->sitename(),
 516                         $this->homelink(),
 517                         $this->generator(),
 518                         $this->caseSetting(),
 519                         $this->namespaces() );
 520                 return "  <siteinfo>\n    " .
 521                         implode( "\n    ", $info ) .
 522                         "\n  </siteinfo>\n";
 523         }
 524
 525         /**
 526          * @return string
 527          */
 528         function sitename() {
 529                 global $wgSitename;
 530                 return Xml::element( 'sitename', array(), $wgSitename );
 531         }
 532
 533         /**
 534          * @return string
 535          */
 536         function generator() {
 537                 global $wgVersion;
 538                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 539         }
 540
 541         /**
 542          * @return string
 543          */
 544         function homelink() {
 545                 return Xml::element( 'base', array(), Title::newMainPage()->getCanonicalURL() );
 546         }
 547
 548         /**
 549          * @return string
 550          */
 551         function caseSetting() {
 552                 global $wgCapitalLinks;
 553                 // "case-insensitive" option is reserved for future
 554                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 555                 return Xml::element( 'case', array(), $sensitivity );
 556         }
 557
 558         /**
 559          * @return string
 560          */
 561         function namespaces() {
 562                 global $wgContLang;
 563                 $spaces = "<namespaces>\n";
 564                 foreach ( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 565                         $spaces .= '      ' .
 566                                 Xml::element( 'namespace',
 567                                         array(
 568                                                 'key' => $ns,
 569                                                 'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 570                                         ), $title ) . "\n";
 571                 }
 572                 $spaces .= "    </namespaces>";
 573                 return $spaces;
 574         }
 575
 576         /**
 577          * Closes the output stream with the closing root element.
 578          * Call when finished dumping things.
 579          *
 580          * @return string
 581          */
 582         function closeStream() {
 583                 return "</mediawiki>\n";
 584         }
 585
 586         /**
 587          * Opens a "<page>" section on the output stream, with data
 588          * from the given database row.
 589          *
 590          * @param $row object
 591          * @return string
 592          * @access private
 593          */
 594         function openPage( $row ) {
 595                 $out = "  <page>\n";
 596                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 597                 $out .= '    ' . Xml::elementClean( 'title', array(), self::canonicalTitle( $title ) ) . "\n";
 598                 $out .= '    ' . Xml::element( 'ns', array(), strval( $row->page_namespace) ) . "\n";
 599                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 600                 if ( $row->page_is_redirect ) {
 601                         $page = WikiPage::factory( $title );
 602                         $redirect = $page->getRedirectTarget();
 603                         if ( $redirect instanceOf Title && $redirect->isValidRedirectTarget() ) {
 604                                 $out .= '    ' . Xml::element( 'redirect', array( 'title' => self::canonicalTitle( $redirect ) ) ) . "\n";
 605                         }
 606                 }
 607
 608                 if ( $row->page_restrictions != '' ) {
 609                         $out .= '    ' . Xml::element( 'restrictions', array(),
 610                                 strval( $row->page_restrictions ) ) . "\n";
 611                 }
 612
 613                 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 614
 615                 return $out;
 616         }
 617
 618         /**
 619          * Closes a "<page>" section on the output stream.
 620          *
 621          * @access private
 622          * @return string
 623          */
 624         function closePage() {
 625                 return "  </page>\n";
 626         }
 627
 628         /**
 629          * Dumps a "<revision>" section on the output stream, with
 630          * data filled in from the given database row.
 631          *
 632          * @param $row object
 633          * @return string
 634          * @access private
 635          */
 636         function writeRevision( $row ) {
 637                 wfProfileIn( __METHOD__ );
 638
 639                 $out = "    <revision>\n";
 640                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 641                 if( isset( $row->rev_parent_id ) && $row->rev_parent_id ) {
 642                         $out .= "      " . Xml::element( 'parentid', null, strval( $row->rev_parent_id ) ) . "\n";
 643                 }
 644
 645                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 646
 647                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_USER ) ) {
 648                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 649                 } else {
 650                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 651                 }
 652
 653                 if ( isset( $row->rev_minor_edit ) && $row->rev_minor_edit ) {
 654                         $out .= "      <minor/>\n";
 655                 }
 656                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_COMMENT ) ) {
 657                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 658                 } elseif ( $row->rev_comment != '' ) {
 659                         $out .= "      " . Xml::elementClean( 'comment', array(), strval( $row->rev_comment ) ) . "\n";
 660                 }
 661
 662                 $text = '';
 663                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_TEXT ) ) {
 664                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 665                 } elseif ( isset( $row->old_text ) ) {
 666                         // Raw text from the database may have invalid chars
 667                         $text = strval( Revision::getRevisionText( $row ) );
 668                         $out .= "      " . Xml::elementClean( 'text',
 669                                 array( 'xml:space' => 'preserve', 'bytes' => intval( $row->rev_len ) ),
 670                                 strval( $text ) ) . "\n";
 671                 } else {
 672                         // Stub output
 673                         $out .= "      " . Xml::element( 'text',
 674                                 array( 'id' => $row->rev_text_id, 'bytes' => intval( $row->rev_len ) ),
 675                                 "" ) . "\n";
 676                 }
 677
 678                 if ( isset( $row->rev_sha1 ) && $row->rev_sha1 && !( $row->rev_deleted & Revision::DELETED_TEXT ) ) {
 679                         $out .= "      " . Xml::element( 'sha1', null, strval( $row->rev_sha1 ) ) . "\n";
 680                 } else {
 681                         $out .= "      <sha1/>\n";
 682                 }
 683
 684                 if ( isset( $row->rev_content_model ) && !is_null( $row->rev_content_model ) ) {
 685                         $content_model = strval( $row->rev_content_model );
 686                 } else {
 687                         // probably using $wgContentHandlerUseDB = false;
 688                         // @todo: test!
 689                         $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 690                         $content_model = ContentHandler::getDefaultModelFor( $title );
 691                 }
 692
 693                 $out .= "      " . Xml::element( 'model', null, strval( $content_model ) ) . "\n";
 694
 695                 if ( isset( $row->rev_content_format ) && !is_null( $row->rev_content_format ) ) {
 696                         $content_format = strval( $row->rev_content_format );
 697                 } else {
 698                         // probably using $wgContentHandlerUseDB = false;
 699                         // @todo: test!
 700                         $content_handler = ContentHandler::getForModelID( $content_model );
 701                         $content_format = $content_handler->getDefaultFormat();
 702                 }
 703
 704                 $out .= "      " . Xml::element( 'format', null, strval( $content_format ) ) . "\n";
 705
 706                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 707
 708                 $out .= "    </revision>\n";
 709
 710                 wfProfileOut( __METHOD__ );
 711                 return $out;
 712         }
 713
 714         /**
 715          * Dumps a "<logitem>" section on the output stream, with
 716          * data filled in from the given database row.
 717          *
 718          * @param $row object
 719          * @return string
 720          * @access private
 721          */
 722         function writeLogItem( $row ) {
 723                 wfProfileIn( __METHOD__ );
 724
 725                 $out = "  <logitem>\n";
 726                 $out .= "    " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 727
 728                 $out .= $this->writeTimestamp( $row->log_timestamp, "    " );
 729
 730                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 731                         $out .= "    " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 732                 } else {
 733                         $out .= $this->writeContributor( $row->log_user, $row->user_name, "    " );
 734                 }
 735
 736                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 737                         $out .= "    " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 738                 } elseif ( $row->log_comment != '' ) {
 739                         $out .= "    " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 740                 }
 741
 742                 $out .= "    " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 743                 $out .= "    " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 744
 745                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 746                         $out .= "    " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 747                 } else {
 748                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 749                         $out .= "    " . Xml::elementClean( 'logtitle', null, self::canonicalTitle( $title ) ) . "\n";
 750                         $out .= "    " . Xml::elementClean( 'params',
 751                                 array( 'xml:space' => 'preserve' ),
 752                                 strval( $row->log_params ) ) . "\n";
 753                 }
 754
 755                 $out .= "  </logitem>\n";
 756
 757                 wfProfileOut( __METHOD__ );
 758                 return $out;
 759         }
 760
 761         /**
 762          * @param $timestamp string
 763          * @param string $indent Default to six spaces
 764          * @return string
 765          */
 766         function writeTimestamp( $timestamp, $indent = "      " ) {
 767                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 768                 return $indent . Xml::element( 'timestamp', null, $ts ) . "\n";
 769         }
 770
 771         /**
 772          * @param $id
 773          * @param $text string
 774          * @param string $indent Default to six spaces
 775          * @return string
 776          */
 777         function writeContributor( $id, $text, $indent = "      " ) {
 778                 $out = $indent . "<contributor>\n";
 779                 if ( $id || !IP::isValid( $text ) ) {
 780                         $out .= $indent . "  " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 781                         $out .= $indent . "  " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 782                 } else {
 783                         $out .= $indent . "  " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 784                 }
 785                 $out .= $indent . "</contributor>\n";
 786                 return $out;
 787         }
 788
 789         /**
 790          * Warning! This data is potentially inconsistent. :(
 791          * @param $row
 792          * @param $dumpContents bool
 793          * @return string
 794          */
 795         function writeUploads( $row, $dumpContents = false ) {
 796                 if ( $row->page_namespace == NS_FILE ) {
 797                         $img = wfLocalFile( $row->page_title );
 798                         if ( $img && $img->exists() ) {
 799                                 $out = '';
 800                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 801                                         $out .= $this->writeUpload( $ver, $dumpContents );
 802                                 }
 803                                 $out .= $this->writeUpload( $img, $dumpContents );
 804                                 return $out;
 805                         }
 806                 }
 807                 return '';
 808         }
 809
 810         /**
 811          * @param $file File
 812          * @param $dumpContents bool
 813          * @return string
 814          */
 815         function writeUpload( $file, $dumpContents = false ) {
 816                 if ( $file->isOld() ) {
 817                         $archiveName = "      " .
 818                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 819                 } else {
 820                         $archiveName = '';
 821                 }
 822                 if ( $dumpContents ) {
 823                         # Dump file as base64
 824                         # Uses only XML-safe characters, so does not need escaping
 825                         $contents = '      <contents encoding="base64">' .
 826                                 chunk_split( base64_encode( file_get_contents( $file->getPath() ) ) ) .
 827                                 "      </contents>\n";
 828                 } else {
 829                         $contents = '';
 830                 }
 831                 if ( $file->isDeleted( File::DELETED_COMMENT ) ) {
 832                         $comment = Xml::element( 'comment', array( 'deleted' => 'deleted' ) );
 833                 } else {
 834                         $comment = Xml::elementClean( 'comment', null, $file->getDescription() );
 835                 }
 836                 return "    <upload>\n" .
 837                         $this->writeTimestamp( $file->getTimestamp() ) .
 838                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 839                         "      " . $comment . "\n" .
 840                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 841                         $archiveName .
 842                         "      " . Xml::element( 'src', null, $file->getCanonicalURL() ) . "\n" .
 843                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 844                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 845                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 846                         $contents .
 847                         "    </upload>\n";
 848         }
 849
 850         /**
 851          * Return prefixed text form of title, but using the content language's
 852          * canonical namespace. This skips any special-casing such as gendered
 853          * user namespaces -- which while useful, are not yet listed in the
 854          * XML "<siteinfo>" data so are unsafe in export.
 855          *
 856          * @param Title $title
 857          * @return string
 858          * @since 1.18
 859          */
 860         public static function canonicalTitle( Title $title ) {
 861                 if ( $title->getInterwiki() ) {
 862                         return $title->getPrefixedText();
 863                 }
 864
 865                 global $wgContLang;
 866                 $prefix = str_replace( '_', ' ', $wgContLang->getNsText( $title->getNamespace() ) );
 867
 868                 if ( $prefix !== '' ) {
 869                         $prefix .= ':';
 870                 }
 871
 872                 return $prefix . $title->getText();
 873         }
 874 }
 875
 876 /**
 877  * Base class for output stream; prints to stdout or buffer or wherever.
 878  * @ingroup Dump
 879  */
 880 class DumpOutput {
 881
 882         /**
 883          * @param $string string
 884          */
 885         function writeOpenStream( $string ) {
 886                 $this->write( $string );
 887         }
 888
 889         /**
 890          * @param $string string
 891          */
 892         function writeCloseStream( $string ) {
 893                 $this->write( $string );
 894         }
 895
 896         /**
 897          * @param $page
 898          * @param $string string
 899          */
 900         function writeOpenPage( $page, $string ) {
 901                 $this->write( $string );
 902         }
 903
 904         /**
 905          * @param $string string
 906          */
 907         function writeClosePage( $string ) {
 908                 $this->write( $string );
 909         }
 910
 911         /**
 912          * @param $rev
 913          * @param $string string
 914          */
 915         function writeRevision( $rev, $string ) {
 916                 $this->write( $string );
 917         }
 918
 919         /**
 920          * @param $rev
 921          * @param $string string
 922          */
 923         function writeLogItem( $rev, $string ) {
 924                 $this->write( $string );
 925         }
 926
 927         /**
 928          * Override to write to a different stream type.
 929          * @param $string string
 930          * @return bool
 931          */
 932         function write( $string ) {
 933                 print $string;
 934         }
 935
 936         /**
 937          * Close the old file, move it to a specified name,
 938          * and reopen new file with the old name. Use this
 939          * for writing out a file in multiple pieces
 940          * at specified checkpoints (e.g. every n hours).
 941          * @param $newname mixed File name. May be a string or an array with one element
 942          */
 943         function closeRenameAndReopen( $newname ) {
 944         }
 945
 946         /**
 947          * Close the old file, and move it to a specified name.
 948          * Use this for the last piece of a file written out
 949          * at specified checkpoints (e.g. every n hours).
 950          * @param $newname mixed File name. May be a string or an array with one element
 951          * @param bool $open If true, a new file with the old filename will be opened again for writing (default: false)
 952          */
 953         function closeAndRename( $newname, $open = false ) {
 954         }
 955
 956         /**
 957          * Returns the name of the file or files which are
 958          * being written to, if there are any.
 959          * @return null
 960          */
 961         function getFilenames() {
 962                 return null;
 963         }
 964 }
 965
 966 /**
 967  * Stream outputter to send data to a file.
 968  * @ingroup Dump
 969  */
 970 class DumpFileOutput extends DumpOutput {
 971         protected $handle = false, $filename;
 972
 973         /**
 974          * @param $file
 975          */
 976         function __construct( $file ) {
 977                 $this->handle = fopen( $file, "wt" );
 978                 $this->filename = $file;
 979         }
 980
 981         /**
 982          * @param $string string
 983          */
 984         function writeCloseStream( $string ) {
 985                 parent::writeCloseStream( $string );
 986                 if ( $this->handle ) {
 987                         fclose( $this->handle );
 988                         $this->handle = false;
 989                 }
 990         }
 991
 992         /**
 993          * @param $string string
 994          */
 995         function write( $string ) {
 996                 fputs( $this->handle, $string );
 997         }
 998
 999         /**
1000          * @param $newname
1001          */
1002         function closeRenameAndReopen( $newname ) {
1003                 $this->closeAndRename( $newname, true );
1004         }
1005
1006         /**
1007          * @param $newname
1008          * @throws MWException
1009          */
1010         function renameOrException( $newname ) {
1011                         if ( !rename( $this->filename, $newname ) ) {
1012                                 throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" );
1013                         }
1014         }
1015
1016         /**
1017          * @param $newname array
1018          * @return mixed
1019          * @throws MWException
1020          */
1021         function checkRenameArgCount( $newname ) {
1022                 if ( is_array( $newname ) ) {
1023                         if ( count( $newname ) > 1 ) {
1024                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
1025                         } else {
1026                                 $newname = $newname[0];
1027                         }
1028                 }
1029                 return $newname;
1030         }
1031
1032         /**
1033          * @param $newname mixed
1034          * @param $open bool
1035          */
1036         function closeAndRename( $newname, $open = false ) {
1037                 $newname = $this->checkRenameArgCount( $newname );
1038                 if ( $newname ) {
1039                         if ( $this->handle ) {
1040                                 fclose( $this->handle );
1041                                 $this->handle = false;
1042                         }
1043                         $this->renameOrException( $newname );
1044                         if ( $open ) {
1045                                 $this->handle = fopen( $this->filename, "wt" );
1046                         }
1047                 }
1048         }
1049
1050         /**
1051          * @return string|null
1052          */
1053         function getFilenames() {
1054                 return $this->filename;
1055         }
1056 }
1057
1058 /**
1059  * Stream outputter to send data to a file via some filter program.
1060  * Even if compression is available in a library, using a separate
1061  * program can allow us to make use of a multi-processor system.
1062  * @ingroup Dump
1063  */
1064 class DumpPipeOutput extends DumpFileOutput {
1065         protected $command, $filename;
1066         protected $procOpenResource = false;
1067
1068         /**
1069          * @param $command
1070          * @param $file null
1071          */
1072         function __construct( $command, $file = null ) {
1073                 if ( !is_null( $file ) ) {
1074                         $command .= " > " . wfEscapeShellArg( $file );
1075                 }
1076
1077                 $this->startCommand( $command );
1078                 $this->command = $command;
1079                 $this->filename = $file;
1080         }
1081
1082         /**
1083          * @param $string string
1084          */
1085         function writeCloseStream( $string ) {
1086                 parent::writeCloseStream( $string );
1087                 if ( $this->procOpenResource ) {
1088                         proc_close( $this->procOpenResource );
1089                         $this->procOpenResource = false;
1090                 }
1091         }
1092
1093         /**
1094          * @param $command
1095          */
1096         function startCommand( $command ) {
1097                 $spec = array(
1098                         0 => array( "pipe", "r" ),
1099                 );
1100                 $pipes = array();
1101                 $this->procOpenResource = proc_open( $command, $spec, $pipes );
1102                 $this->handle = $pipes[0];
1103         }
1104
1105         /**
1106          * @param mixed $newname
1107          */
1108         function closeRenameAndReopen( $newname ) {
1109                 $this->closeAndRename( $newname, true );
1110         }
1111
1112         /**
1113          * @param $newname mixed
1114          * @param $open bool
1115          */
1116         function closeAndRename( $newname, $open = false ) {
1117                 $newname = $this->checkRenameArgCount( $newname );
1118                 if ( $newname ) {
1119                         if ( $this->handle ) {
1120                                 fclose( $this->handle );
1121                                 $this->handle = false;
1122                         }
1123                         if ( $this->procOpenResource ) {
1124                                 proc_close( $this->procOpenResource );
1125                                 $this->procOpenResource = false;
1126                         }
1127                         $this->renameOrException( $newname );
1128                         if ( $open ) {
1129                                 $command = $this->command;
1130                                 $command .= " > " . wfEscapeShellArg( $this->filename );
1131                                 $this->startCommand( $command );
1132                         }
1133                 }
1134         }
1135
1136 }
1137
1138 /**
1139  * Sends dump output via the gzip compressor.
1140  * @ingroup Dump
1141  */
1142 class DumpGZipOutput extends DumpPipeOutput {
1143
1144         /**
1145          * @param $file string
1146          */
1147         function __construct( $file ) {
1148                 parent::__construct( "gzip", $file );
1149         }
1150 }
1151
1152 /**
1153  * Sends dump output via the bgzip2 compressor.
1154  * @ingroup Dump
1155  */
1156 class DumpBZip2Output extends DumpPipeOutput {
1157
1158         /**
1159          * @param $file string
1160          */
1161         function __construct( $file ) {
1162                 parent::__construct( "bzip2", $file );
1163         }
1164 }
1165
1166 /**
1167  * Sends dump output via the p7zip compressor.
1168  * @ingroup Dump
1169  */
1170 class Dump7ZipOutput extends DumpPipeOutput {
1171
1172         /**
1173          * @param $file string
1174          */
1175         function __construct( $file ) {
1176                 $command = $this->setup7zCommand( $file );
1177                 parent::__construct( $command );
1178                 $this->filename = $file;
1179         }
1180
1181         /**
1182          * @param $file string
1183          * @return string
1184          */
1185         function setup7zCommand( $file ) {
1186                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
1187                 // Suppress annoying useless crap from p7zip
1188                 // Unfortunately this could suppress real error messages too
1189                 $command .= ' >' . wfGetNull() . ' 2>&1';
1190                 return( $command );
1191         }
1192
1193         /**
1194          * @param $newname string
1195          * @param $open bool
1196          */
1197         function closeAndRename( $newname, $open = false ) {
1198                 $newname = $this->checkRenameArgCount( $newname );
1199                 if ( $newname ) {
1200                         fclose( $this->handle );
1201                         proc_close( $this->procOpenResource );
1202                         $this->renameOrException( $newname );
1203                         if ( $open ) {
1204                                 $command = $this->setup7zCommand( $this->filename );
1205                                 $this->startCommand( $command );
1206                         }
1207                 }
1208         }
1209 }
1210
1211 /**
1212  * Dump output filter class.
1213  * This just does output filtering and streaming; XML formatting is done
1214  * higher up, so be careful in what you do.
1215  * @ingroup Dump
1216  */
1217 class DumpFilter {
1218
1219         /**
1220          * @var DumpOutput
1221          * FIXME will need to be made protected whenever legacy code
1222          * is updated.
1223          */
1224         public $sink;
1225
1226         /**
1227          * @var bool
1228          */
1229         protected $sendingThisPage;
1230
1231         /**
1232          * @param $sink DumpOutput
1233          */
1234         function __construct( &$sink ) {
1235                 $this->sink =& $sink;
1236         }
1237
1238         /**
1239          * @param $string string
1240          */
1241         function writeOpenStream( $string ) {
1242                 $this->sink->writeOpenStream( $string );
1243         }
1244
1245         /**
1246          * @param $string string
1247          */
1248         function writeCloseStream( $string ) {
1249                 $this->sink->writeCloseStream( $string );
1250         }
1251
1252         /**
1253          * @param $page
1254          * @param $string string
1255          */
1256         function writeOpenPage( $page, $string ) {
1257                 $this->sendingThisPage = $this->pass( $page, $string );
1258                 if ( $this->sendingThisPage ) {
1259                         $this->sink->writeOpenPage( $page, $string );
1260                 }
1261         }
1262
1263         /**
1264          * @param $string string
1265          */
1266         function writeClosePage( $string ) {
1267                 if ( $this->sendingThisPage ) {
1268                         $this->sink->writeClosePage( $string );
1269                         $this->sendingThisPage = false;
1270                 }
1271         }
1272
1273         /**
1274          * @param $rev
1275          * @param $string string
1276          */
1277         function writeRevision( $rev, $string ) {
1278                 if ( $this->sendingThisPage ) {
1279                         $this->sink->writeRevision( $rev, $string );
1280                 }
1281         }
1282
1283         /**
1284          * @param $rev
1285          * @param $string string
1286          */
1287         function writeLogItem( $rev, $string ) {
1288                 $this->sink->writeRevision( $rev, $string );
1289         }
1290
1291         /**
1292          * @param $newname string
1293          */
1294         function closeRenameAndReopen( $newname ) {
1295                 $this->sink->closeRenameAndReopen( $newname );
1296         }
1297
1298         /**
1299          * @param $newname string
1300          * @param $open bool
1301          */
1302         function closeAndRename( $newname, $open = false ) {
1303                 $this->sink->closeAndRename( $newname, $open );
1304         }
1305
1306         /**
1307          * @return array
1308          */
1309         function getFilenames() {
1310                 return $this->sink->getFilenames();
1311         }
1312
1313         /**
1314          * Override for page-based filter types.
1315          * @param $page
1316          * @return bool
1317          */
1318         function pass( $page ) {
1319                 return true;
1320         }
1321 }
1322
1323 /**
1324  * Simple dump output filter to exclude all talk pages.
1325  * @ingroup Dump
1326  */
1327 class DumpNotalkFilter extends DumpFilter {
1328
1329         /**
1330          * @param $page
1331          * @return bool
1332          */
1333         function pass( $page ) {
1334                 return !MWNamespace::isTalk( $page->page_namespace );
1335         }
1336 }
1337
1338 /**
1339  * Dump output filter to include or exclude pages in a given set of namespaces.
1340  * @ingroup Dump
1341  */
1342 class DumpNamespaceFilter extends DumpFilter {
1343         var $invert = false;
1344         var $namespaces = array();
1345
1346         /**
1347          * @param $sink DumpOutput
1348          * @param $param
1349          * @throws MWException
1350          */
1351         function __construct( &$sink, $param ) {
1352                 parent::__construct( $sink );
1353
1354                 $constants = array(
1355                         "NS_MAIN"           => NS_MAIN,
1356                         "NS_TALK"           => NS_TALK,
1357                         "NS_USER"           => NS_USER,
1358                         "NS_USER_TALK"      => NS_USER_TALK,
1359                         "NS_PROJECT"        => NS_PROJECT,
1360                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
1361                         "NS_FILE"           => NS_FILE,
1362                         "NS_FILE_TALK"      => NS_FILE_TALK,
1363                         "NS_IMAGE"          => NS_IMAGE, // NS_IMAGE is an alias for NS_FILE
1364                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
1365                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
1366                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
1367                         "NS_TEMPLATE"       => NS_TEMPLATE,
1368                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
1369                         "NS_HELP"           => NS_HELP,
1370                         "NS_HELP_TALK"      => NS_HELP_TALK,
1371                         "NS_CATEGORY"       => NS_CATEGORY,
1372                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
1373
1374                 if ( $param { 0 } == '!' ) {
1375                         $this->invert = true;
1376                         $param = substr( $param, 1 );
1377                 }
1378
1379                 foreach ( explode( ',', $param ) as $key ) {
1380                         $key = trim( $key );
1381                         if ( isset( $constants[$key] ) ) {
1382                                 $ns = $constants[$key];
1383                                 $this->namespaces[$ns] = true;
1384                         } elseif ( is_numeric( $key ) ) {
1385                                 $ns = intval( $key );
1386                                 $this->namespaces[$ns] = true;
1387                         } else {
1388                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
1389                         }
1390                 }
1391         }
1392
1393         /**
1394          * @param $page
1395          * @return bool
1396          */
1397         function pass( $page ) {
1398                 $match = isset( $this->namespaces[$page->page_namespace] );
1399                 return $this->invert xor $match;
1400         }
1401 }
1402
1403 /**
1404  * Dump output filter to include only the last revision in each page sequence.
1405  * @ingroup Dump
1406  */
1407 class DumpLatestFilter extends DumpFilter {
1408         var $page, $pageString, $rev, $revString;
1409
1410         /**
1411          * @param $page
1412          * @param $string string
1413          */
1414         function writeOpenPage( $page, $string ) {
1415                 $this->page = $page;
1416                 $this->pageString = $string;
1417         }
1418
1419         /**
1420          * @param $string string
1421          */
1422         function writeClosePage( $string ) {
1423                 if ( $this->rev ) {
1424                         $this->sink->writeOpenPage( $this->page, $this->pageString );
1425                         $this->sink->writeRevision( $this->rev, $this->revString );
1426                         $this->sink->writeClosePage( $string );
1427                 }
1428                 $this->rev = null;
1429                 $this->revString = null;
1430                 $this->page = null;
1431                 $this->pageString = null;
1432         }
1433
1434         /**
1435          * @param $rev
1436          * @param $string string
1437          */
1438         function writeRevision( $rev, $string ) {
1439                 if ( $rev->rev_id == $this->page->page_latest ) {
1440                         $this->rev = $rev;
1441                         $this->revString = $string;
1442                 }
1443         }
1444 }
1445
1446 /**
1447  * Base class for output stream; prints to stdout or buffer or wherever.
1448  * @ingroup Dump
1449  */
1450 class DumpMultiWriter {
1451
1452         /**
1453          * @param $sinks
1454          */
1455         function __construct( $sinks ) {
1456                 $this->sinks = $sinks;
1457                 $this->count = count( $sinks );
1458         }
1459
1460         /**
1461          * @param $string string
1462          */
1463         function writeOpenStream( $string ) {
1464                 for ( $i = 0; $i < $this->count; $i++ ) {
1465                         $this->sinks[$i]->writeOpenStream( $string );
1466                 }
1467         }
1468
1469         /**
1470          * @param $string string
1471          */
1472         function writeCloseStream( $string ) {
1473                 for ( $i = 0; $i < $this->count; $i++ ) {
1474                         $this->sinks[$i]->writeCloseStream( $string );
1475                 }
1476         }
1477
1478         /**
1479          * @param $page
1480          * @param $string string
1481          */
1482         function writeOpenPage( $page, $string ) {
1483                 for ( $i = 0; $i < $this->count; $i++ ) {
1484                         $this->sinks[$i]->writeOpenPage( $page, $string );
1485                 }
1486         }
1487
1488         /**
1489          * @param $string
1490          */
1491         function writeClosePage( $string ) {
1492                 for ( $i = 0; $i < $this->count; $i++ ) {
1493                         $this->sinks[$i]->writeClosePage( $string );
1494                 }
1495         }
1496
1497         /**
1498          * @param $rev
1499          * @param $string
1500          */
1501         function writeRevision( $rev, $string ) {
1502                 for ( $i = 0; $i < $this->count; $i++ ) {
1503                         $this->sinks[$i]->writeRevision( $rev, $string );
1504                 }
1505         }
1506
1507         /**
1508          * @param $newnames
1509          */
1510         function closeRenameAndReopen( $newnames ) {
1511                 $this->closeAndRename( $newnames, true );
1512         }
1513
1514         /**
1515          * @param $newnames array
1516          * @param bool $open
1517          */
1518         function closeAndRename( $newnames, $open = false ) {
1519                 for ( $i = 0; $i < $this->count; $i++ ) {
1520                         $this->sinks[$i]->closeAndRename( $newnames[$i], $open );
1521                 }
1522         }
1523
1524         /**
1525          * @return array
1526          */
1527         function getFilenames() {
1528                 $filenames = array();
1529                 for ( $i = 0; $i < $this->count; $i++ ) {
1530                         $filenames[] = $this->sinks[$i]->getFilenames();
1531                 }
1532                 return $filenames;
1533         }
1534
1535 }
1536
1537 /**
1538  * @param $string string
1539  * @return string
1540  */
1541 function xmlsafe( $string ) {
1542         wfProfileIn( __FUNCTION__ );
1543
1544         /**
1545          * The page may contain old data which has not been properly normalized.
1546          * Invalid UTF-8 sequences or forbidden control characters will make our
1547          * XML output invalid, so be sure to strip them out.
1548          */
1549         $string = UtfNormal::cleanUp( $string );
1550
1551         $string = htmlspecialchars( $string );
1552         wfProfileOut( __FUNCTION__ );
1553         return $string;
1554 }