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 ) {
 253                                 $where[] = $hideLogs;
 254                         }
 255                         # Add on any caller specified conditions
 256                         if ( $cond ) {
 257                                 $where[] = $cond;
 258                         }
 259                         # Get logging table name for logging.* clause
 260                         $logging = $this->db->tableName( 'logging' );
 261
 262                         if ( $this->buffer == WikiExporter::STREAM ) {
 263                                 $prev = $this->db->bufferResults( false );
 264                         }
 265                         $wrapper = null; // Assuring $wrapper is not undefined, if exception occurs early
 266                         try {
 267                                 $result = $this->db->select( array( 'logging', 'user' ),
 268                                         array( "{$logging}.*", 'user_name' ), // grab the user name
 269                                         $where,
 270                                         __METHOD__,
 271                                         array( 'ORDER BY' => 'log_id', 'USE INDEX' => array( 'logging' => 'PRIMARY' ) )
 272                                 );
 273                                 $wrapper = $this->db->resultObject( $result );
 274                                 $this->outputLogStream( $wrapper );
 275                                 if ( $this->buffer == WikiExporter::STREAM ) {
 276                                         $this->db->bufferResults( $prev );
 277                                 }
 278                         } catch ( Exception $e ) {
 279                                 // Throwing the exception does not reliably free the resultset, and
 280                                 // would also leave the connection in unbuffered mode.
 281
 282                                 // Freeing result
 283                                 try {
 284                                         if ( $wrapper ) {
 285                                                 $wrapper->free();
 286                                         }
 287                                 } catch ( Exception $e2 ) {
 288                                         // Already in panic mode -> ignoring $e2 as $e has
 289                                         // higher priority
 290                                 }
 291
 292                                 // Putting database back in previous buffer mode
 293                                 try {
 294                                         if ( $this->buffer == WikiExporter::STREAM ) {
 295                                                 $this->db->bufferResults( $prev );
 296                                         }
 297                                 } catch ( Exception $e2 ) {
 298                                         // Already in panic mode -> ignoring $e2 as $e has
 299                                         // higher priority
 300                                 }
 301
 302                                 // Inform caller about problem
 303                                 wfProfileOut( __METHOD__ );
 304                                 throw $e;
 305                         }
 306                 # For page dumps...
 307                 } else {
 308                         $tables = array( 'page', 'revision' );
 309                         $opts = array( 'ORDER BY' => 'page_id ASC' );
 310                         $opts['USE INDEX'] = array();
 311                         $join = array();
 312                         if ( is_array( $this->history ) ) {
 313                                 # Time offset/limit for all pages/history...
 314                                 $revJoin = 'page_id=rev_page';
 315                                 # Set time order
 316                                 if ( $this->history['dir'] == 'asc' ) {
 317                                         $op = '>';
 318                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 319                                 } else {
 320                                         $op = '<';
 321                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 322                                 }
 323                                 # Set offset
 324                                 if ( !empty( $this->history['offset'] ) ) {
 325                                         $revJoin .= " AND rev_timestamp $op " .
 326                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 327                                 }
 328                                 $join['revision'] = array( 'INNER JOIN', $revJoin );
 329                                 # Set query limit
 330                                 if ( !empty( $this->history['limit'] ) ) {
 331                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 332                                 }
 333                         } elseif ( $this->history & WikiExporter::FULL ) {
 334                                 # Full history dumps...
 335                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 336                         } elseif ( $this->history & WikiExporter::CURRENT ) {
 337                                 # Latest revision dumps...
 338                                 if ( $this->list_authors && $cond != '' ) { // List authors, if so desired
 339                                         $this->do_list_authors( $cond );
 340                                 }
 341                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 342                         } elseif ( $this->history & WikiExporter::STABLE ) {
 343                                 # "Stable" revision dumps...
 344                                 # Default JOIN, to be overridden...
 345                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 346                                 # One, and only one hook should set this, and return false
 347                                 if ( wfRunHooks( 'WikiExporter::dumpStableQuery', array( &$tables, &$opts, &$join ) ) ) {
 348                                         wfProfileOut( __METHOD__ );
 349                                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 350                                 }
 351                         } elseif ( $this->history & WikiExporter::RANGE ) {
 352                                 # Dump of revisions within a specified range
 353                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 354                                 $opts['ORDER BY'] = array( 'rev_page ASC', 'rev_id ASC' );
 355                         } else {
 356                                 # Unknown history specification parameter?
 357                                 wfProfileOut( __METHOD__ );
 358                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 359                         }
 360                         # Query optimization hacks
 361                         if ( $cond == '' ) {
 362                                 $opts[] = 'STRAIGHT_JOIN';
 363                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 364                         }
 365                         # Build text join options
 366                         if ( $this->text != WikiExporter::STUB ) { // 1-pass
 367                                 $tables[] = 'text';
 368                                 $join['text'] = array( 'INNER JOIN', 'rev_text_id=old_id' );
 369                         }
 370
 371                         if ( $this->buffer == WikiExporter::STREAM ) {
 372                                 $prev = $this->db->bufferResults( false );
 373                         }
 374
 375                         $wrapper = null; // Assuring $wrapper is not undefined, if exception occurs early
 376                         try {
 377                                 wfRunHooks( 'ModifyExportQuery',
 378                                                 array( $this->db, &$tables, &$cond, &$opts, &$join ) );
 379
 380                                 # Do the query!
 381                                 $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 382                                 $wrapper = $this->db->resultObject( $result );
 383                                 # Output dump results
 384                                 $this->outputPageStream( $wrapper );
 385
 386                                 if ( $this->buffer == WikiExporter::STREAM ) {
 387                                         $this->db->bufferResults( $prev );
 388                                 }
 389                         } catch ( Exception $e ) {
 390                                 // Throwing the exception does not reliably free the resultset, and
 391                                 // would also leave the connection in unbuffered mode.
 392
 393                                 // Freeing result
 394                                 try {
 395                                         if ( $wrapper ) {
 396                                                 $wrapper->free();
 397                                         }
 398                                 } catch ( Exception $e2 ) {
 399                                         // Already in panic mode -> ignoring $e2 as $e has
 400                                         // higher priority
 401                                 }
 402
 403                                 // Putting database back in previous buffer mode
 404                                 try {
 405                                         if ( $this->buffer == WikiExporter::STREAM ) {
 406                                                 $this->db->bufferResults( $prev );
 407                                         }
 408                                 } catch ( Exception $e2 ) {
 409                                         // Already in panic mode -> ignoring $e2 as $e has
 410                                         // higher priority
 411                                 }
 412
 413                                 // Inform caller about problem
 414                                 throw $e;
 415                         }
 416                 }
 417                 wfProfileOut( __METHOD__ );
 418         }
 419
 420         /**
 421          * Runs through a query result set dumping page and revision records.
 422          * The result set should be sorted/grouped by page to avoid duplicate
 423          * page records in the output.
 424          *
 425          * Should be safe for
 426          * streaming (non-buffered) queries, as long as it was made on a
 427          * separate database connection not managed by LoadBalancer; some
 428          * blob storage types will make queries to pull source data.
 429          *
 430          * @param $resultset ResultWrapper
 431          */
 432         protected function outputPageStream( $resultset ) {
 433                 $last = null;
 434                 foreach ( $resultset as $row ) {
 435                         if ( $last === null ||
 436                                 $last->page_namespace != $row->page_namespace ||
 437                                 $last->page_title != $row->page_title ) {
 438                                 if ( $last !== null ) {
 439                                         $output = '';
 440                                         if ( $this->dumpUploads ) {
 441                                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 442                                         }
 443                                         $output .= $this->writer->closePage();
 444                                         $this->sink->writeClosePage( $output );
 445                                 }
 446                                 $output = $this->writer->openPage( $row );
 447                                 $this->sink->writeOpenPage( $row, $output );
 448                                 $last = $row;
 449                         }
 450                         $output = $this->writer->writeRevision( $row );
 451                         $this->sink->writeRevision( $row, $output );
 452                 }
 453                 if ( $last !== null ) {
 454                         $output = '';
 455                         if ( $this->dumpUploads ) {
 456                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 457                         }
 458                         $output .= $this->author_list;
 459                         $output .= $this->writer->closePage();
 460                         $this->sink->writeClosePage( $output );
 461                 }
 462         }
 463
 464         /**
 465          * @param $resultset array
 466          */
 467         protected function outputLogStream( $resultset ) {
 468                 foreach ( $resultset as $row ) {
 469                         $output = $this->writer->writeLogItem( $row );
 470                         $this->sink->writeLogItem( $row, $output );
 471                 }
 472         }
 473 }
 474
 475 /**
 476  * @ingroup Dump
 477  */
 478 class XmlDumpWriter {
 479         /**
 480          * Returns the export schema version.
 481          * @deprecated in 1.20; use WikiExporter::schemaVersion() instead
 482          * @return string
 483          */
 484         function schemaVersion() {
 485                 wfDeprecated( __METHOD__, '1.20' );
 486                 return WikiExporter::schemaVersion();
 487         }
 488
 489         /**
 490          * Opens the XML output stream's root "<mediawiki>" element.
 491          * This does not include an xml directive, so is safe to include
 492          * as a subelement in a larger XML stream. Namespace and XML Schema
 493          * references are included.
 494          *
 495          * Output will be encoded in UTF-8.
 496          *
 497          * @return string
 498          */
 499         function openStream() {
 500                 global $wgLanguageCode;
 501                 $ver = WikiExporter::schemaVersion();
 502                 return Xml::element( 'mediawiki', array(
 503                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 504                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 505                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 506                                                 "http://www.mediawiki.org/xml/export-$ver.xsd", #TODO: how do we get a new version up there?
 507                         'version'            => $ver,
 508                         'xml:lang'           => $wgLanguageCode ),
 509                         null ) .
 510                         "\n" .
 511                         $this->siteInfo();
 512         }
 513
 514         /**
 515          * @return string
 516          */
 517         function siteInfo() {
 518                 $info = array(
 519                         $this->sitename(),
 520                         $this->homelink(),
 521                         $this->generator(),
 522                         $this->caseSetting(),
 523                         $this->namespaces() );
 524                 return "  <siteinfo>\n    " .
 525                         implode( "\n    ", $info ) .
 526                         "\n  </siteinfo>\n";
 527         }
 528
 529         /**
 530          * @return string
 531          */
 532         function sitename() {
 533                 global $wgSitename;
 534                 return Xml::element( 'sitename', array(), $wgSitename );
 535         }
 536
 537         /**
 538          * @return string
 539          */
 540         function generator() {
 541                 global $wgVersion;
 542                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 543         }
 544
 545         /**
 546          * @return string
 547          */
 548         function homelink() {
 549                 return Xml::element( 'base', array(), Title::newMainPage()->getCanonicalURL() );
 550         }
 551
 552         /**
 553          * @return string
 554          */
 555         function caseSetting() {
 556                 global $wgCapitalLinks;
 557                 // "case-insensitive" option is reserved for future
 558                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 559                 return Xml::element( 'case', array(), $sensitivity );
 560         }
 561
 562         /**
 563          * @return string
 564          */
 565         function namespaces() {
 566                 global $wgContLang;
 567                 $spaces = "<namespaces>\n";
 568                 foreach ( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 569                         $spaces .= '      ' .
 570                                 Xml::element( 'namespace',
 571                                         array(
 572                                                 'key' => $ns,
 573                                                 'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 574                                         ), $title ) . "\n";
 575                 }
 576                 $spaces .= "    </namespaces>";
 577                 return $spaces;
 578         }
 579
 580         /**
 581          * Closes the output stream with the closing root element.
 582          * Call when finished dumping things.
 583          *
 584          * @return string
 585          */
 586         function closeStream() {
 587                 return "</mediawiki>\n";
 588         }
 589
 590         /**
 591          * Opens a "<page>" section on the output stream, with data
 592          * from the given database row.
 593          *
 594          * @param $row object
 595          * @return string
 596          * @access private
 597          */
 598         function openPage( $row ) {
 599                 $out = "  <page>\n";
 600                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 601                 $out .= '    ' . Xml::elementClean( 'title', array(), self::canonicalTitle( $title ) ) . "\n";
 602                 $out .= '    ' . Xml::element( 'ns', array(), strval( $row->page_namespace ) ) . "\n";
 603                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 604                 if ( $row->page_is_redirect ) {
 605                         $page = WikiPage::factory( $title );
 606                         $redirect = $page->getRedirectTarget();
 607                         if ( $redirect instanceOf Title && $redirect->isValidRedirectTarget() ) {
 608                                 $out .= '    ' . Xml::element( 'redirect', array( 'title' => self::canonicalTitle( $redirect ) ) ) . "\n";
 609                         }
 610                 }
 611
 612                 if ( $row->page_restrictions != '' ) {
 613                         $out .= '    ' . Xml::element( 'restrictions', array(),
 614                                 strval( $row->page_restrictions ) ) . "\n";
 615                 }
 616
 617                 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 618
 619                 return $out;
 620         }
 621
 622         /**
 623          * Closes a "<page>" section on the output stream.
 624          *
 625          * @access private
 626          * @return string
 627          */
 628         function closePage() {
 629                 return "  </page>\n";
 630         }
 631
 632         /**
 633          * Dumps a "<revision>" section on the output stream, with
 634          * data filled in from the given database row.
 635          *
 636          * @param $row object
 637          * @return string
 638          * @access private
 639          */
 640         function writeRevision( $row ) {
 641                 wfProfileIn( __METHOD__ );
 642
 643                 $out = "    <revision>\n";
 644                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 645                 if ( isset( $row->rev_parent_id ) && $row->rev_parent_id ) {
 646                         $out .= "      " . Xml::element( 'parentid', null, strval( $row->rev_parent_id ) ) . "\n";
 647                 }
 648
 649                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 650
 651                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_USER ) ) {
 652                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 653                 } else {
 654                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 655                 }
 656
 657                 if ( isset( $row->rev_minor_edit ) && $row->rev_minor_edit ) {
 658                         $out .= "      <minor/>\n";
 659                 }
 660                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_COMMENT ) ) {
 661                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 662                 } elseif ( $row->rev_comment != '' ) {
 663                         $out .= "      " . Xml::elementClean( 'comment', array(), strval( $row->rev_comment ) ) . "\n";
 664                 }
 665
 666                 $text = '';
 667                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_TEXT ) ) {
 668                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 669                 } elseif ( isset( $row->old_text ) ) {
 670                         // Raw text from the database may have invalid chars
 671                         $text = strval( Revision::getRevisionText( $row ) );
 672                         $out .= "      " . Xml::elementClean( 'text',
 673                                 array( 'xml:space' => 'preserve', 'bytes' => intval( $row->rev_len ) ),
 674                                 strval( $text ) ) . "\n";
 675                 } else {
 676                         // Stub output
 677                         $out .= "      " . Xml::element( 'text',
 678                                 array( 'id' => $row->rev_text_id, 'bytes' => intval( $row->rev_len ) ),
 679                                 "" ) . "\n";
 680                 }
 681
 682                 if ( isset( $row->rev_sha1 ) && $row->rev_sha1 && !( $row->rev_deleted & Revision::DELETED_TEXT ) ) {
 683                         $out .= "      " . Xml::element( 'sha1', null, strval( $row->rev_sha1 ) ) . "\n";
 684                 } else {
 685                         $out .= "      <sha1/>\n";
 686                 }
 687
 688                 if ( isset( $row->rev_content_model ) && !is_null( $row->rev_content_model ) ) {
 689                         $content_model = strval( $row->rev_content_model );
 690                 } else {
 691                         // probably using $wgContentHandlerUseDB = false;
 692                         // @todo test!
 693                         $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 694                         $content_model = ContentHandler::getDefaultModelFor( $title );
 695                 }
 696
 697                 $out .= "      " . Xml::element( 'model', null, strval( $content_model ) ) . "\n";
 698
 699                 if ( isset( $row->rev_content_format ) && !is_null( $row->rev_content_format ) ) {
 700                         $content_format = strval( $row->rev_content_format );
 701                 } else {
 702                         // probably using $wgContentHandlerUseDB = false;
 703                         // @todo test!
 704                         $content_handler = ContentHandler::getForModelID( $content_model );
 705                         $content_format = $content_handler->getDefaultFormat();
 706                 }
 707
 708                 $out .= "      " . Xml::element( 'format', null, strval( $content_format ) ) . "\n";
 709
 710                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 711
 712                 $out .= "    </revision>\n";
 713
 714                 wfProfileOut( __METHOD__ );
 715                 return $out;
 716         }
 717
 718         /**
 719          * Dumps a "<logitem>" section on the output stream, with
 720          * data filled in from the given database row.
 721          *
 722          * @param $row object
 723          * @return string
 724          * @access private
 725          */
 726         function writeLogItem( $row ) {
 727                 wfProfileIn( __METHOD__ );
 728
 729                 $out = "  <logitem>\n";
 730                 $out .= "    " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 731
 732                 $out .= $this->writeTimestamp( $row->log_timestamp, "    " );
 733
 734                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 735                         $out .= "    " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 736                 } else {
 737                         $out .= $this->writeContributor( $row->log_user, $row->user_name, "    " );
 738                 }
 739
 740                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 741                         $out .= "    " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 742                 } elseif ( $row->log_comment != '' ) {
 743                         $out .= "    " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 744                 }
 745
 746                 $out .= "    " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 747                 $out .= "    " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 748
 749                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 750                         $out .= "    " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 751                 } else {
 752                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 753                         $out .= "    " . Xml::elementClean( 'logtitle', null, self::canonicalTitle( $title ) ) . "\n";
 754                         $out .= "    " . Xml::elementClean( 'params',
 755                                 array( 'xml:space' => 'preserve' ),
 756                                 strval( $row->log_params ) ) . "\n";
 757                 }
 758
 759                 $out .= "  </logitem>\n";
 760
 761                 wfProfileOut( __METHOD__ );
 762                 return $out;
 763         }
 764
 765         /**
 766          * @param $timestamp string
 767          * @param string $indent Default to six spaces
 768          * @return string
 769          */
 770         function writeTimestamp( $timestamp, $indent = "      " ) {
 771                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 772                 return $indent . Xml::element( 'timestamp', null, $ts ) . "\n";
 773         }
 774
 775         /**
 776          * @param $id
 777          * @param $text string
 778          * @param string $indent Default to six spaces
 779          * @return string
 780          */
 781         function writeContributor( $id, $text, $indent = "      " ) {
 782                 $out = $indent . "<contributor>\n";
 783                 if ( $id || !IP::isValid( $text ) ) {
 784                         $out .= $indent . "  " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 785                         $out .= $indent . "  " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 786                 } else {
 787                         $out .= $indent . "  " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 788                 }
 789                 $out .= $indent . "</contributor>\n";
 790                 return $out;
 791         }
 792
 793         /**
 794          * Warning! This data is potentially inconsistent. :(
 795          * @param $row
 796          * @param $dumpContents bool
 797          * @return string
 798          */
 799         function writeUploads( $row, $dumpContents = false ) {
 800                 if ( $row->page_namespace == NS_FILE ) {
 801                         $img = wfLocalFile( $row->page_title );
 802                         if ( $img && $img->exists() ) {
 803                                 $out = '';
 804                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 805                                         $out .= $this->writeUpload( $ver, $dumpContents );
 806                                 }
 807                                 $out .= $this->writeUpload( $img, $dumpContents );
 808                                 return $out;
 809                         }
 810                 }
 811                 return '';
 812         }
 813
 814         /**
 815          * @param $file File
 816          * @param $dumpContents bool
 817          * @return string
 818          */
 819         function writeUpload( $file, $dumpContents = false ) {
 820                 if ( $file->isOld() ) {
 821                         $archiveName = "      " .
 822                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 823                 } else {
 824                         $archiveName = '';
 825                 }
 826                 if ( $dumpContents ) {
 827                         # Dump file as base64
 828                         # Uses only XML-safe characters, so does not need escaping
 829                         $contents = '      <contents encoding="base64">' .
 830                                 chunk_split( base64_encode( file_get_contents( $file->getPath() ) ) ) .
 831                                 "      </contents>\n";
 832                 } else {
 833                         $contents = '';
 834                 }
 835                 if ( $file->isDeleted( File::DELETED_COMMENT ) ) {
 836                         $comment = Xml::element( 'comment', array( 'deleted' => 'deleted' ) );
 837                 } else {
 838                         $comment = Xml::elementClean( 'comment', null, $file->getDescription() );
 839                 }
 840                 return "    <upload>\n" .
 841                         $this->writeTimestamp( $file->getTimestamp() ) .
 842                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 843                         "      " . $comment . "\n" .
 844                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 845                         $archiveName .
 846                         "      " . Xml::element( 'src', null, $file->getCanonicalURL() ) . "\n" .
 847                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 848                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 849                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 850                         $contents .
 851                         "    </upload>\n";
 852         }
 853
 854         /**
 855          * Return prefixed text form of title, but using the content language's
 856          * canonical namespace. This skips any special-casing such as gendered
 857          * user namespaces -- which while useful, are not yet listed in the
 858          * XML "<siteinfo>" data so are unsafe in export.
 859          *
 860          * @param Title $title
 861          * @return string
 862          * @since 1.18
 863          */
 864         public static function canonicalTitle( Title $title ) {
 865                 if ( $title->getInterwiki() ) {
 866                         return $title->getPrefixedText();
 867                 }
 868
 869                 global $wgContLang;
 870                 $prefix = str_replace( '_', ' ', $wgContLang->getNsText( $title->getNamespace() ) );
 871
 872                 if ( $prefix !== '' ) {
 873                         $prefix .= ':';
 874                 }
 875
 876                 return $prefix . $title->getText();
 877         }
 878 }
 879
 880 /**
 881  * Base class for output stream; prints to stdout or buffer or wherever.
 882  * @ingroup Dump
 883  */
 884 class DumpOutput {
 885
 886         /**
 887          * @param $string string
 888          */
 889         function writeOpenStream( $string ) {
 890                 $this->write( $string );
 891         }
 892
 893         /**
 894          * @param $string string
 895          */
 896         function writeCloseStream( $string ) {
 897                 $this->write( $string );
 898         }
 899
 900         /**
 901          * @param $page
 902          * @param $string string
 903          */
 904         function writeOpenPage( $page, $string ) {
 905                 $this->write( $string );
 906         }
 907
 908         /**
 909          * @param $string string
 910          */
 911         function writeClosePage( $string ) {
 912                 $this->write( $string );
 913         }
 914
 915         /**
 916          * @param $rev
 917          * @param $string string
 918          */
 919         function writeRevision( $rev, $string ) {
 920                 $this->write( $string );
 921         }
 922
 923         /**
 924          * @param $rev
 925          * @param $string string
 926          */
 927         function writeLogItem( $rev, $string ) {
 928                 $this->write( $string );
 929         }
 930
 931         /**
 932          * Override to write to a different stream type.
 933          * @param $string string
 934          * @return bool
 935          */
 936         function write( $string ) {
 937                 print $string;
 938         }
 939
 940         /**
 941          * Close the old file, move it to a specified name,
 942          * and reopen new file with the old name. Use this
 943          * for writing out a file in multiple pieces
 944          * at specified checkpoints (e.g. every n hours).
 945          * @param $newname mixed File name. May be a string or an array with one element
 946          */
 947         function closeRenameAndReopen( $newname ) {
 948         }
 949
 950         /**
 951          * Close the old file, and move it to a specified name.
 952          * Use this for the last piece of a file written out
 953          * at specified checkpoints (e.g. every n hours).
 954          * @param $newname mixed File name. May be a string or an array with one element
 955          * @param bool $open If true, a new file with the old filename will be opened again for writing (default: false)
 956          */
 957         function closeAndRename( $newname, $open = false ) {
 958         }
 959
 960         /**
 961          * Returns the name of the file or files which are
 962          * being written to, if there are any.
 963          * @return null
 964          */
 965         function getFilenames() {
 966                 return null;
 967         }
 968 }
 969
 970 /**
 971  * Stream outputter to send data to a file.
 972  * @ingroup Dump
 973  */
 974 class DumpFileOutput extends DumpOutput {
 975         protected $handle = false, $filename;
 976
 977         /**
 978          * @param $file
 979          */
 980         function __construct( $file ) {
 981                 $this->handle = fopen( $file, "wt" );
 982                 $this->filename = $file;
 983         }
 984
 985         /**
 986          * @param $string string
 987          */
 988         function writeCloseStream( $string ) {
 989                 parent::writeCloseStream( $string );
 990                 if ( $this->handle ) {
 991                         fclose( $this->handle );
 992                         $this->handle = false;
 993                 }
 994         }
 995
 996         /**
 997          * @param $string string
 998          */
 999         function write( $string ) {
1000                 fputs( $this->handle, $string );
1001         }
1002
1003         /**
1004          * @param $newname
1005          */
1006         function closeRenameAndReopen( $newname ) {
1007                 $this->closeAndRename( $newname, true );
1008         }
1009
1010         /**
1011          * @param $newname
1012          * @throws MWException
1013          */
1014         function renameOrException( $newname ) {
1015                         if ( !rename( $this->filename, $newname ) ) {
1016                                 throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" );
1017                         }
1018         }
1019
1020         /**
1021          * @param $newname array
1022          * @return mixed
1023          * @throws MWException
1024          */
1025         function checkRenameArgCount( $newname ) {
1026                 if ( is_array( $newname ) ) {
1027                         if ( count( $newname ) > 1 ) {
1028                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
1029                         } else {
1030                                 $newname = $newname[0];
1031                         }
1032                 }
1033                 return $newname;
1034         }
1035
1036         /**
1037          * @param $newname mixed
1038          * @param $open bool
1039          */
1040         function closeAndRename( $newname, $open = false ) {
1041                 $newname = $this->checkRenameArgCount( $newname );
1042                 if ( $newname ) {
1043                         if ( $this->handle ) {
1044                                 fclose( $this->handle );
1045                                 $this->handle = false;
1046                         }
1047                         $this->renameOrException( $newname );
1048                         if ( $open ) {
1049                                 $this->handle = fopen( $this->filename, "wt" );
1050                         }
1051                 }
1052         }
1053
1054         /**
1055          * @return string|null
1056          */
1057         function getFilenames() {
1058                 return $this->filename;
1059         }
1060 }
1061
1062 /**
1063  * Stream outputter to send data to a file via some filter program.
1064  * Even if compression is available in a library, using a separate
1065  * program can allow us to make use of a multi-processor system.
1066  * @ingroup Dump
1067  */
1068 class DumpPipeOutput extends DumpFileOutput {
1069         protected $command, $filename;
1070         protected $procOpenResource = false;
1071
1072         /**
1073          * @param $command
1074          * @param $file null
1075          */
1076         function __construct( $command, $file = null ) {
1077                 if ( !is_null( $file ) ) {
1078                         $command .= " > " . wfEscapeShellArg( $file );
1079                 }
1080
1081                 $this->startCommand( $command );
1082                 $this->command = $command;
1083                 $this->filename = $file;
1084         }
1085
1086         /**
1087          * @param $string string
1088          */
1089         function writeCloseStream( $string ) {
1090                 parent::writeCloseStream( $string );
1091                 if ( $this->procOpenResource ) {
1092                         proc_close( $this->procOpenResource );
1093                         $this->procOpenResource = false;
1094                 }
1095         }
1096
1097         /**
1098          * @param $command
1099          */
1100         function startCommand( $command ) {
1101                 $spec = array(
1102                         0 => array( "pipe", "r" ),
1103                 );
1104                 $pipes = array();
1105                 $this->procOpenResource = proc_open( $command, $spec, $pipes );
1106                 $this->handle = $pipes[0];
1107         }
1108
1109         /**
1110          * @param mixed $newname
1111          */
1112         function closeRenameAndReopen( $newname ) {
1113                 $this->closeAndRename( $newname, true );
1114         }
1115
1116         /**
1117          * @param $newname mixed
1118          * @param $open bool
1119          */
1120         function closeAndRename( $newname, $open = false ) {
1121                 $newname = $this->checkRenameArgCount( $newname );
1122                 if ( $newname ) {
1123                         if ( $this->handle ) {
1124                                 fclose( $this->handle );
1125                                 $this->handle = false;
1126                         }
1127                         if ( $this->procOpenResource ) {
1128                                 proc_close( $this->procOpenResource );
1129                                 $this->procOpenResource = false;
1130                         }
1131                         $this->renameOrException( $newname );
1132                         if ( $open ) {
1133                                 $command = $this->command;
1134                                 $command .= " > " . wfEscapeShellArg( $this->filename );
1135                                 $this->startCommand( $command );
1136                         }
1137                 }
1138         }
1139
1140 }
1141
1142 /**
1143  * Sends dump output via the gzip compressor.
1144  * @ingroup Dump
1145  */
1146 class DumpGZipOutput extends DumpPipeOutput {
1147
1148         /**
1149          * @param $file string
1150          */
1151         function __construct( $file ) {
1152                 parent::__construct( "gzip", $file );
1153         }
1154 }
1155
1156 /**
1157  * Sends dump output via the bgzip2 compressor.
1158  * @ingroup Dump
1159  */
1160 class DumpBZip2Output extends DumpPipeOutput {
1161
1162         /**
1163          * @param $file string
1164          */
1165         function __construct( $file ) {
1166                 parent::__construct( "bzip2", $file );
1167         }
1168 }
1169
1170 /**
1171  * Sends dump output via the p7zip compressor.
1172  * @ingroup Dump
1173  */
1174 class Dump7ZipOutput extends DumpPipeOutput {
1175
1176         /**
1177          * @param $file string
1178          */
1179         function __construct( $file ) {
1180                 $command = $this->setup7zCommand( $file );
1181                 parent::__construct( $command );
1182                 $this->filename = $file;
1183         }
1184
1185         /**
1186          * @param $file string
1187          * @return string
1188          */
1189         function setup7zCommand( $file ) {
1190                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
1191                 // Suppress annoying useless crap from p7zip
1192                 // Unfortunately this could suppress real error messages too
1193                 $command .= ' >' . wfGetNull() . ' 2>&1';
1194                 return $command;
1195         }
1196
1197         /**
1198          * @param $newname string
1199          * @param $open bool
1200          */
1201         function closeAndRename( $newname, $open = false ) {
1202                 $newname = $this->checkRenameArgCount( $newname );
1203                 if ( $newname ) {
1204                         fclose( $this->handle );
1205                         proc_close( $this->procOpenResource );
1206                         $this->renameOrException( $newname );
1207                         if ( $open ) {
1208                                 $command = $this->setup7zCommand( $this->filename );
1209                                 $this->startCommand( $command );
1210                         }
1211                 }
1212         }
1213 }
1214
1215 /**
1216  * Dump output filter class.
1217  * This just does output filtering and streaming; XML formatting is done
1218  * higher up, so be careful in what you do.
1219  * @ingroup Dump
1220  */
1221 class DumpFilter {
1222
1223         /**
1224          * @var DumpOutput
1225          * FIXME will need to be made protected whenever legacy code
1226          * is updated.
1227          */
1228         public $sink;
1229
1230         /**
1231          * @var bool
1232          */
1233         protected $sendingThisPage;
1234
1235         /**
1236          * @param $sink DumpOutput
1237          */
1238         function __construct( &$sink ) {
1239                 $this->sink =& $sink;
1240         }
1241
1242         /**
1243          * @param $string string
1244          */
1245         function writeOpenStream( $string ) {
1246                 $this->sink->writeOpenStream( $string );
1247         }
1248
1249         /**
1250          * @param $string string
1251          */
1252         function writeCloseStream( $string ) {
1253                 $this->sink->writeCloseStream( $string );
1254         }
1255
1256         /**
1257          * @param $page
1258          * @param $string string
1259          */
1260         function writeOpenPage( $page, $string ) {
1261                 $this->sendingThisPage = $this->pass( $page, $string );
1262                 if ( $this->sendingThisPage ) {
1263                         $this->sink->writeOpenPage( $page, $string );
1264                 }
1265         }
1266
1267         /**
1268          * @param $string string
1269          */
1270         function writeClosePage( $string ) {
1271                 if ( $this->sendingThisPage ) {
1272                         $this->sink->writeClosePage( $string );
1273                         $this->sendingThisPage = false;
1274                 }
1275         }
1276
1277         /**
1278          * @param $rev
1279          * @param $string string
1280          */
1281         function writeRevision( $rev, $string ) {
1282                 if ( $this->sendingThisPage ) {
1283                         $this->sink->writeRevision( $rev, $string );
1284                 }
1285         }
1286
1287         /**
1288          * @param $rev
1289          * @param $string string
1290          */
1291         function writeLogItem( $rev, $string ) {
1292                 $this->sink->writeRevision( $rev, $string );
1293         }
1294
1295         /**
1296          * @param $newname string
1297          */
1298         function closeRenameAndReopen( $newname ) {
1299                 $this->sink->closeRenameAndReopen( $newname );
1300         }
1301
1302         /**
1303          * @param $newname string
1304          * @param $open bool
1305          */
1306         function closeAndRename( $newname, $open = false ) {
1307                 $this->sink->closeAndRename( $newname, $open );
1308         }
1309
1310         /**
1311          * @return array
1312          */
1313         function getFilenames() {
1314                 return $this->sink->getFilenames();
1315         }
1316
1317         /**
1318          * Override for page-based filter types.
1319          * @param $page
1320          * @return bool
1321          */
1322         function pass( $page ) {
1323                 return true;
1324         }
1325 }
1326
1327 /**
1328  * Simple dump output filter to exclude all talk pages.
1329  * @ingroup Dump
1330  */
1331 class DumpNotalkFilter extends DumpFilter {
1332
1333         /**
1334          * @param $page
1335          * @return bool
1336          */
1337         function pass( $page ) {
1338                 return !MWNamespace::isTalk( $page->page_namespace );
1339         }
1340 }
1341
1342 /**
1343  * Dump output filter to include or exclude pages in a given set of namespaces.
1344  * @ingroup Dump
1345  */
1346 class DumpNamespaceFilter extends DumpFilter {
1347         var $invert = false;
1348         var $namespaces = array();
1349
1350         /**
1351          * @param $sink DumpOutput
1352          * @param $param
1353          * @throws MWException
1354          */
1355         function __construct( &$sink, $param ) {
1356                 parent::__construct( $sink );
1357
1358                 $constants = array(
1359                         "NS_MAIN"           => NS_MAIN,
1360                         "NS_TALK"           => NS_TALK,
1361                         "NS_USER"           => NS_USER,
1362                         "NS_USER_TALK"      => NS_USER_TALK,
1363                         "NS_PROJECT"        => NS_PROJECT,
1364                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
1365                         "NS_FILE"           => NS_FILE,
1366                         "NS_FILE_TALK"      => NS_FILE_TALK,
1367                         "NS_IMAGE"          => NS_IMAGE, // NS_IMAGE is an alias for NS_FILE
1368                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
1369                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
1370                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
1371                         "NS_TEMPLATE"       => NS_TEMPLATE,
1372                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
1373                         "NS_HELP"           => NS_HELP,
1374                         "NS_HELP_TALK"      => NS_HELP_TALK,
1375                         "NS_CATEGORY"       => NS_CATEGORY,
1376                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
1377
1378                 if ( $param { 0 } == '!' ) {
1379                         $this->invert = true;
1380                         $param = substr( $param, 1 );
1381                 }
1382
1383                 foreach ( explode( ',', $param ) as $key ) {
1384                         $key = trim( $key );
1385                         if ( isset( $constants[$key] ) ) {
1386                                 $ns = $constants[$key];
1387                                 $this->namespaces[$ns] = true;
1388                         } elseif ( is_numeric( $key ) ) {
1389                                 $ns = intval( $key );
1390                                 $this->namespaces[$ns] = true;
1391                         } else {
1392                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
1393                         }
1394                 }
1395         }
1396
1397         /**
1398          * @param $page
1399          * @return bool
1400          */
1401         function pass( $page ) {
1402                 $match = isset( $this->namespaces[$page->page_namespace] );
1403                 return $this->invert xor $match;
1404         }
1405 }
1406
1407 /**
1408  * Dump output filter to include only the last revision in each page sequence.
1409  * @ingroup Dump
1410  */
1411 class DumpLatestFilter extends DumpFilter {
1412         var $page, $pageString, $rev, $revString;
1413
1414         /**
1415          * @param $page
1416          * @param $string string
1417          */
1418         function writeOpenPage( $page, $string ) {
1419                 $this->page = $page;
1420                 $this->pageString = $string;
1421         }
1422
1423         /**
1424          * @param $string string
1425          */
1426         function writeClosePage( $string ) {
1427                 if ( $this->rev ) {
1428                         $this->sink->writeOpenPage( $this->page, $this->pageString );
1429                         $this->sink->writeRevision( $this->rev, $this->revString );
1430                         $this->sink->writeClosePage( $string );
1431                 }
1432                 $this->rev = null;
1433                 $this->revString = null;
1434                 $this->page = null;
1435                 $this->pageString = null;
1436         }
1437
1438         /**
1439          * @param $rev
1440          * @param $string string
1441          */
1442         function writeRevision( $rev, $string ) {
1443                 if ( $rev->rev_id == $this->page->page_latest ) {
1444                         $this->rev = $rev;
1445                         $this->revString = $string;
1446                 }
1447         }
1448 }
1449
1450 /**
1451  * Base class for output stream; prints to stdout or buffer or wherever.
1452  * @ingroup Dump
1453  */
1454 class DumpMultiWriter {
1455
1456         /**
1457          * @param $sinks
1458          */
1459         function __construct( $sinks ) {
1460                 $this->sinks = $sinks;
1461                 $this->count = count( $sinks );
1462         }
1463
1464         /**
1465          * @param $string string
1466          */
1467         function writeOpenStream( $string ) {
1468                 for ( $i = 0; $i < $this->count; $i++ ) {
1469                         $this->sinks[$i]->writeOpenStream( $string );
1470                 }
1471         }
1472
1473         /**
1474          * @param $string string
1475          */
1476         function writeCloseStream( $string ) {
1477                 for ( $i = 0; $i < $this->count; $i++ ) {
1478                         $this->sinks[$i]->writeCloseStream( $string );
1479                 }
1480         }
1481
1482         /**
1483          * @param $page
1484          * @param $string string
1485          */
1486         function writeOpenPage( $page, $string ) {
1487                 for ( $i = 0; $i < $this->count; $i++ ) {
1488                         $this->sinks[$i]->writeOpenPage( $page, $string );
1489                 }
1490         }
1491
1492         /**
1493          * @param $string
1494          */
1495         function writeClosePage( $string ) {
1496                 for ( $i = 0; $i < $this->count; $i++ ) {
1497                         $this->sinks[$i]->writeClosePage( $string );
1498                 }
1499         }
1500
1501         /**
1502          * @param $rev
1503          * @param $string
1504          */
1505         function writeRevision( $rev, $string ) {
1506                 for ( $i = 0; $i < $this->count; $i++ ) {
1507                         $this->sinks[$i]->writeRevision( $rev, $string );
1508                 }
1509         }
1510
1511         /**
1512          * @param $newnames
1513          */
1514         function closeRenameAndReopen( $newnames ) {
1515                 $this->closeAndRename( $newnames, true );
1516         }
1517
1518         /**
1519          * @param $newnames array
1520          * @param bool $open
1521          */
1522         function closeAndRename( $newnames, $open = false ) {
1523                 for ( $i = 0; $i < $this->count; $i++ ) {
1524                         $this->sinks[$i]->closeAndRename( $newnames[$i], $open );
1525                 }
1526         }
1527
1528         /**
1529          * @return array
1530          */
1531         function getFilenames() {
1532                 $filenames = array();
1533                 for ( $i = 0; $i < $this->count; $i++ ) {
1534                         $filenames[] = $this->sinks[$i]->getFilenames();
1535                 }
1536                 return $filenames;
1537         }
1538
1539 }
1540
1541 /**
1542  * @param $string string
1543  * @return string
1544  */
1545 function xmlsafe( $string ) {
1546         wfProfileIn( __FUNCTION__ );
1547
1548         /**
1549          * The page may contain old data which has not been properly normalized.
1550          * Invalid UTF-8 sequences or forbidden control characters will make our
1551          * XML output invalid, so be sure to strip them out.
1552          */
1553         $string = UtfNormal::cleanUp( $string );
1554
1555         $string = htmlspecialchars( $string );
1556         wfProfileOut( __FUNCTION__ );
1557         return $string;
1558 }