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                         $be = $file->getRepo()->getBackend();
 828                         # Dump file as base64
 829                         # Uses only XML-safe characters, so does not need escaping
 830                         # @TODO: too bad this loads the contents into memory (script might swap)
 831                         $contents = '      <contents encoding="base64">' .
 832                                 chunk_split( base64_encode(
 833                                         $be->getFileContents( array( 'src' => $file->getPath() ) ) ) ) .
 834                                 "      </contents>\n";
 835                 } else {
 836                         $contents = '';
 837                 }
 838                 if ( $file->isDeleted( File::DELETED_COMMENT ) ) {
 839                         $comment = Xml::element( 'comment', array( 'deleted' => 'deleted' ) );
 840                 } else {
 841                         $comment = Xml::elementClean( 'comment', null, $file->getDescription() );
 842                 }
 843                 return "    <upload>\n" .
 844                         $this->writeTimestamp( $file->getTimestamp() ) .
 845                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 846                         "      " . $comment . "\n" .
 847                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 848                         $archiveName .
 849                         "      " . Xml::element( 'src', null, $file->getCanonicalURL() ) . "\n" .
 850                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 851                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 852                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 853                         $contents .
 854                         "    </upload>\n";
 855         }
 856
 857         /**
 858          * Return prefixed text form of title, but using the content language's
 859          * canonical namespace. This skips any special-casing such as gendered
 860          * user namespaces -- which while useful, are not yet listed in the
 861          * XML "<siteinfo>" data so are unsafe in export.
 862          *
 863          * @param Title $title
 864          * @return string
 865          * @since 1.18
 866          */
 867         public static function canonicalTitle( Title $title ) {
 868                 if ( $title->getInterwiki() ) {
 869                         return $title->getPrefixedText();
 870                 }
 871
 872                 global $wgContLang;
 873                 $prefix = str_replace( '_', ' ', $wgContLang->getNsText( $title->getNamespace() ) );
 874
 875                 if ( $prefix !== '' ) {
 876                         $prefix .= ':';
 877                 }
 878
 879                 return $prefix . $title->getText();
 880         }
 881 }
 882
 883 /**
 884  * Base class for output stream; prints to stdout or buffer or wherever.
 885  * @ingroup Dump
 886  */
 887 class DumpOutput {
 888
 889         /**
 890          * @param $string string
 891          */
 892         function writeOpenStream( $string ) {
 893                 $this->write( $string );
 894         }
 895
 896         /**
 897          * @param $string string
 898          */
 899         function writeCloseStream( $string ) {
 900                 $this->write( $string );
 901         }
 902
 903         /**
 904          * @param $page
 905          * @param $string string
 906          */
 907         function writeOpenPage( $page, $string ) {
 908                 $this->write( $string );
 909         }
 910
 911         /**
 912          * @param $string string
 913          */
 914         function writeClosePage( $string ) {
 915                 $this->write( $string );
 916         }
 917
 918         /**
 919          * @param $rev
 920          * @param $string string
 921          */
 922         function writeRevision( $rev, $string ) {
 923                 $this->write( $string );
 924         }
 925
 926         /**
 927          * @param $rev
 928          * @param $string string
 929          */
 930         function writeLogItem( $rev, $string ) {
 931                 $this->write( $string );
 932         }
 933
 934         /**
 935          * Override to write to a different stream type.
 936          * @param $string string
 937          * @return bool
 938          */
 939         function write( $string ) {
 940                 print $string;
 941         }
 942
 943         /**
 944          * Close the old file, move it to a specified name,
 945          * and reopen new file with the old name. Use this
 946          * for writing out a file in multiple pieces
 947          * at specified checkpoints (e.g. every n hours).
 948          * @param $newname mixed File name. May be a string or an array with one element
 949          */
 950         function closeRenameAndReopen( $newname ) {
 951         }
 952
 953         /**
 954          * Close the old file, and move it to a specified name.
 955          * Use this for the last piece of a file written out
 956          * at specified checkpoints (e.g. every n hours).
 957          * @param $newname mixed File name. May be a string or an array with one element
 958          * @param bool $open If true, a new file with the old filename will be opened again for writing (default: false)
 959          */
 960         function closeAndRename( $newname, $open = false ) {
 961         }
 962
 963         /**
 964          * Returns the name of the file or files which are
 965          * being written to, if there are any.
 966          * @return null
 967          */
 968         function getFilenames() {
 969                 return null;
 970         }
 971 }
 972
 973 /**
 974  * Stream outputter to send data to a file.
 975  * @ingroup Dump
 976  */
 977 class DumpFileOutput extends DumpOutput {
 978         protected $handle = false, $filename;
 979
 980         /**
 981          * @param $file
 982          */
 983         function __construct( $file ) {
 984                 $this->handle = fopen( $file, "wt" );
 985                 $this->filename = $file;
 986         }
 987
 988         /**
 989          * @param $string string
 990          */
 991         function writeCloseStream( $string ) {
 992                 parent::writeCloseStream( $string );
 993                 if ( $this->handle ) {
 994                         fclose( $this->handle );
 995                         $this->handle = false;
 996                 }
 997         }
 998
 999         /**
1000          * @param $string string
1001          */
1002         function write( $string ) {
1003                 fputs( $this->handle, $string );
1004         }
1005
1006         /**
1007          * @param $newname
1008          */
1009         function closeRenameAndReopen( $newname ) {
1010                 $this->closeAndRename( $newname, true );
1011         }
1012
1013         /**
1014          * @param $newname
1015          * @throws MWException
1016          */
1017         function renameOrException( $newname ) {
1018                         if ( !rename( $this->filename, $newname ) ) {
1019                                 throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" );
1020                         }
1021         }
1022
1023         /**
1024          * @param $newname array
1025          * @return mixed
1026          * @throws MWException
1027          */
1028         function checkRenameArgCount( $newname ) {
1029                 if ( is_array( $newname ) ) {
1030                         if ( count( $newname ) > 1 ) {
1031                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
1032                         } else {
1033                                 $newname = $newname[0];
1034                         }
1035                 }
1036                 return $newname;
1037         }
1038
1039         /**
1040          * @param $newname mixed
1041          * @param $open bool
1042          */
1043         function closeAndRename( $newname, $open = false ) {
1044                 $newname = $this->checkRenameArgCount( $newname );
1045                 if ( $newname ) {
1046                         if ( $this->handle ) {
1047                                 fclose( $this->handle );
1048                                 $this->handle = false;
1049                         }
1050                         $this->renameOrException( $newname );
1051                         if ( $open ) {
1052                                 $this->handle = fopen( $this->filename, "wt" );
1053                         }
1054                 }
1055         }
1056
1057         /**
1058          * @return string|null
1059          */
1060         function getFilenames() {
1061                 return $this->filename;
1062         }
1063 }
1064
1065 /**
1066  * Stream outputter to send data to a file via some filter program.
1067  * Even if compression is available in a library, using a separate
1068  * program can allow us to make use of a multi-processor system.
1069  * @ingroup Dump
1070  */
1071 class DumpPipeOutput extends DumpFileOutput {
1072         protected $command, $filename;
1073         protected $procOpenResource = false;
1074
1075         /**
1076          * @param $command
1077          * @param $file null
1078          */
1079         function __construct( $command, $file = null ) {
1080                 if ( !is_null( $file ) ) {
1081                         $command .= " > " . wfEscapeShellArg( $file );
1082                 }
1083
1084                 $this->startCommand( $command );
1085                 $this->command = $command;
1086                 $this->filename = $file;
1087         }
1088
1089         /**
1090          * @param $string string
1091          */
1092         function writeCloseStream( $string ) {
1093                 parent::writeCloseStream( $string );
1094                 if ( $this->procOpenResource ) {
1095                         proc_close( $this->procOpenResource );
1096                         $this->procOpenResource = false;
1097                 }
1098         }
1099
1100         /**
1101          * @param $command
1102          */
1103         function startCommand( $command ) {
1104                 $spec = array(
1105                         0 => array( "pipe", "r" ),
1106                 );
1107                 $pipes = array();
1108                 $this->procOpenResource = proc_open( $command, $spec, $pipes );
1109                 $this->handle = $pipes[0];
1110         }
1111
1112         /**
1113          * @param mixed $newname
1114          */
1115         function closeRenameAndReopen( $newname ) {
1116                 $this->closeAndRename( $newname, true );
1117         }
1118
1119         /**
1120          * @param $newname mixed
1121          * @param $open bool
1122          */
1123         function closeAndRename( $newname, $open = false ) {
1124                 $newname = $this->checkRenameArgCount( $newname );
1125                 if ( $newname ) {
1126                         if ( $this->handle ) {
1127                                 fclose( $this->handle );
1128                                 $this->handle = false;
1129                         }
1130                         if ( $this->procOpenResource ) {
1131                                 proc_close( $this->procOpenResource );
1132                                 $this->procOpenResource = false;
1133                         }
1134                         $this->renameOrException( $newname );
1135                         if ( $open ) {
1136                                 $command = $this->command;
1137                                 $command .= " > " . wfEscapeShellArg( $this->filename );
1138                                 $this->startCommand( $command );
1139                         }
1140                 }
1141         }
1142
1143 }
1144
1145 /**
1146  * Sends dump output via the gzip compressor.
1147  * @ingroup Dump
1148  */
1149 class DumpGZipOutput extends DumpPipeOutput {
1150
1151         /**
1152          * @param $file string
1153          */
1154         function __construct( $file ) {
1155                 parent::__construct( "gzip", $file );
1156         }
1157 }
1158
1159 /**
1160  * Sends dump output via the bgzip2 compressor.
1161  * @ingroup Dump
1162  */
1163 class DumpBZip2Output extends DumpPipeOutput {
1164
1165         /**
1166          * @param $file string
1167          */
1168         function __construct( $file ) {
1169                 parent::__construct( "bzip2", $file );
1170         }
1171 }
1172
1173 /**
1174  * Sends dump output via the p7zip compressor.
1175  * @ingroup Dump
1176  */
1177 class Dump7ZipOutput extends DumpPipeOutput {
1178
1179         /**
1180          * @param $file string
1181          */
1182         function __construct( $file ) {
1183                 $command = $this->setup7zCommand( $file );
1184                 parent::__construct( $command );
1185                 $this->filename = $file;
1186         }
1187
1188         /**
1189          * @param $file string
1190          * @return string
1191          */
1192         function setup7zCommand( $file ) {
1193                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
1194                 // Suppress annoying useless crap from p7zip
1195                 // Unfortunately this could suppress real error messages too
1196                 $command .= ' >' . wfGetNull() . ' 2>&1';
1197                 return $command;
1198         }
1199
1200         /**
1201          * @param $newname string
1202          * @param $open bool
1203          */
1204         function closeAndRename( $newname, $open = false ) {
1205                 $newname = $this->checkRenameArgCount( $newname );
1206                 if ( $newname ) {
1207                         fclose( $this->handle );
1208                         proc_close( $this->procOpenResource );
1209                         $this->renameOrException( $newname );
1210                         if ( $open ) {
1211                                 $command = $this->setup7zCommand( $this->filename );
1212                                 $this->startCommand( $command );
1213                         }
1214                 }
1215         }
1216 }
1217
1218 /**
1219  * Dump output filter class.
1220  * This just does output filtering and streaming; XML formatting is done
1221  * higher up, so be careful in what you do.
1222  * @ingroup Dump
1223  */
1224 class DumpFilter {
1225
1226         /**
1227          * @var DumpOutput
1228          * FIXME will need to be made protected whenever legacy code
1229          * is updated.
1230          */
1231         public $sink;
1232
1233         /**
1234          * @var bool
1235          */
1236         protected $sendingThisPage;
1237
1238         /**
1239          * @param $sink DumpOutput
1240          */
1241         function __construct( &$sink ) {
1242                 $this->sink =& $sink;
1243         }
1244
1245         /**
1246          * @param $string string
1247          */
1248         function writeOpenStream( $string ) {
1249                 $this->sink->writeOpenStream( $string );
1250         }
1251
1252         /**
1253          * @param $string string
1254          */
1255         function writeCloseStream( $string ) {
1256                 $this->sink->writeCloseStream( $string );
1257         }
1258
1259         /**
1260          * @param $page
1261          * @param $string string
1262          */
1263         function writeOpenPage( $page, $string ) {
1264                 $this->sendingThisPage = $this->pass( $page, $string );
1265                 if ( $this->sendingThisPage ) {
1266                         $this->sink->writeOpenPage( $page, $string );
1267                 }
1268         }
1269
1270         /**
1271          * @param $string string
1272          */
1273         function writeClosePage( $string ) {
1274                 if ( $this->sendingThisPage ) {
1275                         $this->sink->writeClosePage( $string );
1276                         $this->sendingThisPage = false;
1277                 }
1278         }
1279
1280         /**
1281          * @param $rev
1282          * @param $string string
1283          */
1284         function writeRevision( $rev, $string ) {
1285                 if ( $this->sendingThisPage ) {
1286                         $this->sink->writeRevision( $rev, $string );
1287                 }
1288         }
1289
1290         /**
1291          * @param $rev
1292          * @param $string string
1293          */
1294         function writeLogItem( $rev, $string ) {
1295                 $this->sink->writeRevision( $rev, $string );
1296         }
1297
1298         /**
1299          * @param $newname string
1300          */
1301         function closeRenameAndReopen( $newname ) {
1302                 $this->sink->closeRenameAndReopen( $newname );
1303         }
1304
1305         /**
1306          * @param $newname string
1307          * @param $open bool
1308          */
1309         function closeAndRename( $newname, $open = false ) {
1310                 $this->sink->closeAndRename( $newname, $open );
1311         }
1312
1313         /**
1314          * @return array
1315          */
1316         function getFilenames() {
1317                 return $this->sink->getFilenames();
1318         }
1319
1320         /**
1321          * Override for page-based filter types.
1322          * @param $page
1323          * @return bool
1324          */
1325         function pass( $page ) {
1326                 return true;
1327         }
1328 }
1329
1330 /**
1331  * Simple dump output filter to exclude all talk pages.
1332  * @ingroup Dump
1333  */
1334 class DumpNotalkFilter extends DumpFilter {
1335
1336         /**
1337          * @param $page
1338          * @return bool
1339          */
1340         function pass( $page ) {
1341                 return !MWNamespace::isTalk( $page->page_namespace );
1342         }
1343 }
1344
1345 /**
1346  * Dump output filter to include or exclude pages in a given set of namespaces.
1347  * @ingroup Dump
1348  */
1349 class DumpNamespaceFilter extends DumpFilter {
1350         var $invert = false;
1351         var $namespaces = array();
1352
1353         /**
1354          * @param $sink DumpOutput
1355          * @param $param
1356          * @throws MWException
1357          */
1358         function __construct( &$sink, $param ) {
1359                 parent::__construct( $sink );
1360
1361                 $constants = array(
1362                         "NS_MAIN"           => NS_MAIN,
1363                         "NS_TALK"           => NS_TALK,
1364                         "NS_USER"           => NS_USER,
1365                         "NS_USER_TALK"      => NS_USER_TALK,
1366                         "NS_PROJECT"        => NS_PROJECT,
1367                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
1368                         "NS_FILE"           => NS_FILE,
1369                         "NS_FILE_TALK"      => NS_FILE_TALK,
1370                         "NS_IMAGE"          => NS_IMAGE, // NS_IMAGE is an alias for NS_FILE
1371                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
1372                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
1373                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
1374                         "NS_TEMPLATE"       => NS_TEMPLATE,
1375                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
1376                         "NS_HELP"           => NS_HELP,
1377                         "NS_HELP_TALK"      => NS_HELP_TALK,
1378                         "NS_CATEGORY"       => NS_CATEGORY,
1379                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
1380
1381                 if ( $param { 0 } == '!' ) {
1382                         $this->invert = true;
1383                         $param = substr( $param, 1 );
1384                 }
1385
1386                 foreach ( explode( ',', $param ) as $key ) {
1387                         $key = trim( $key );
1388                         if ( isset( $constants[$key] ) ) {
1389                                 $ns = $constants[$key];
1390                                 $this->namespaces[$ns] = true;
1391                         } elseif ( is_numeric( $key ) ) {
1392                                 $ns = intval( $key );
1393                                 $this->namespaces[$ns] = true;
1394                         } else {
1395                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
1396                         }
1397                 }
1398         }
1399
1400         /**
1401          * @param $page
1402          * @return bool
1403          */
1404         function pass( $page ) {
1405                 $match = isset( $this->namespaces[$page->page_namespace] );
1406                 return $this->invert xor $match;
1407         }
1408 }
1409
1410 /**
1411  * Dump output filter to include only the last revision in each page sequence.
1412  * @ingroup Dump
1413  */
1414 class DumpLatestFilter extends DumpFilter {
1415         var $page, $pageString, $rev, $revString;
1416
1417         /**
1418          * @param $page
1419          * @param $string string
1420          */
1421         function writeOpenPage( $page, $string ) {
1422                 $this->page = $page;
1423                 $this->pageString = $string;
1424         }
1425
1426         /**
1427          * @param $string string
1428          */
1429         function writeClosePage( $string ) {
1430                 if ( $this->rev ) {
1431                         $this->sink->writeOpenPage( $this->page, $this->pageString );
1432                         $this->sink->writeRevision( $this->rev, $this->revString );
1433                         $this->sink->writeClosePage( $string );
1434                 }
1435                 $this->rev = null;
1436                 $this->revString = null;
1437                 $this->page = null;
1438                 $this->pageString = null;
1439         }
1440
1441         /**
1442          * @param $rev
1443          * @param $string string
1444          */
1445         function writeRevision( $rev, $string ) {
1446                 if ( $rev->rev_id == $this->page->page_latest ) {
1447                         $this->rev = $rev;
1448                         $this->revString = $string;
1449                 }
1450         }
1451 }
1452
1453 /**
1454  * Base class for output stream; prints to stdout or buffer or wherever.
1455  * @ingroup Dump
1456  */
1457 class DumpMultiWriter {
1458
1459         /**
1460          * @param $sinks
1461          */
1462         function __construct( $sinks ) {
1463                 $this->sinks = $sinks;
1464                 $this->count = count( $sinks );
1465         }
1466
1467         /**
1468          * @param $string string
1469          */
1470         function writeOpenStream( $string ) {
1471                 for ( $i = 0; $i < $this->count; $i++ ) {
1472                         $this->sinks[$i]->writeOpenStream( $string );
1473                 }
1474         }
1475
1476         /**
1477          * @param $string string
1478          */
1479         function writeCloseStream( $string ) {
1480                 for ( $i = 0; $i < $this->count; $i++ ) {
1481                         $this->sinks[$i]->writeCloseStream( $string );
1482                 }
1483         }
1484
1485         /**
1486          * @param $page
1487          * @param $string string
1488          */
1489         function writeOpenPage( $page, $string ) {
1490                 for ( $i = 0; $i < $this->count; $i++ ) {
1491                         $this->sinks[$i]->writeOpenPage( $page, $string );
1492                 }
1493         }
1494
1495         /**
1496          * @param $string
1497          */
1498         function writeClosePage( $string ) {
1499                 for ( $i = 0; $i < $this->count; $i++ ) {
1500                         $this->sinks[$i]->writeClosePage( $string );
1501                 }
1502         }
1503
1504         /**
1505          * @param $rev
1506          * @param $string
1507          */
1508         function writeRevision( $rev, $string ) {
1509                 for ( $i = 0; $i < $this->count; $i++ ) {
1510                         $this->sinks[$i]->writeRevision( $rev, $string );
1511                 }
1512         }
1513
1514         /**
1515          * @param $newnames
1516          */
1517         function closeRenameAndReopen( $newnames ) {
1518                 $this->closeAndRename( $newnames, true );
1519         }
1520
1521         /**
1522          * @param $newnames array
1523          * @param bool $open
1524          */
1525         function closeAndRename( $newnames, $open = false ) {
1526                 for ( $i = 0; $i < $this->count; $i++ ) {
1527                         $this->sinks[$i]->closeAndRename( $newnames[$i], $open );
1528                 }
1529         }
1530
1531         /**
1532          * @return array
1533          */
1534         function getFilenames() {
1535                 $filenames = array();
1536                 for ( $i = 0; $i < $this->count; $i++ ) {
1537                         $filenames[] = $this->sinks[$i]->getFilenames();
1538                 }
1539                 return $filenames;
1540         }
1541
1542 }
1543
1544 /**
1545  * @param $string string
1546  * @return string
1547  */
1548 function xmlsafe( $string ) {
1549         wfProfileIn( __FUNCTION__ );
1550
1551         /**
1552          * The page may contain old data which has not been properly normalized.
1553          * Invalid UTF-8 sequences or forbidden control characters will make our
1554          * XML output invalid, so be sure to strip them out.
1555          */
1556         $string = UtfNormal::cleanUp( $string );
1557
1558         $string = htmlspecialchars( $string );
1559         wfProfileOut( __FUNCTION__ );
1560         return $string;
1561 }