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-$ver/ " .
 506                                 #TODO: how do we get a new version up there?
 507                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 508                         'version'            => $ver,
 509                         'xml:lang'           => $wgLanguageCode ),
 510                         null ) .
 511                         "\n" .
 512                         $this->siteInfo();
 513         }
 514
 515         /**
 516          * @return string
 517          */
 518         function siteInfo() {
 519                 $info = array(
 520                         $this->sitename(),
 521                         $this->homelink(),
 522                         $this->generator(),
 523                         $this->caseSetting(),
 524                         $this->namespaces() );
 525                 return "  <siteinfo>\n    " .
 526                         implode( "\n    ", $info ) .
 527                         "\n  </siteinfo>\n";
 528         }
 529
 530         /**
 531          * @return string
 532          */
 533         function sitename() {
 534                 global $wgSitename;
 535                 return Xml::element( 'sitename', array(), $wgSitename );
 536         }
 537
 538         /**
 539          * @return string
 540          */
 541         function generator() {
 542                 global $wgVersion;
 543                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 544         }
 545
 546         /**
 547          * @return string
 548          */
 549         function homelink() {
 550                 return Xml::element( 'base', array(), Title::newMainPage()->getCanonicalURL() );
 551         }
 552
 553         /**
 554          * @return string
 555          */
 556         function caseSetting() {
 557                 global $wgCapitalLinks;
 558                 // "case-insensitive" option is reserved for future
 559                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 560                 return Xml::element( 'case', array(), $sensitivity );
 561         }
 562
 563         /**
 564          * @return string
 565          */
 566         function namespaces() {
 567                 global $wgContLang;
 568                 $spaces = "<namespaces>\n";
 569                 foreach ( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 570                         $spaces .= '      ' .
 571                                 Xml::element( 'namespace',
 572                                         array(
 573                                                 'key' => $ns,
 574                                                 'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 575                                         ), $title ) . "\n";
 576                 }
 577                 $spaces .= "    </namespaces>";
 578                 return $spaces;
 579         }
 580
 581         /**
 582          * Closes the output stream with the closing root element.
 583          * Call when finished dumping things.
 584          *
 585          * @return string
 586          */
 587         function closeStream() {
 588                 return "</mediawiki>\n";
 589         }
 590
 591         /**
 592          * Opens a "<page>" section on the output stream, with data
 593          * from the given database row.
 594          *
 595          * @param $row object
 596          * @return string
 597          * @access private
 598          */
 599         function openPage( $row ) {
 600                 $out = "  <page>\n";
 601                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 602                 $out .= '    ' . Xml::elementClean( 'title', array(), self::canonicalTitle( $title ) ) . "\n";
 603                 $out .= '    ' . Xml::element( 'ns', array(), strval( $row->page_namespace ) ) . "\n";
 604                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 605                 if ( $row->page_is_redirect ) {
 606                         $page = WikiPage::factory( $title );
 607                         $redirect = $page->getRedirectTarget();
 608                         if ( $redirect instanceof Title && $redirect->isValidRedirectTarget() ) {
 609                                 $out .= '    ' . Xml::element( 'redirect', array( 'title' => self::canonicalTitle( $redirect ) ) ) . "\n";
 610                         }
 611                 }
 612
 613                 if ( $row->page_restrictions != '' ) {
 614                         $out .= '    ' . Xml::element( 'restrictions', array(),
 615                                 strval( $row->page_restrictions ) ) . "\n";
 616                 }
 617
 618                 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 619
 620                 return $out;
 621         }
 622
 623         /**
 624          * Closes a "<page>" section on the output stream.
 625          *
 626          * @access private
 627          * @return string
 628          */
 629         function closePage() {
 630                 return "  </page>\n";
 631         }
 632
 633         /**
 634          * Dumps a "<revision>" section on the output stream, with
 635          * data filled in from the given database row.
 636          *
 637          * @param $row object
 638          * @return string
 639          * @access private
 640          */
 641         function writeRevision( $row ) {
 642                 wfProfileIn( __METHOD__ );
 643
 644                 $out = "    <revision>\n";
 645                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 646                 if ( isset( $row->rev_parent_id ) && $row->rev_parent_id ) {
 647                         $out .= "      " . Xml::element( 'parentid', null, strval( $row->rev_parent_id ) ) . "\n";
 648                 }
 649
 650                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 651
 652                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_USER ) ) {
 653                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 654                 } else {
 655                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 656                 }
 657
 658                 if ( isset( $row->rev_minor_edit ) && $row->rev_minor_edit ) {
 659                         $out .= "      <minor/>\n";
 660                 }
 661                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_COMMENT ) ) {
 662                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 663                 } elseif ( $row->rev_comment != '' ) {
 664                         $out .= "      " . Xml::elementClean( 'comment', array(), strval( $row->rev_comment ) ) . "\n";
 665                 }
 666
 667                 $text = '';
 668                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_TEXT ) ) {
 669                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 670                 } elseif ( isset( $row->old_text ) ) {
 671                         // Raw text from the database may have invalid chars
 672                         $text = strval( Revision::getRevisionText( $row ) );
 673                         $out .= "      " . Xml::elementClean( 'text',
 674                                 array( 'xml:space' => 'preserve', 'bytes' => intval( $row->rev_len ) ),
 675                                 strval( $text ) ) . "\n";
 676                 } else {
 677                         // Stub output
 678                         $out .= "      " . Xml::element( 'text',
 679                                 array( 'id' => $row->rev_text_id, 'bytes' => intval( $row->rev_len ) ),
 680                                 "" ) . "\n";
 681                 }
 682
 683                 if ( isset( $row->rev_sha1 ) && $row->rev_sha1 && !( $row->rev_deleted & Revision::DELETED_TEXT ) ) {
 684                         $out .= "      " . Xml::element( 'sha1', null, strval( $row->rev_sha1 ) ) . "\n";
 685                 } else {
 686                         $out .= "      <sha1/>\n";
 687                 }
 688
 689                 if ( isset( $row->rev_content_model ) && !is_null( $row->rev_content_model ) ) {
 690                         $content_model = strval( $row->rev_content_model );
 691                 } else {
 692                         // probably using $wgContentHandlerUseDB = false;
 693                         // @todo test!
 694                         $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 695                         $content_model = ContentHandler::getDefaultModelFor( $title );
 696                 }
 697
 698                 $out .= "      " . Xml::element( 'model', null, strval( $content_model ) ) . "\n";
 699
 700                 if ( isset( $row->rev_content_format ) && !is_null( $row->rev_content_format ) ) {
 701                         $content_format = strval( $row->rev_content_format );
 702                 } else {
 703                         // probably using $wgContentHandlerUseDB = false;
 704                         // @todo test!
 705                         $content_handler = ContentHandler::getForModelID( $content_model );
 706                         $content_format = $content_handler->getDefaultFormat();
 707                 }
 708
 709                 $out .= "      " . Xml::element( 'format', null, strval( $content_format ) ) . "\n";
 710
 711                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 712
 713                 $out .= "    </revision>\n";
 714
 715                 wfProfileOut( __METHOD__ );
 716                 return $out;
 717         }
 718
 719         /**
 720          * Dumps a "<logitem>" section on the output stream, with
 721          * data filled in from the given database row.
 722          *
 723          * @param $row object
 724          * @return string
 725          * @access private
 726          */
 727         function writeLogItem( $row ) {
 728                 wfProfileIn( __METHOD__ );
 729
 730                 $out = "  <logitem>\n";
 731                 $out .= "    " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 732
 733                 $out .= $this->writeTimestamp( $row->log_timestamp, "    " );
 734
 735                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 736                         $out .= "    " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 737                 } else {
 738                         $out .= $this->writeContributor( $row->log_user, $row->user_name, "    " );
 739                 }
 740
 741                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 742                         $out .= "    " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 743                 } elseif ( $row->log_comment != '' ) {
 744                         $out .= "    " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 745                 }
 746
 747                 $out .= "    " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 748                 $out .= "    " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 749
 750                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 751                         $out .= "    " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 752                 } else {
 753                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 754                         $out .= "    " . Xml::elementClean( 'logtitle', null, self::canonicalTitle( $title ) ) . "\n";
 755                         $out .= "    " . Xml::elementClean( 'params',
 756                                 array( 'xml:space' => 'preserve' ),
 757                                 strval( $row->log_params ) ) . "\n";
 758                 }
 759
 760                 $out .= "  </logitem>\n";
 761
 762                 wfProfileOut( __METHOD__ );
 763                 return $out;
 764         }
 765
 766         /**
 767          * @param $timestamp string
 768          * @param string $indent Default to six spaces
 769          * @return string
 770          */
 771         function writeTimestamp( $timestamp, $indent = "      " ) {
 772                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 773                 return $indent . Xml::element( 'timestamp', null, $ts ) . "\n";
 774         }
 775
 776         /**
 777          * @param $id
 778          * @param $text string
 779          * @param string $indent Default to six spaces
 780          * @return string
 781          */
 782         function writeContributor( $id, $text, $indent = "      " ) {
 783                 $out = $indent . "<contributor>\n";
 784                 if ( $id || !IP::isValid( $text ) ) {
 785                         $out .= $indent . "  " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 786                         $out .= $indent . "  " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 787                 } else {
 788                         $out .= $indent . "  " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 789                 }
 790                 $out .= $indent . "</contributor>\n";
 791                 return $out;
 792         }
 793
 794         /**
 795          * Warning! This data is potentially inconsistent. :(
 796          * @param $row
 797          * @param $dumpContents bool
 798          * @return string
 799          */
 800         function writeUploads( $row, $dumpContents = false ) {
 801                 if ( $row->page_namespace == NS_FILE ) {
 802                         $img = wfLocalFile( $row->page_title );
 803                         if ( $img && $img->exists() ) {
 804                                 $out = '';
 805                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 806                                         $out .= $this->writeUpload( $ver, $dumpContents );
 807                                 }
 808                                 $out .= $this->writeUpload( $img, $dumpContents );
 809                                 return $out;
 810                         }
 811                 }
 812                 return '';
 813         }
 814
 815         /**
 816          * @param $file File
 817          * @param $dumpContents bool
 818          * @return string
 819          */
 820         function writeUpload( $file, $dumpContents = false ) {
 821                 if ( $file->isOld() ) {
 822                         $archiveName = "      " .
 823                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 824                 } else {
 825                         $archiveName = '';
 826                 }
 827                 if ( $dumpContents ) {
 828                         $be = $file->getRepo()->getBackend();
 829                         # Dump file as base64
 830                         # Uses only XML-safe characters, so does not need escaping
 831                         # @TODO: too bad this loads the contents into memory (script might swap)
 832                         $contents = '      <contents encoding="base64">' .
 833                                 chunk_split( base64_encode(
 834                                         $be->getFileContents( array( 'src' => $file->getPath() ) ) ) ) .
 835                                 "      </contents>\n";
 836                 } else {
 837                         $contents = '';
 838                 }
 839                 if ( $file->isDeleted( File::DELETED_COMMENT ) ) {
 840                         $comment = Xml::element( 'comment', array( 'deleted' => 'deleted' ) );
 841                 } else {
 842                         $comment = Xml::elementClean( 'comment', null, $file->getDescription() );
 843                 }
 844                 return "    <upload>\n" .
 845                         $this->writeTimestamp( $file->getTimestamp() ) .
 846                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 847                         "      " . $comment . "\n" .
 848                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 849                         $archiveName .
 850                         "      " . Xml::element( 'src', null, $file->getCanonicalURL() ) . "\n" .
 851                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 852                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 853                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 854                         $contents .
 855                         "    </upload>\n";
 856         }
 857
 858         /**
 859          * Return prefixed text form of title, but using the content language's
 860          * canonical namespace. This skips any special-casing such as gendered
 861          * user namespaces -- which while useful, are not yet listed in the
 862          * XML "<siteinfo>" data so are unsafe in export.
 863          *
 864          * @param Title $title
 865          * @return string
 866          * @since 1.18
 867          */
 868         public static function canonicalTitle( Title $title ) {
 869                 if ( $title->isExternal() ) {
 870                         return $title->getPrefixedText();
 871                 }
 872
 873                 global $wgContLang;
 874                 $prefix = str_replace( '_', ' ', $wgContLang->getNsText( $title->getNamespace() ) );
 875
 876                 if ( $prefix !== '' ) {
 877                         $prefix .= ':';
 878                 }
 879
 880                 return $prefix . $title->getText();
 881         }
 882 }
 883
 884 /**
 885  * Base class for output stream; prints to stdout or buffer or wherever.
 886  * @ingroup Dump
 887  */
 888 class DumpOutput {
 889
 890         /**
 891          * @param $string string
 892          */
 893         function writeOpenStream( $string ) {
 894                 $this->write( $string );
 895         }
 896
 897         /**
 898          * @param $string string
 899          */
 900         function writeCloseStream( $string ) {
 901                 $this->write( $string );
 902         }
 903
 904         /**
 905          * @param $page
 906          * @param $string string
 907          */
 908         function writeOpenPage( $page, $string ) {
 909                 $this->write( $string );
 910         }
 911
 912         /**
 913          * @param $string string
 914          */
 915         function writeClosePage( $string ) {
 916                 $this->write( $string );
 917         }
 918
 919         /**
 920          * @param $rev
 921          * @param $string string
 922          */
 923         function writeRevision( $rev, $string ) {
 924                 $this->write( $string );
 925         }
 926
 927         /**
 928          * @param $rev
 929          * @param $string string
 930          */
 931         function writeLogItem( $rev, $string ) {
 932                 $this->write( $string );
 933         }
 934
 935         /**
 936          * Override to write to a different stream type.
 937          * @param $string string
 938          * @return bool
 939          */
 940         function write( $string ) {
 941                 print $string;
 942         }
 943
 944         /**
 945          * Close the old file, move it to a specified name,
 946          * and reopen new file with the old name. Use this
 947          * for writing out a file in multiple pieces
 948          * at specified checkpoints (e.g. every n hours).
 949          * @param $newname mixed File name. May be a string or an array with one element
 950          */
 951         function closeRenameAndReopen( $newname ) {
 952         }
 953
 954         /**
 955          * Close the old file, and move it to a specified name.
 956          * Use this for the last piece of a file written out
 957          * at specified checkpoints (e.g. every n hours).
 958          * @param $newname mixed File name. May be a string or an array with one element
 959          * @param bool $open If true, a new file with the old filename will be opened again for writing (default: false)
 960          */
 961         function closeAndRename( $newname, $open = false ) {
 962         }
 963
 964         /**
 965          * Returns the name of the file or files which are
 966          * being written to, if there are any.
 967          * @return null
 968          */
 969         function getFilenames() {
 970                 return null;
 971         }
 972 }
 973
 974 /**
 975  * Stream outputter to send data to a file.
 976  * @ingroup Dump
 977  */
 978 class DumpFileOutput extends DumpOutput {
 979         protected $handle = false, $filename;
 980
 981         /**
 982          * @param $file
 983          */
 984         function __construct( $file ) {
 985                 $this->handle = fopen( $file, "wt" );
 986                 $this->filename = $file;
 987         }
 988
 989         /**
 990          * @param $string string
 991          */
 992         function writeCloseStream( $string ) {
 993                 parent::writeCloseStream( $string );
 994                 if ( $this->handle ) {
 995                         fclose( $this->handle );
 996                         $this->handle = false;
 997                 }
 998         }
 999
1000         /**
1001          * @param $string string
1002          */
1003         function write( $string ) {
1004                 fputs( $this->handle, $string );
1005         }
1006
1007         /**
1008          * @param $newname
1009          */
1010         function closeRenameAndReopen( $newname ) {
1011                 $this->closeAndRename( $newname, true );
1012         }
1013
1014         /**
1015          * @param $newname
1016          * @throws MWException
1017          */
1018         function renameOrException( $newname ) {
1019                         if ( !rename( $this->filename, $newname ) ) {
1020                                 throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" );
1021                         }
1022         }
1023
1024         /**
1025          * @param $newname array
1026          * @return mixed
1027          * @throws MWException
1028          */
1029         function checkRenameArgCount( $newname ) {
1030                 if ( is_array( $newname ) ) {
1031                         if ( count( $newname ) > 1 ) {
1032                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
1033                         } else {
1034                                 $newname = $newname[0];
1035                         }
1036                 }
1037                 return $newname;
1038         }
1039
1040         /**
1041          * @param $newname mixed
1042          * @param $open bool
1043          */
1044         function closeAndRename( $newname, $open = false ) {
1045                 $newname = $this->checkRenameArgCount( $newname );
1046                 if ( $newname ) {
1047                         if ( $this->handle ) {
1048                                 fclose( $this->handle );
1049                                 $this->handle = false;
1050                         }
1051                         $this->renameOrException( $newname );
1052                         if ( $open ) {
1053                                 $this->handle = fopen( $this->filename, "wt" );
1054                         }
1055                 }
1056         }
1057
1058         /**
1059          * @return string|null
1060          */
1061         function getFilenames() {
1062                 return $this->filename;
1063         }
1064 }
1065
1066 /**
1067  * Stream outputter to send data to a file via some filter program.
1068  * Even if compression is available in a library, using a separate
1069  * program can allow us to make use of a multi-processor system.
1070  * @ingroup Dump
1071  */
1072 class DumpPipeOutput extends DumpFileOutput {
1073         protected $command, $filename;
1074         protected $procOpenResource = false;
1075
1076         /**
1077          * @param $command
1078          * @param $file null
1079          */
1080         function __construct( $command, $file = null ) {
1081                 if ( !is_null( $file ) ) {
1082                         $command .= " > " . wfEscapeShellArg( $file );
1083                 }
1084
1085                 $this->startCommand( $command );
1086                 $this->command = $command;
1087                 $this->filename = $file;
1088         }
1089
1090         /**
1091          * @param $string string
1092          */
1093         function writeCloseStream( $string ) {
1094                 parent::writeCloseStream( $string );
1095                 if ( $this->procOpenResource ) {
1096                         proc_close( $this->procOpenResource );
1097                         $this->procOpenResource = false;
1098                 }
1099         }
1100
1101         /**
1102          * @param $command
1103          */
1104         function startCommand( $command ) {
1105                 $spec = array(
1106                         0 => array( "pipe", "r" ),
1107                 );
1108                 $pipes = array();
1109                 $this->procOpenResource = proc_open( $command, $spec, $pipes );
1110                 $this->handle = $pipes[0];
1111         }
1112
1113         /**
1114          * @param mixed $newname
1115          */
1116         function closeRenameAndReopen( $newname ) {
1117                 $this->closeAndRename( $newname, true );
1118         }
1119
1120         /**
1121          * @param $newname mixed
1122          * @param $open bool
1123          */
1124         function closeAndRename( $newname, $open = false ) {
1125                 $newname = $this->checkRenameArgCount( $newname );
1126                 if ( $newname ) {
1127                         if ( $this->handle ) {
1128                                 fclose( $this->handle );
1129                                 $this->handle = false;
1130                         }
1131                         if ( $this->procOpenResource ) {
1132                                 proc_close( $this->procOpenResource );
1133                                 $this->procOpenResource = false;
1134                         }
1135                         $this->renameOrException( $newname );
1136                         if ( $open ) {
1137                                 $command = $this->command;
1138                                 $command .= " > " . wfEscapeShellArg( $this->filename );
1139                                 $this->startCommand( $command );
1140                         }
1141                 }
1142         }
1143
1144 }
1145
1146 /**
1147  * Sends dump output via the gzip compressor.
1148  * @ingroup Dump
1149  */
1150 class DumpGZipOutput extends DumpPipeOutput {
1151
1152         /**
1153          * @param $file string
1154          */
1155         function __construct( $file ) {
1156                 parent::__construct( "gzip", $file );
1157         }
1158 }
1159
1160 /**
1161  * Sends dump output via the bgzip2 compressor.
1162  * @ingroup Dump
1163  */
1164 class DumpBZip2Output extends DumpPipeOutput {
1165
1166         /**
1167          * @param $file string
1168          */
1169         function __construct( $file ) {
1170                 parent::__construct( "bzip2", $file );
1171         }
1172 }
1173
1174 /**
1175  * Sends dump output via the p7zip compressor.
1176  * @ingroup Dump
1177  */
1178 class Dump7ZipOutput extends DumpPipeOutput {
1179
1180         /**
1181          * @param $file string
1182          */
1183         function __construct( $file ) {
1184                 $command = $this->setup7zCommand( $file );
1185                 parent::__construct( $command );
1186                 $this->filename = $file;
1187         }
1188
1189         /**
1190          * @param $file string
1191          * @return string
1192          */
1193         function setup7zCommand( $file ) {
1194                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
1195                 // Suppress annoying useless crap from p7zip
1196                 // Unfortunately this could suppress real error messages too
1197                 $command .= ' >' . wfGetNull() . ' 2>&1';
1198                 return $command;
1199         }
1200
1201         /**
1202          * @param $newname string
1203          * @param $open bool
1204          */
1205         function closeAndRename( $newname, $open = false ) {
1206                 $newname = $this->checkRenameArgCount( $newname );
1207                 if ( $newname ) {
1208                         fclose( $this->handle );
1209                         proc_close( $this->procOpenResource );
1210                         $this->renameOrException( $newname );
1211                         if ( $open ) {
1212                                 $command = $this->setup7zCommand( $this->filename );
1213                                 $this->startCommand( $command );
1214                         }
1215                 }
1216         }
1217 }
1218
1219 /**
1220  * Dump output filter class.
1221  * This just does output filtering and streaming; XML formatting is done
1222  * higher up, so be careful in what you do.
1223  * @ingroup Dump
1224  */
1225 class DumpFilter {
1226
1227         /**
1228          * @var DumpOutput
1229          * FIXME will need to be made protected whenever legacy code
1230          * is updated.
1231          */
1232         public $sink;
1233
1234         /**
1235          * @var bool
1236          */
1237         protected $sendingThisPage;
1238
1239         /**
1240          * @param $sink DumpOutput
1241          */
1242         function __construct( &$sink ) {
1243                 $this->sink =& $sink;
1244         }
1245
1246         /**
1247          * @param $string string
1248          */
1249         function writeOpenStream( $string ) {
1250                 $this->sink->writeOpenStream( $string );
1251         }
1252
1253         /**
1254          * @param $string string
1255          */
1256         function writeCloseStream( $string ) {
1257                 $this->sink->writeCloseStream( $string );
1258         }
1259
1260         /**
1261          * @param $page
1262          * @param $string string
1263          */
1264         function writeOpenPage( $page, $string ) {
1265                 $this->sendingThisPage = $this->pass( $page, $string );
1266                 if ( $this->sendingThisPage ) {
1267                         $this->sink->writeOpenPage( $page, $string );
1268                 }
1269         }
1270
1271         /**
1272          * @param $string string
1273          */
1274         function writeClosePage( $string ) {
1275                 if ( $this->sendingThisPage ) {
1276                         $this->sink->writeClosePage( $string );
1277                         $this->sendingThisPage = false;
1278                 }
1279         }
1280
1281         /**
1282          * @param $rev
1283          * @param $string string
1284          */
1285         function writeRevision( $rev, $string ) {
1286                 if ( $this->sendingThisPage ) {
1287                         $this->sink->writeRevision( $rev, $string );
1288                 }
1289         }
1290
1291         /**
1292          * @param $rev
1293          * @param $string string
1294          */
1295         function writeLogItem( $rev, $string ) {
1296                 $this->sink->writeRevision( $rev, $string );
1297         }
1298
1299         /**
1300          * @param $newname string
1301          */
1302         function closeRenameAndReopen( $newname ) {
1303                 $this->sink->closeRenameAndReopen( $newname );
1304         }
1305
1306         /**
1307          * @param $newname string
1308          * @param $open bool
1309          */
1310         function closeAndRename( $newname, $open = false ) {
1311                 $this->sink->closeAndRename( $newname, $open );
1312         }
1313
1314         /**
1315          * @return array
1316          */
1317         function getFilenames() {
1318                 return $this->sink->getFilenames();
1319         }
1320
1321         /**
1322          * Override for page-based filter types.
1323          * @param $page
1324          * @return bool
1325          */
1326         function pass( $page ) {
1327                 return true;
1328         }
1329 }
1330
1331 /**
1332  * Simple dump output filter to exclude all talk pages.
1333  * @ingroup Dump
1334  */
1335 class DumpNotalkFilter extends DumpFilter {
1336
1337         /**
1338          * @param $page
1339          * @return bool
1340          */
1341         function pass( $page ) {
1342                 return !MWNamespace::isTalk( $page->page_namespace );
1343         }
1344 }
1345
1346 /**
1347  * Dump output filter to include or exclude pages in a given set of namespaces.
1348  * @ingroup Dump
1349  */
1350 class DumpNamespaceFilter extends DumpFilter {
1351         var $invert = false;
1352         var $namespaces = array();
1353
1354         /**
1355          * @param $sink DumpOutput
1356          * @param $param
1357          * @throws MWException
1358          */
1359         function __construct( &$sink, $param ) {
1360                 parent::__construct( $sink );
1361
1362                 $constants = array(
1363                         "NS_MAIN"           => NS_MAIN,
1364                         "NS_TALK"           => NS_TALK,
1365                         "NS_USER"           => NS_USER,
1366                         "NS_USER_TALK"      => NS_USER_TALK,
1367                         "NS_PROJECT"        => NS_PROJECT,
1368                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
1369                         "NS_FILE"           => NS_FILE,
1370                         "NS_FILE_TALK"      => NS_FILE_TALK,
1371                         "NS_IMAGE"          => NS_IMAGE, // NS_IMAGE is an alias for NS_FILE
1372                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
1373                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
1374                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
1375                         "NS_TEMPLATE"       => NS_TEMPLATE,
1376                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
1377                         "NS_HELP"           => NS_HELP,
1378                         "NS_HELP_TALK"      => NS_HELP_TALK,
1379                         "NS_CATEGORY"       => NS_CATEGORY,
1380                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
1381
1382                 if ( $param { 0 } == '!' ) {
1383                         $this->invert = true;
1384                         $param = substr( $param, 1 );
1385                 }
1386
1387                 foreach ( explode( ',', $param ) as $key ) {
1388                         $key = trim( $key );
1389                         if ( isset( $constants[$key] ) ) {
1390                                 $ns = $constants[$key];
1391                                 $this->namespaces[$ns] = true;
1392                         } elseif ( is_numeric( $key ) ) {
1393                                 $ns = intval( $key );
1394                                 $this->namespaces[$ns] = true;
1395                         } else {
1396                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
1397                         }
1398                 }
1399         }
1400
1401         /**
1402          * @param $page
1403          * @return bool
1404          */
1405         function pass( $page ) {
1406                 $match = isset( $this->namespaces[$page->page_namespace] );
1407                 return $this->invert xor $match;
1408         }
1409 }
1410
1411 /**
1412  * Dump output filter to include only the last revision in each page sequence.
1413  * @ingroup Dump
1414  */
1415 class DumpLatestFilter extends DumpFilter {
1416         var $page, $pageString, $rev, $revString;
1417
1418         /**
1419          * @param $page
1420          * @param $string string
1421          */
1422         function writeOpenPage( $page, $string ) {
1423                 $this->page = $page;
1424                 $this->pageString = $string;
1425         }
1426
1427         /**
1428          * @param $string string
1429          */
1430         function writeClosePage( $string ) {
1431                 if ( $this->rev ) {
1432                         $this->sink->writeOpenPage( $this->page, $this->pageString );
1433                         $this->sink->writeRevision( $this->rev, $this->revString );
1434                         $this->sink->writeClosePage( $string );
1435                 }
1436                 $this->rev = null;
1437                 $this->revString = null;
1438                 $this->page = null;
1439                 $this->pageString = null;
1440         }
1441
1442         /**
1443          * @param $rev
1444          * @param $string string
1445          */
1446         function writeRevision( $rev, $string ) {
1447                 if ( $rev->rev_id == $this->page->page_latest ) {
1448                         $this->rev = $rev;
1449                         $this->revString = $string;
1450                 }
1451         }
1452 }
1453
1454 /**
1455  * Base class for output stream; prints to stdout or buffer or wherever.
1456  * @ingroup Dump
1457  */
1458 class DumpMultiWriter {
1459
1460         /**
1461          * @param $sinks
1462          */
1463         function __construct( $sinks ) {
1464                 $this->sinks = $sinks;
1465                 $this->count = count( $sinks );
1466         }
1467
1468         /**
1469          * @param $string string
1470          */
1471         function writeOpenStream( $string ) {
1472                 for ( $i = 0; $i < $this->count; $i++ ) {
1473                         $this->sinks[$i]->writeOpenStream( $string );
1474                 }
1475         }
1476
1477         /**
1478          * @param $string string
1479          */
1480         function writeCloseStream( $string ) {
1481                 for ( $i = 0; $i < $this->count; $i++ ) {
1482                         $this->sinks[$i]->writeCloseStream( $string );
1483                 }
1484         }
1485
1486         /**
1487          * @param $page
1488          * @param $string string
1489          */
1490         function writeOpenPage( $page, $string ) {
1491                 for ( $i = 0; $i < $this->count; $i++ ) {
1492                         $this->sinks[$i]->writeOpenPage( $page, $string );
1493                 }
1494         }
1495
1496         /**
1497          * @param $string
1498          */
1499         function writeClosePage( $string ) {
1500                 for ( $i = 0; $i < $this->count; $i++ ) {
1501                         $this->sinks[$i]->writeClosePage( $string );
1502                 }
1503         }
1504
1505         /**
1506          * @param $rev
1507          * @param $string
1508          */
1509         function writeRevision( $rev, $string ) {
1510                 for ( $i = 0; $i < $this->count; $i++ ) {
1511                         $this->sinks[$i]->writeRevision( $rev, $string );
1512                 }
1513         }
1514
1515         /**
1516          * @param $newnames
1517          */
1518         function closeRenameAndReopen( $newnames ) {
1519                 $this->closeAndRename( $newnames, true );
1520         }
1521
1522         /**
1523          * @param $newnames array
1524          * @param bool $open
1525          */
1526         function closeAndRename( $newnames, $open = false ) {
1527                 for ( $i = 0; $i < $this->count; $i++ ) {
1528                         $this->sinks[$i]->closeAndRename( $newnames[$i], $open );
1529                 }
1530         }
1531
1532         /**
1533          * @return array
1534          */
1535         function getFilenames() {
1536                 $filenames = array();
1537                 for ( $i = 0; $i < $this->count; $i++ ) {
1538                         $filenames[] = $this->sinks[$i]->getFilenames();
1539                 }
1540                 return $filenames;
1541         }
1542
1543 }
1544
1545 /**
1546  * @param $string string
1547  * @return string
1548  */
1549 function xmlsafe( $string ) {
1550         wfProfileIn( __FUNCTION__ );
1551
1552         /**
1553          * The page may contain old data which has not been properly normalized.
1554          * Invalid UTF-8 sequences or forbidden control characters will make our
1555          * XML output invalid, so be sure to strip them out.
1556          */
1557         $string = UtfNormal::cleanUp( $string );
1558
1559         $string = htmlspecialchars( $string );
1560         wfProfileOut( __FUNCTION__ );
1561         return $string;
1562 }