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  * https://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 DatabaseBase $db
  77          * @param int|array $history One of WikiExporter::FULL, WikiExporter::CURRENT,
  78          *   WikiExporter::RANGE or WikiExporter::STABLE, or an associative array:
  79          *   - offset: non-inclusive offset at which to start the query
  80          *   - limit: maximum number of rows to return
  81          *   - dir: "asc" or "desc" timestamp order
  82          * @param int $buffer One of WikiExporter::BUFFER or WikiExporter::STREAM
  83          * @param int $text One of WikiExporter::TEXT or WikiExporter::STUB
  84          */
  85         function __construct( $db, $history = WikiExporter::CURRENT,
  86                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  87                 $this->db = $db;
  88                 $this->history = $history;
  89                 $this->buffer = $buffer;
  90                 $this->writer = new XmlDumpWriter();
  91                 $this->sink = new DumpOutput();
  92                 $this->text = $text;
  93         }
  94
  95         /**
  96          * Set the DumpOutput or DumpFilter object which will receive
  97          * various row objects and XML output for filtering. Filters
  98          * can be chained or used as callbacks.
  99          *
 100          * @param DumpOutput $sink
 101          */
 102         public function setOutputSink( &$sink ) {
 103                 $this->sink =& $sink;
 104         }
 105
 106         public function openStream() {
 107                 $output = $this->writer->openStream();
 108                 $this->sink->writeOpenStream( $output );
 109         }
 110
 111         public function closeStream() {
 112                 $output = $this->writer->closeStream();
 113                 $this->sink->writeCloseStream( $output );
 114         }
 115
 116         /**
 117          * Dumps a series of page and revision records for all pages
 118          * in the database, either including complete history or only
 119          * the most recent version.
 120          */
 121         public function allPages() {
 122                 $this->dumpFrom( '' );
 123         }
 124
 125         /**
 126          * Dumps a series of page and revision records for those pages
 127          * in the database falling within the page_id range given.
 128          * @param int $start Inclusive lower limit (this id is included)
 129          * @param int $end Exclusive upper limit (this id is not included)
 130          *   If 0, no upper limit.
 131          */
 132         public function pagesByRange( $start, $end ) {
 133                 $condition = 'page_id >= ' . intval( $start );
 134                 if ( $end ) {
 135                         $condition .= ' AND page_id < ' . intval( $end );
 136                 }
 137                 $this->dumpFrom( $condition );
 138         }
 139
 140         /**
 141          * Dumps a series of page and revision records for those pages
 142          * in the database with revisions falling within the rev_id range given.
 143          * @param int $start Inclusive lower limit (this id is included)
 144          * @param int $end Exclusive upper limit (this id is not included)
 145          *   If 0, no upper limit.
 146          */
 147         public function revsByRange( $start, $end ) {
 148                 $condition = 'rev_id >= ' . intval( $start );
 149                 if ( $end ) {
 150                         $condition .= ' AND rev_id < ' . intval( $end );
 151                 }
 152                 $this->dumpFrom( $condition );
 153         }
 154
 155         /**
 156          * @param Title $title
 157          */
 158         public function pageByTitle( $title ) {
 159                 $this->dumpFrom(
 160                         'page_namespace=' . $title->getNamespace() .
 161                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 162         }
 163
 164         /**
 165          * @param string $name
 166          * @throws MWException
 167          */
 168         public function pageByName( $name ) {
 169                 $title = Title::newFromText( $name );
 170                 if ( is_null( $title ) ) {
 171                         throw new MWException( "Can't export invalid title" );
 172                 } else {
 173                         $this->pageByTitle( $title );
 174                 }
 175         }
 176
 177         /**
 178          * @param array $names
 179          */
 180         public function pagesByName( $names ) {
 181                 foreach ( $names as $name ) {
 182                         $this->pageByName( $name );
 183                 }
 184         }
 185
 186         public function allLogs() {
 187                 $this->dumpFrom( '' );
 188         }
 189
 190         /**
 191          * @param int $start
 192          * @param int $end
 193          */
 194         public function logsByRange( $start, $end ) {
 195                 $condition = 'log_id >= ' . intval( $start );
 196                 if ( $end ) {
 197                         $condition .= ' AND log_id < ' . intval( $end );
 198                 }
 199                 $this->dumpFrom( $condition );
 200         }
 201
 202         /**
 203          * Generates the distinct list of authors of an article
 204          * Not called by default (depends on $this->list_authors)
 205          * Can be set by Special:Export when not exporting whole history
 206          *
 207          * @param array $cond
 208          */
 209         protected function do_list_authors( $cond ) {
 210                 wfProfileIn( __METHOD__ );
 211                 $this->author_list = "<contributors>";
 212                 // rev_deleted
 213
 214                 $res = $this->db->select(
 215                         array( 'page', 'revision' ),
 216                         array( 'DISTINCT rev_user_text', 'rev_user' ),
 217                         array(
 218                                 $this->db->bitAnd( 'rev_deleted', Revision::DELETED_USER ) . ' = 0',
 219                                 $cond,
 220                                 'page_id = rev_id',
 221                         ),
 222                         __METHOD__
 223                 );
 224
 225                 foreach ( $res as $row ) {
 226                         $this->author_list .= "<contributor>" .
 227                                 "<username>" .
 228                                 htmlentities( $row->rev_user_text ) .
 229                                 "</username>" .
 230                                 "<id>" .
 231                                 $row->rev_user .
 232                                 "</id>" .
 233                                 "</contributor>";
 234                 }
 235                 $this->author_list .= "</contributors>";
 236                 wfProfileOut( __METHOD__ );
 237         }
 238
 239         /**
 240          * @param string $cond
 241          * @throws MWException
 242          * @throws Exception
 243          */
 244         protected function dumpFrom( $cond = '' ) {
 245                 wfProfileIn( __METHOD__ );
 246                 # For logging dumps...
 247                 if ( $this->history & self::LOGS ) {
 248                         $where = array( 'user_id = log_user' );
 249                         # Hide private logs
 250                         $hideLogs = LogEventsList::getExcludeClause( $this->db );
 251                         if ( $hideLogs ) {
 252                                 $where[] = $hideLogs;
 253                         }
 254                         # Add on any caller specified conditions
 255                         if ( $cond ) {
 256                                 $where[] = $cond;
 257                         }
 258                         # Get logging table name for logging.* clause
 259                         $logging = $this->db->tableName( 'logging' );
 260
 261                         if ( $this->buffer == WikiExporter::STREAM ) {
 262                                 $prev = $this->db->bufferResults( false );
 263                         }
 264                         $wrapper = null; // Assuring $wrapper is not undefined, if exception occurs early
 265                         try {
 266                                 $result = $this->db->select( array( 'logging', 'user' ),
 267                                         array( "{$logging}.*", 'user_name' ), // grab the user name
 268                                         $where,
 269                                         __METHOD__,
 270                                         array( 'ORDER BY' => 'log_id', 'USE INDEX' => array( 'logging' => 'PRIMARY' ) )
 271                                 );
 272                                 $wrapper = $this->db->resultObject( $result );
 273                                 $this->outputLogStream( $wrapper );
 274                                 if ( $this->buffer == WikiExporter::STREAM ) {
 275                                         $this->db->bufferResults( $prev );
 276                                 }
 277                         } catch ( Exception $e ) {
 278                                 // Throwing the exception does not reliably free the resultset, and
 279                                 // would also leave the connection in unbuffered mode.
 280
 281                                 // Freeing result
 282                                 try {
 283                                         if ( $wrapper ) {
 284                                                 $wrapper->free();
 285                                         }
 286                                 } catch ( Exception $e2 ) {
 287                                         // Already in panic mode -> ignoring $e2 as $e has
 288                                         // higher priority
 289                                 }
 290
 291                                 // Putting database back in previous buffer mode
 292                                 try {
 293                                         if ( $this->buffer == WikiExporter::STREAM ) {
 294                                                 $this->db->bufferResults( $prev );
 295                                         }
 296                                 } catch ( Exception $e2 ) {
 297                                         // Already in panic mode -> ignoring $e2 as $e has
 298                                         // higher priority
 299                                 }
 300
 301                                 // Inform caller about problem
 302                                 wfProfileOut( __METHOD__ );
 303                                 throw $e;
 304                         }
 305                 # For page dumps...
 306                 } else {
 307                         $tables = array( 'page', 'revision' );
 308                         $opts = array( 'ORDER BY' => 'page_id ASC' );
 309                         $opts['USE INDEX'] = array();
 310                         $join = array();
 311                         if ( is_array( $this->history ) ) {
 312                                 # Time offset/limit for all pages/history...
 313                                 $revJoin = 'page_id=rev_page';
 314                                 # Set time order
 315                                 if ( $this->history['dir'] == 'asc' ) {
 316                                         $op = '>';
 317                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 318                                 } else {
 319                                         $op = '<';
 320                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 321                                 }
 322                                 # Set offset
 323                                 if ( !empty( $this->history['offset'] ) ) {
 324                                         $revJoin .= " AND rev_timestamp $op " .
 325                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 326                                 }
 327                                 $join['revision'] = array( 'INNER JOIN', $revJoin );
 328                                 # Set query limit
 329                                 if ( !empty( $this->history['limit'] ) ) {
 330                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 331                                 }
 332                         } elseif ( $this->history & WikiExporter::FULL ) {
 333                                 # Full history dumps...
 334                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 335                         } elseif ( $this->history & WikiExporter::CURRENT ) {
 336                                 # Latest revision dumps...
 337                                 if ( $this->list_authors && $cond != '' ) { // List authors, if so desired
 338                                         $this->do_list_authors( $cond );
 339                                 }
 340                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 341                         } elseif ( $this->history & WikiExporter::STABLE ) {
 342                                 # "Stable" revision dumps...
 343                                 # Default JOIN, to be overridden...
 344                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 345                                 # One, and only one hook should set this, and return false
 346                                 if ( wfRunHooks( 'WikiExporter::dumpStableQuery', array( &$tables, &$opts, &$join ) ) ) {
 347                                         wfProfileOut( __METHOD__ );
 348                                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 349                                 }
 350                         } elseif ( $this->history & WikiExporter::RANGE ) {
 351                                 # Dump of revisions within a specified range
 352                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 353                                 $opts['ORDER BY'] = array( 'rev_page ASC', 'rev_id ASC' );
 354                         } else {
 355                                 # Unknown history specification parameter?
 356                                 wfProfileOut( __METHOD__ );
 357                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 358                         }
 359                         # Query optimization hacks
 360                         if ( $cond == '' ) {
 361                                 $opts[] = 'STRAIGHT_JOIN';
 362                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 363                         }
 364                         # Build text join options
 365                         if ( $this->text != WikiExporter::STUB ) { // 1-pass
 366                                 $tables[] = 'text';
 367                                 $join['text'] = array( 'INNER JOIN', 'rev_text_id=old_id' );
 368                         }
 369
 370                         if ( $this->buffer == WikiExporter::STREAM ) {
 371                                 $prev = $this->db->bufferResults( false );
 372                         }
 373
 374                         $wrapper = null; // Assuring $wrapper is not undefined, if exception occurs early
 375                         try {
 376                                 wfRunHooks( 'ModifyExportQuery',
 377                                                 array( $this->db, &$tables, &$cond, &$opts, &$join ) );
 378
 379                                 # Do the query!
 380                                 $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 381                                 $wrapper = $this->db->resultObject( $result );
 382                                 # Output dump results
 383                                 $this->outputPageStream( $wrapper );
 384
 385                                 if ( $this->buffer == WikiExporter::STREAM ) {
 386                                         $this->db->bufferResults( $prev );
 387                                 }
 388                         } catch ( Exception $e ) {
 389                                 // Throwing the exception does not reliably free the resultset, and
 390                                 // would also leave the connection in unbuffered mode.
 391
 392                                 // Freeing result
 393                                 try {
 394                                         if ( $wrapper ) {
 395                                                 $wrapper->free();
 396                                         }
 397                                 } catch ( Exception $e2 ) {
 398                                         // Already in panic mode -> ignoring $e2 as $e has
 399                                         // higher priority
 400                                 }
 401
 402                                 // Putting database back in previous buffer mode
 403                                 try {
 404                                         if ( $this->buffer == WikiExporter::STREAM ) {
 405                                                 $this->db->bufferResults( $prev );
 406                                         }
 407                                 } catch ( Exception $e2 ) {
 408                                         // Already in panic mode -> ignoring $e2 as $e has
 409                                         // higher priority
 410                                 }
 411
 412                                 // Inform caller about problem
 413                                 throw $e;
 414                         }
 415                 }
 416                 wfProfileOut( __METHOD__ );
 417         }
 418
 419         /**
 420          * Runs through a query result set dumping page and revision records.
 421          * The result set should be sorted/grouped by page to avoid duplicate
 422          * page records in the output.
 423          *
 424          * Should be safe for
 425          * streaming (non-buffered) queries, as long as it was made on a
 426          * separate database connection not managed by LoadBalancer; some
 427          * blob storage types will make queries to pull source data.
 428          *
 429          * @param ResultWrapper $resultset
 430          */
 431         protected function outputPageStream( $resultset ) {
 432                 $last = null;
 433                 foreach ( $resultset as $row ) {
 434                         if ( $last === null ||
 435                                 $last->page_namespace != $row->page_namespace ||
 436                                 $last->page_title != $row->page_title ) {
 437                                 if ( $last !== null ) {
 438                                         $output = '';
 439                                         if ( $this->dumpUploads ) {
 440                                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 441                                         }
 442                                         $output .= $this->writer->closePage();
 443                                         $this->sink->writeClosePage( $output );
 444                                 }
 445                                 $output = $this->writer->openPage( $row );
 446                                 $this->sink->writeOpenPage( $row, $output );
 447                                 $last = $row;
 448                         }
 449                         $output = $this->writer->writeRevision( $row );
 450                         $this->sink->writeRevision( $row, $output );
 451                 }
 452                 if ( $last !== null ) {
 453                         $output = '';
 454                         if ( $this->dumpUploads ) {
 455                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 456                         }
 457                         $output .= $this->author_list;
 458                         $output .= $this->writer->closePage();
 459                         $this->sink->writeClosePage( $output );
 460                 }
 461         }
 462
 463         /**
 464          * @param array $resultset
 465          */
 466         protected function outputLogStream( $resultset ) {
 467                 foreach ( $resultset as $row ) {
 468                         $output = $this->writer->writeLogItem( $row );
 469                         $this->sink->writeLogItem( $row, $output );
 470                 }
 471         }
 472 }
 473
 474 /**
 475  * @ingroup Dump
 476  */
 477 class XmlDumpWriter {
 478         /**
 479          * Returns the export schema version.
 480          * @deprecated since 1.20; use WikiExporter::schemaVersion() instead
 481          * @return string
 482          */
 483         function schemaVersion() {
 484                 wfDeprecated( __METHOD__, '1.20' );
 485                 return WikiExporter::schemaVersion();
 486         }
 487
 488         /**
 489          * Opens the XML output stream's root "<mediawiki>" element.
 490          * This does not include an xml directive, so is safe to include
 491          * as a subelement in a larger XML stream. Namespace and XML Schema
 492          * references are included.
 493          *
 494          * Output will be encoded in UTF-8.
 495          *
 496          * @return string
 497          */
 498         function openStream() {
 499                 global $wgLanguageCode;
 500                 $ver = WikiExporter::schemaVersion();
 501                 return Xml::element( 'mediawiki', array(
 502                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 503                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 504                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 505                                 #TODO: how do we get a new version up there?
 506                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 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 object $row
 595          * @return string
 596          */
 597         public function openPage( $row ) {
 598                 $out = "  <page>\n";
 599                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 600                 $out .= '    ' . Xml::elementClean( 'title', array(), self::canonicalTitle( $title ) ) . "\n";
 601                 $out .= '    ' . Xml::element( 'ns', array(), strval( $row->page_namespace ) ) . "\n";
 602                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 603                 if ( $row->page_is_redirect ) {
 604                         $page = WikiPage::factory( $title );
 605                         $redirect = $page->getRedirectTarget();
 606                         if ( $redirect instanceof Title && $redirect->isValidRedirectTarget() ) {
 607                                 $out .= '    ';
 608                                 $out .= Xml::element( 'redirect', array( 'title' => self::canonicalTitle( $redirect ) ) );
 609                                 $out .= "\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 object $row
 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 )
 684                         && $row->rev_sha1
 685                         && !( $row->rev_deleted & Revision::DELETED_TEXT )
 686                 ) {
 687                         $out .= "      " . Xml::element( 'sha1', null, strval( $row->rev_sha1 ) ) . "\n";
 688                 } else {
 689                         $out .= "      <sha1/>\n";
 690                 }
 691
 692                 if ( isset( $row->rev_content_model ) && !is_null( $row->rev_content_model ) ) {
 693                         $content_model = strval( $row->rev_content_model );
 694                 } else {
 695                         // probably using $wgContentHandlerUseDB = false;
 696                         // @todo test!
 697                         $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 698                         $content_model = ContentHandler::getDefaultModelFor( $title );
 699                 }
 700
 701                 $out .= "      " . Xml::element( 'model', null, strval( $content_model ) ) . "\n";
 702
 703                 if ( isset( $row->rev_content_format ) && !is_null( $row->rev_content_format ) ) {
 704                         $content_format = strval( $row->rev_content_format );
 705                 } else {
 706                         // probably using $wgContentHandlerUseDB = false;
 707                         // @todo test!
 708                         $content_handler = ContentHandler::getForModelID( $content_model );
 709                         $content_format = $content_handler->getDefaultFormat();
 710                 }
 711
 712                 $out .= "      " . Xml::element( 'format', null, strval( $content_format ) ) . "\n";
 713
 714                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 715
 716                 $out .= "    </revision>\n";
 717
 718                 wfProfileOut( __METHOD__ );
 719                 return $out;
 720         }
 721
 722         /**
 723          * Dumps a "<logitem>" section on the output stream, with
 724          * data filled in from the given database row.
 725          *
 726          * @param object $row
 727          * @return string
 728          * @access private
 729          */
 730         function writeLogItem( $row ) {
 731                 wfProfileIn( __METHOD__ );
 732
 733                 $out = "  <logitem>\n";
 734                 $out .= "    " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 735
 736                 $out .= $this->writeTimestamp( $row->log_timestamp, "    " );
 737
 738                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 739                         $out .= "    " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 740                 } else {
 741                         $out .= $this->writeContributor( $row->log_user, $row->user_name, "    " );
 742                 }
 743
 744                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 745                         $out .= "    " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 746                 } elseif ( $row->log_comment != '' ) {
 747                         $out .= "    " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 748                 }
 749
 750                 $out .= "    " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 751                 $out .= "    " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 752
 753                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 754                         $out .= "    " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 755                 } else {
 756                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 757                         $out .= "    " . Xml::elementClean( 'logtitle', null, self::canonicalTitle( $title ) ) . "\n";
 758                         $out .= "    " . Xml::elementClean( 'params',
 759                                 array( 'xml:space' => 'preserve' ),
 760                                 strval( $row->log_params ) ) . "\n";
 761                 }
 762
 763                 $out .= "  </logitem>\n";
 764
 765                 wfProfileOut( __METHOD__ );
 766                 return $out;
 767         }
 768
 769         /**
 770          * @param string $timestamp
 771          * @param string $indent Default to six spaces
 772          * @return string
 773          */
 774         function writeTimestamp( $timestamp, $indent = "      " ) {
 775                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 776                 return $indent . Xml::element( 'timestamp', null, $ts ) . "\n";
 777         }
 778
 779         /**
 780          * @param int $id
 781          * @param string $text
 782          * @param string $indent Default to six spaces
 783          * @return string
 784          */
 785         function writeContributor( $id, $text, $indent = "      " ) {
 786                 $out = $indent . "<contributor>\n";
 787                 if ( $id || !IP::isValid( $text ) ) {
 788                         $out .= $indent . "  " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 789                         $out .= $indent . "  " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 790                 } else {
 791                         $out .= $indent . "  " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 792                 }
 793                 $out .= $indent . "</contributor>\n";
 794                 return $out;
 795         }
 796
 797         /**
 798          * Warning! This data is potentially inconsistent. :(
 799          * @param object $row
 800          * @param bool $dumpContents
 801          * @return string
 802          */
 803         function writeUploads( $row, $dumpContents = false ) {
 804                 if ( $row->page_namespace == NS_FILE ) {
 805                         $img = wfLocalFile( $row->page_title );
 806                         if ( $img && $img->exists() ) {
 807                                 $out = '';
 808                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 809                                         $out .= $this->writeUpload( $ver, $dumpContents );
 810                                 }
 811                                 $out .= $this->writeUpload( $img, $dumpContents );
 812                                 return $out;
 813                         }
 814                 }
 815                 return '';
 816         }
 817
 818         /**
 819          * @param File $file
 820          * @param bool $dumpContents
 821          * @return string
 822          */
 823         function writeUpload( $file, $dumpContents = false ) {
 824                 if ( $file->isOld() ) {
 825                         $archiveName = "      " .
 826                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 827                 } else {
 828                         $archiveName = '';
 829                 }
 830                 if ( $dumpContents ) {
 831                         $be = $file->getRepo()->getBackend();
 832                         # Dump file as base64
 833                         # Uses only XML-safe characters, so does not need escaping
 834                         # @todo Too bad this loads the contents into memory (script might swap)
 835                         $contents = '      <contents encoding="base64">' .
 836                                 chunk_split( base64_encode(
 837                                         $be->getFileContents( array( 'src' => $file->getPath() ) ) ) ) .
 838                                 "      </contents>\n";
 839                 } else {
 840                         $contents = '';
 841                 }
 842                 if ( $file->isDeleted( File::DELETED_COMMENT ) ) {
 843                         $comment = Xml::element( 'comment', array( 'deleted' => 'deleted' ) );
 844                 } else {
 845                         $comment = Xml::elementClean( 'comment', null, $file->getDescription() );
 846                 }
 847                 return "    <upload>\n" .
 848                         $this->writeTimestamp( $file->getTimestamp() ) .
 849                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 850                         "      " . $comment . "\n" .
 851                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 852                         $archiveName .
 853                         "      " . Xml::element( 'src', null, $file->getCanonicalURL() ) . "\n" .
 854                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 855                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 856                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 857                         $contents .
 858                         "    </upload>\n";
 859         }
 860
 861         /**
 862          * Return prefixed text form of title, but using the content language's
 863          * canonical namespace. This skips any special-casing such as gendered
 864          * user namespaces -- which while useful, are not yet listed in the
 865          * XML "<siteinfo>" data so are unsafe in export.
 866          *
 867          * @param Title $title
 868          * @return string
 869          * @since 1.18
 870          */
 871         public static function canonicalTitle( Title $title ) {
 872                 if ( $title->isExternal() ) {
 873                         return $title->getPrefixedText();
 874                 }
 875
 876                 global $wgContLang;
 877                 $prefix = str_replace( '_', ' ', $wgContLang->getNsText( $title->getNamespace() ) );
 878
 879                 if ( $prefix !== '' ) {
 880                         $prefix .= ':';
 881                 }
 882
 883                 return $prefix . $title->getText();
 884         }
 885 }
 886
 887 /**
 888  * Base class for output stream; prints to stdout or buffer or wherever.
 889  * @ingroup Dump
 890  */
 891 class DumpOutput {
 892
 893         /**
 894          * @param string $string
 895          */
 896         function writeOpenStream( $string ) {
 897                 $this->write( $string );
 898         }
 899
 900         /**
 901          * @param string $string
 902          */
 903         function writeCloseStream( $string ) {
 904                 $this->write( $string );
 905         }
 906
 907         /**
 908          * @param object $page
 909          * @param string $string
 910          */
 911         function writeOpenPage( $page, $string ) {
 912                 $this->write( $string );
 913         }
 914
 915         /**
 916          * @param string $string
 917          */
 918         function writeClosePage( $string ) {
 919                 $this->write( $string );
 920         }
 921
 922         /**
 923          * @param object $rev
 924          * @param string $string
 925          */
 926         function writeRevision( $rev, $string ) {
 927                 $this->write( $string );
 928         }
 929
 930         /**
 931          * @param object $rev
 932          * @param string $string
 933          */
 934         function writeLogItem( $rev, $string ) {
 935                 $this->write( $string );
 936         }
 937
 938         /**
 939          * Override to write to a different stream type.
 940          * @param string $string
 941          * @return bool
 942          */
 943         function write( $string ) {
 944                 print $string;
 945         }
 946
 947         /**
 948          * Close the old file, move it to a specified name,
 949          * and reopen new file with the old name. Use this
 950          * for writing out a file in multiple pieces
 951          * at specified checkpoints (e.g. every n hours).
 952          * @param string|array $newname File name. May be a string or an array with one element
 953          */
 954         function closeRenameAndReopen( $newname ) {
 955         }
 956
 957         /**
 958          * Close the old file, and move it to a specified name.
 959          * Use this for the last piece of a file written out
 960          * at specified checkpoints (e.g. every n hours).
 961          * @param string|array $newname File name. May be a string or an array with one element
 962          * @param bool $open If true, a new file with the old filename will be opened
 963          *   again for writing (default: false)
 964          */
 965         function closeAndRename( $newname, $open = false ) {
 966         }
 967
 968         /**
 969          * Returns the name of the file or files which are
 970          * being written to, if there are any.
 971          * @return null
 972          */
 973         function getFilenames() {
 974                 return null;
 975         }
 976 }
 977
 978 /**
 979  * Stream outputter to send data to a file.
 980  * @ingroup Dump
 981  */
 982 class DumpFileOutput extends DumpOutput {
 983         protected $handle = false, $filename;
 984
 985         /**
 986          * @param string $file
 987          */
 988         function __construct( $file ) {
 989                 $this->handle = fopen( $file, "wt" );
 990                 $this->filename = $file;
 991         }
 992
 993         /**
 994          * @param string $string
 995          */
 996         function writeCloseStream( $string ) {
 997                 parent::writeCloseStream( $string );
 998                 if ( $this->handle ) {
 999                         fclose( $this->handle );
1000                         $this->handle = false;
1001                 }
1002         }
1003
1004         /**
1005          * @param string $string
1006          */
1007         function write( $string ) {
1008                 fputs( $this->handle, $string );
1009         }
1010
1011         /**
1012          * @param string $newname
1013          */
1014         function closeRenameAndReopen( $newname ) {
1015                 $this->closeAndRename( $newname, true );
1016         }
1017
1018         /**
1019          * @param string $newname
1020          * @throws MWException
1021          */
1022         function renameOrException( $newname ) {
1023                         if ( !rename( $this->filename, $newname ) ) {
1024                                 throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" );
1025                         }
1026         }
1027
1028         /**
1029          * @param array $newname
1030          * @return string
1031          * @throws MWException
1032          */
1033         function checkRenameArgCount( $newname ) {
1034                 if ( is_array( $newname ) ) {
1035                         if ( count( $newname ) > 1 ) {
1036                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
1037                         } else {
1038                                 $newname = $newname[0];
1039                         }
1040                 }
1041                 return $newname;
1042         }
1043
1044         /**
1045          * @param string $newname
1046          * @param bool $open
1047          */
1048         function closeAndRename( $newname, $open = false ) {
1049                 $newname = $this->checkRenameArgCount( $newname );
1050                 if ( $newname ) {
1051                         if ( $this->handle ) {
1052                                 fclose( $this->handle );
1053                                 $this->handle = false;
1054                         }
1055                         $this->renameOrException( $newname );
1056                         if ( $open ) {
1057                                 $this->handle = fopen( $this->filename, "wt" );
1058                         }
1059                 }
1060         }
1061
1062         /**
1063          * @return string|null
1064          */
1065         function getFilenames() {
1066                 return $this->filename;
1067         }
1068 }
1069
1070 /**
1071  * Stream outputter to send data to a file via some filter program.
1072  * Even if compression is available in a library, using a separate
1073  * program can allow us to make use of a multi-processor system.
1074  * @ingroup Dump
1075  */
1076 class DumpPipeOutput extends DumpFileOutput {
1077         protected $command, $filename;
1078         protected $procOpenResource = false;
1079
1080         /**
1081          * @param string $command
1082          * @param string $file
1083          */
1084         function __construct( $command, $file = null ) {
1085                 if ( !is_null( $file ) ) {
1086                         $command .= " > " . wfEscapeShellArg( $file );
1087                 }
1088
1089                 $this->startCommand( $command );
1090                 $this->command = $command;
1091                 $this->filename = $file;
1092         }
1093
1094         /**
1095          * @param string $string
1096          */
1097         function writeCloseStream( $string ) {
1098                 parent::writeCloseStream( $string );
1099                 if ( $this->procOpenResource ) {
1100                         proc_close( $this->procOpenResource );
1101                         $this->procOpenResource = false;
1102                 }
1103         }
1104
1105         /**
1106          * @param string $command
1107          */
1108         function startCommand( $command ) {
1109                 $spec = array(
1110                         0 => array( "pipe", "r" ),
1111                 );
1112                 $pipes = array();
1113                 $this->procOpenResource = proc_open( $command, $spec, $pipes );
1114                 $this->handle = $pipes[0];
1115         }
1116
1117         /**
1118          * @param string $newname
1119          */
1120         function closeRenameAndReopen( $newname ) {
1121                 $this->closeAndRename( $newname, true );
1122         }
1123
1124         /**
1125          * @param string $newname
1126          * @param bool $open
1127          */
1128         function closeAndRename( $newname, $open = false ) {
1129                 $newname = $this->checkRenameArgCount( $newname );
1130                 if ( $newname ) {
1131                         if ( $this->handle ) {
1132                                 fclose( $this->handle );
1133                                 $this->handle = false;
1134                         }
1135                         if ( $this->procOpenResource ) {
1136                                 proc_close( $this->procOpenResource );
1137                                 $this->procOpenResource = false;
1138                         }
1139                         $this->renameOrException( $newname );
1140                         if ( $open ) {
1141                                 $command = $this->command;
1142                                 $command .= " > " . wfEscapeShellArg( $this->filename );
1143                                 $this->startCommand( $command );
1144                         }
1145                 }
1146         }
1147
1148 }
1149
1150 /**
1151  * Sends dump output via the gzip compressor.
1152  * @ingroup Dump
1153  */
1154 class DumpGZipOutput extends DumpPipeOutput {
1155
1156         /**
1157          * @param string $file
1158          */
1159         function __construct( $file ) {
1160                 parent::__construct( "gzip", $file );
1161         }
1162 }
1163
1164 /**
1165  * Sends dump output via the bgzip2 compressor.
1166  * @ingroup Dump
1167  */
1168 class DumpBZip2Output extends DumpPipeOutput {
1169
1170         /**
1171          * @param string $file
1172          */
1173         function __construct( $file ) {
1174                 parent::__construct( "bzip2", $file );
1175         }
1176 }
1177
1178 /**
1179  * Sends dump output via the p7zip compressor.
1180  * @ingroup Dump
1181  */
1182 class Dump7ZipOutput extends DumpPipeOutput {
1183
1184         /**
1185          * @param string $file
1186          */
1187         function __construct( $file ) {
1188                 $command = $this->setup7zCommand( $file );
1189                 parent::__construct( $command );
1190                 $this->filename = $file;
1191         }
1192
1193         /**
1194          * @param string $file
1195          * @return string
1196          */
1197         function setup7zCommand( $file ) {
1198                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
1199                 // Suppress annoying useless crap from p7zip
1200                 // Unfortunately this could suppress real error messages too
1201                 $command .= ' >' . wfGetNull() . ' 2>&1';
1202                 return $command;
1203         }
1204
1205         /**
1206          * @param string $newname
1207          * @param bool $open
1208          */
1209         function closeAndRename( $newname, $open = false ) {
1210                 $newname = $this->checkRenameArgCount( $newname );
1211                 if ( $newname ) {
1212                         fclose( $this->handle );
1213                         proc_close( $this->procOpenResource );
1214                         $this->renameOrException( $newname );
1215                         if ( $open ) {
1216                                 $command = $this->setup7zCommand( $this->filename );
1217                                 $this->startCommand( $command );
1218                         }
1219                 }
1220         }
1221 }
1222
1223 /**
1224  * Dump output filter class.
1225  * This just does output filtering and streaming; XML formatting is done
1226  * higher up, so be careful in what you do.
1227  * @ingroup Dump
1228  */
1229 class DumpFilter {
1230
1231         /**
1232          * @var DumpOutput
1233          * FIXME will need to be made protected whenever legacy code
1234          * is updated.
1235          */
1236         public $sink;
1237
1238         /**
1239          * @var bool
1240          */
1241         protected $sendingThisPage;
1242
1243         /**
1244          * @param DumpOutput $sink
1245          */
1246         function __construct( &$sink ) {
1247                 $this->sink =& $sink;
1248         }
1249
1250         /**
1251          * @param string $string
1252          */
1253         function writeOpenStream( $string ) {
1254                 $this->sink->writeOpenStream( $string );
1255         }
1256
1257         /**
1258          * @param string $string
1259          */
1260         function writeCloseStream( $string ) {
1261                 $this->sink->writeCloseStream( $string );
1262         }
1263
1264         /**
1265          * @param object $page
1266          * @param string $string
1267          */
1268         function writeOpenPage( $page, $string ) {
1269                 $this->sendingThisPage = $this->pass( $page, $string );
1270                 if ( $this->sendingThisPage ) {
1271                         $this->sink->writeOpenPage( $page, $string );
1272                 }
1273         }
1274
1275         /**
1276          * @param string $string
1277          */
1278         function writeClosePage( $string ) {
1279                 if ( $this->sendingThisPage ) {
1280                         $this->sink->writeClosePage( $string );
1281                         $this->sendingThisPage = false;
1282                 }
1283         }
1284
1285         /**
1286          * @param object $rev
1287          * @param string $string
1288          */
1289         function writeRevision( $rev, $string ) {
1290                 if ( $this->sendingThisPage ) {
1291                         $this->sink->writeRevision( $rev, $string );
1292                 }
1293         }
1294
1295         /**
1296          * @param object $rev
1297          * @param string $string
1298          */
1299         function writeLogItem( $rev, $string ) {
1300                 $this->sink->writeRevision( $rev, $string );
1301         }
1302
1303         /**
1304          * @param string $newname
1305          */
1306         function closeRenameAndReopen( $newname ) {
1307                 $this->sink->closeRenameAndReopen( $newname );
1308         }
1309
1310         /**
1311          * @param string $newname
1312          * @param bool $open
1313          */
1314         function closeAndRename( $newname, $open = false ) {
1315                 $this->sink->closeAndRename( $newname, $open );
1316         }
1317
1318         /**
1319          * @return array
1320          */
1321         function getFilenames() {
1322                 return $this->sink->getFilenames();
1323         }
1324
1325         /**
1326          * Override for page-based filter types.
1327          * @param object $page
1328          * @return bool
1329          */
1330         function pass( $page ) {
1331                 return true;
1332         }
1333 }
1334
1335 /**
1336  * Simple dump output filter to exclude all talk pages.
1337  * @ingroup Dump
1338  */
1339 class DumpNotalkFilter extends DumpFilter {
1340
1341         /**
1342          * @param object $page
1343          * @return bool
1344          */
1345         function pass( $page ) {
1346                 return !MWNamespace::isTalk( $page->page_namespace );
1347         }
1348 }
1349
1350 /**
1351  * Dump output filter to include or exclude pages in a given set of namespaces.
1352  * @ingroup Dump
1353  */
1354 class DumpNamespaceFilter extends DumpFilter {
1355         var $invert = false;
1356         var $namespaces = array();
1357
1358         /**
1359          * @param DumpOutput $sink
1360          * @param array $param
1361          * @throws MWException
1362          */
1363         function __construct( &$sink, $param ) {
1364                 parent::__construct( $sink );
1365
1366                 $constants = array(
1367                         "NS_MAIN"           => NS_MAIN,
1368                         "NS_TALK"           => NS_TALK,
1369                         "NS_USER"           => NS_USER,
1370                         "NS_USER_TALK"      => NS_USER_TALK,
1371                         "NS_PROJECT"        => NS_PROJECT,
1372                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
1373                         "NS_FILE"           => NS_FILE,
1374                         "NS_FILE_TALK"      => NS_FILE_TALK,
1375                         "NS_IMAGE"          => NS_IMAGE, // NS_IMAGE is an alias for NS_FILE
1376                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
1377                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
1378                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
1379                         "NS_TEMPLATE"       => NS_TEMPLATE,
1380                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
1381                         "NS_HELP"           => NS_HELP,
1382                         "NS_HELP_TALK"      => NS_HELP_TALK,
1383                         "NS_CATEGORY"       => NS_CATEGORY,
1384                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
1385
1386                 if ( $param { 0 } == '!' ) {
1387                         $this->invert = true;
1388                         $param = substr( $param, 1 );
1389                 }
1390
1391                 foreach ( explode( ',', $param ) as $key ) {
1392                         $key = trim( $key );
1393                         if ( isset( $constants[$key] ) ) {
1394                                 $ns = $constants[$key];
1395                                 $this->namespaces[$ns] = true;
1396                         } elseif ( is_numeric( $key ) ) {
1397                                 $ns = intval( $key );
1398                                 $this->namespaces[$ns] = true;
1399                         } else {
1400                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
1401                         }
1402                 }
1403         }
1404
1405         /**
1406          * @param object $page
1407          * @return bool
1408          */
1409         function pass( $page ) {
1410                 $match = isset( $this->namespaces[$page->page_namespace] );
1411                 return $this->invert xor $match;
1412         }
1413 }
1414
1415 /**
1416  * Dump output filter to include only the last revision in each page sequence.
1417  * @ingroup Dump
1418  */
1419 class DumpLatestFilter extends DumpFilter {
1420         var $page, $pageString, $rev, $revString;
1421
1422         /**
1423          * @param object $page
1424          * @param string $string
1425          */
1426         function writeOpenPage( $page, $string ) {
1427                 $this->page = $page;
1428                 $this->pageString = $string;
1429         }
1430
1431         /**
1432          * @param string $string
1433          */
1434         function writeClosePage( $string ) {
1435                 if ( $this->rev ) {
1436                         $this->sink->writeOpenPage( $this->page, $this->pageString );
1437                         $this->sink->writeRevision( $this->rev, $this->revString );
1438                         $this->sink->writeClosePage( $string );
1439                 }
1440                 $this->rev = null;
1441                 $this->revString = null;
1442                 $this->page = null;
1443                 $this->pageString = null;
1444         }
1445
1446         /**
1447          * @param object $rev
1448          * @param string $string
1449          */
1450         function writeRevision( $rev, $string ) {
1451                 if ( $rev->rev_id == $this->page->page_latest ) {
1452                         $this->rev = $rev;
1453                         $this->revString = $string;
1454                 }
1455         }
1456 }
1457
1458 /**
1459  * Base class for output stream; prints to stdout or buffer or wherever.
1460  * @ingroup Dump
1461  */
1462 class DumpMultiWriter {
1463
1464         /**
1465          * @param array $sinks
1466          */
1467         function __construct( $sinks ) {
1468                 $this->sinks = $sinks;
1469                 $this->count = count( $sinks );
1470         }
1471
1472         /**
1473          * @param string $string
1474          */
1475         function writeOpenStream( $string ) {
1476                 for ( $i = 0; $i < $this->count; $i++ ) {
1477                         $this->sinks[$i]->writeOpenStream( $string );
1478                 }
1479         }
1480
1481         /**
1482          * @param string $string
1483          */
1484         function writeCloseStream( $string ) {
1485                 for ( $i = 0; $i < $this->count; $i++ ) {
1486                         $this->sinks[$i]->writeCloseStream( $string );
1487                 }
1488         }
1489
1490         /**
1491          * @param object $page
1492          * @param string $string
1493          */
1494         function writeOpenPage( $page, $string ) {
1495                 for ( $i = 0; $i < $this->count; $i++ ) {
1496                         $this->sinks[$i]->writeOpenPage( $page, $string );
1497                 }
1498         }
1499
1500         /**
1501          * @param string $string
1502          */
1503         function writeClosePage( $string ) {
1504                 for ( $i = 0; $i < $this->count; $i++ ) {
1505                         $this->sinks[$i]->writeClosePage( $string );
1506                 }
1507         }
1508
1509         /**
1510          * @param object $rev
1511          * @param string $string
1512          */
1513         function writeRevision( $rev, $string ) {
1514                 for ( $i = 0; $i < $this->count; $i++ ) {
1515                         $this->sinks[$i]->writeRevision( $rev, $string );
1516                 }
1517         }
1518
1519         /**
1520          * @param array $newnames
1521          */
1522         function closeRenameAndReopen( $newnames ) {
1523                 $this->closeAndRename( $newnames, true );
1524         }
1525
1526         /**
1527          * @param array $newnames
1528          * @param bool $open
1529          */
1530         function closeAndRename( $newnames, $open = false ) {
1531                 for ( $i = 0; $i < $this->count; $i++ ) {
1532                         $this->sinks[$i]->closeAndRename( $newnames[$i], $open );
1533                 }
1534         }
1535
1536         /**
1537          * @return array
1538          */
1539         function getFilenames() {
1540                 $filenames = array();
1541                 for ( $i = 0; $i < $this->count; $i++ ) {
1542                         $filenames[] = $this->sinks[$i]->getFilenames();
1543                 }
1544                 return $filenames;
1545         }
1546
1547 }
1548
1549 /**
1550  * @param string $string
1551  * @return string
1552  */
1553 function xmlsafe( $string ) {
1554         wfProfileIn( __FUNCTION__ );
1555
1556         /**
1557          * The page may contain old data which has not been properly normalized.
1558          * Invalid UTF-8 sequences or forbidden control characters will make our
1559          * XML output invalid, so be sure to strip them out.
1560          */
1561         $string = UtfNormal::cleanUp( $string );
1562
1563         $string = htmlspecialchars( $string );
1564         wfProfileOut( __FUNCTION__ );
1565         return $string;
1566 }