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 bool Return distinct author list (when not returning full history) */
  35         public $list_authors = false;
  36
  37         /** @var bool */
  38         public $dumpUploads = false;
  39
  40         /** @var bool */
  41         public $dumpUploadFileContents = false;
  42
  43         /** @var string */
  44         protected $author_list = "";
  45
  46         const FULL = 1;
  47         const CURRENT = 2;
  48         const STABLE = 4; // extension defined
  49         const LOGS = 8;
  50         const RANGE = 16;
  51
  52         const BUFFER = 0;
  53         const STREAM = 1;
  54
  55         const TEXT = 0;
  56         const STUB = 1;
  57
  58         /** @var int */
  59         protected $buffer;
  60
  61         /** @var int */
  62         protected $text;
  63
  64         /** @var DumpOutput */
  65         protected $sink;
  66
  67         /**
  68          * Returns the export schema version.
  69          * @return string
  70          */
  71         public static function schemaVersion() {
  72                 return "0.8";
  73         }
  74
  75         /**
  76          * If using WikiExporter::STREAM to stream a large amount of data,
  77          * provide a database connection which is not managed by
  78          * LoadBalancer to read from: some history blob types will
  79          * make additional queries to pull source data while the
  80          * main query is still running.
  81          *
  82          * @param DatabaseBase $db
  83          * @param int|array $history One of WikiExporter::FULL, WikiExporter::CURRENT,
  84          *   WikiExporter::RANGE or WikiExporter::STABLE, or an associative array:
  85          *   - offset: non-inclusive offset at which to start the query
  86          *   - limit: maximum number of rows to return
  87          *   - dir: "asc" or "desc" timestamp order
  88          * @param int $buffer One of WikiExporter::BUFFER or WikiExporter::STREAM
  89          * @param int $text One of WikiExporter::TEXT or WikiExporter::STUB
  90          */
  91         function __construct( $db, $history = WikiExporter::CURRENT,
  92                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  93                 $this->db = $db;
  94                 $this->history = $history;
  95                 $this->buffer = $buffer;
  96                 $this->writer = new XmlDumpWriter();
  97                 $this->sink = new DumpOutput();
  98                 $this->text = $text;
  99         }
 100
 101         /**
 102          * Set the DumpOutput or DumpFilter object which will receive
 103          * various row objects and XML output for filtering. Filters
 104          * can be chained or used as callbacks.
 105          *
 106          * @param DumpOutput $sink
 107          */
 108         public function setOutputSink( &$sink ) {
 109                 $this->sink =& $sink;
 110         }
 111
 112         public function openStream() {
 113                 $output = $this->writer->openStream();
 114                 $this->sink->writeOpenStream( $output );
 115         }
 116
 117         public function closeStream() {
 118                 $output = $this->writer->closeStream();
 119                 $this->sink->writeCloseStream( $output );
 120         }
 121
 122         /**
 123          * Dumps a series of page and revision records for all pages
 124          * in the database, either including complete history or only
 125          * the most recent version.
 126          */
 127         public function allPages() {
 128                 $this->dumpFrom( '' );
 129         }
 130
 131         /**
 132          * Dumps a series of page and revision records for those pages
 133          * in the database falling within the page_id range given.
 134          * @param int $start Inclusive lower limit (this id is included)
 135          * @param int $end Exclusive upper limit (this id is not included)
 136          *   If 0, no upper limit.
 137          */
 138         public function pagesByRange( $start, $end ) {
 139                 $condition = 'page_id >= ' . intval( $start );
 140                 if ( $end ) {
 141                         $condition .= ' AND page_id < ' . intval( $end );
 142                 }
 143                 $this->dumpFrom( $condition );
 144         }
 145
 146         /**
 147          * Dumps a series of page and revision records for those pages
 148          * in the database with revisions falling within the rev_id range given.
 149          * @param int $start Inclusive lower limit (this id is included)
 150          * @param int $end Exclusive upper limit (this id is not included)
 151          *   If 0, no upper limit.
 152          */
 153         public function revsByRange( $start, $end ) {
 154                 $condition = 'rev_id >= ' . intval( $start );
 155                 if ( $end ) {
 156                         $condition .= ' AND rev_id < ' . intval( $end );
 157                 }
 158                 $this->dumpFrom( $condition );
 159         }
 160
 161         /**
 162          * @param Title $title
 163          */
 164         public function pageByTitle( $title ) {
 165                 $this->dumpFrom(
 166                         'page_namespace=' . $title->getNamespace() .
 167                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 168         }
 169
 170         /**
 171          * @param string $name
 172          * @throws MWException
 173          */
 174         public function pageByName( $name ) {
 175                 $title = Title::newFromText( $name );
 176                 if ( is_null( $title ) ) {
 177                         throw new MWException( "Can't export invalid title" );
 178                 } else {
 179                         $this->pageByTitle( $title );
 180                 }
 181         }
 182
 183         /**
 184          * @param array $names
 185          */
 186         public function pagesByName( $names ) {
 187                 foreach ( $names as $name ) {
 188                         $this->pageByName( $name );
 189                 }
 190         }
 191
 192         public function allLogs() {
 193                 $this->dumpFrom( '' );
 194         }
 195
 196         /**
 197          * @param int $start
 198          * @param int $end
 199          */
 200         public function logsByRange( $start, $end ) {
 201                 $condition = 'log_id >= ' . intval( $start );
 202                 if ( $end ) {
 203                         $condition .= ' AND log_id < ' . intval( $end );
 204                 }
 205                 $this->dumpFrom( $condition );
 206         }
 207
 208         /**
 209          * Generates the distinct list of authors of an article
 210          * Not called by default (depends on $this->list_authors)
 211          * Can be set by Special:Export when not exporting whole history
 212          *
 213          * @param array $cond
 214          */
 215         protected function do_list_authors( $cond ) {
 216                 wfProfileIn( __METHOD__ );
 217                 $this->author_list = "<contributors>";
 218                 // rev_deleted
 219
 220                 $res = $this->db->select(
 221                         array( 'page', 'revision' ),
 222                         array( 'DISTINCT rev_user_text', 'rev_user' ),
 223                         array(
 224                                 $this->db->bitAnd( 'rev_deleted', Revision::DELETED_USER ) . ' = 0',
 225                                 $cond,
 226                                 'page_id = rev_id',
 227                         ),
 228                         __METHOD__
 229                 );
 230
 231                 foreach ( $res as $row ) {
 232                         $this->author_list .= "<contributor>" .
 233                                 "<username>" .
 234                                 htmlentities( $row->rev_user_text ) .
 235                                 "</username>" .
 236                                 "<id>" .
 237                                 $row->rev_user .
 238                                 "</id>" .
 239                                 "</contributor>";
 240                 }
 241                 $this->author_list .= "</contributors>";
 242                 wfProfileOut( __METHOD__ );
 243         }
 244
 245         /**
 246          * @param string $cond
 247          * @throws MWException
 248          * @throws Exception
 249          */
 250         protected function dumpFrom( $cond = '' ) {
 251                 wfProfileIn( __METHOD__ );
 252                 # For logging dumps...
 253                 if ( $this->history & self::LOGS ) {
 254                         $where = array( 'user_id = log_user' );
 255                         # Hide private logs
 256                         $hideLogs = LogEventsList::getExcludeClause( $this->db );
 257                         if ( $hideLogs ) {
 258                                 $where[] = $hideLogs;
 259                         }
 260                         # Add on any caller specified conditions
 261                         if ( $cond ) {
 262                                 $where[] = $cond;
 263                         }
 264                         # Get logging table name for logging.* clause
 265                         $logging = $this->db->tableName( 'logging' );
 266
 267                         if ( $this->buffer == WikiExporter::STREAM ) {
 268                                 $prev = $this->db->bufferResults( false );
 269                         }
 270                         $wrapper = null; // Assuring $wrapper is not undefined, if exception occurs early
 271                         try {
 272                                 $result = $this->db->select( array( 'logging', 'user' ),
 273                                         array( "{$logging}.*", 'user_name' ), // grab the user name
 274                                         $where,
 275                                         __METHOD__,
 276                                         array( 'ORDER BY' => 'log_id', 'USE INDEX' => array( 'logging' => 'PRIMARY' ) )
 277                                 );
 278                                 $wrapper = $this->db->resultObject( $result );
 279                                 $this->outputLogStream( $wrapper );
 280                                 if ( $this->buffer == WikiExporter::STREAM ) {
 281                                         $this->db->bufferResults( $prev );
 282                                 }
 283                         } catch ( Exception $e ) {
 284                                 // Throwing the exception does not reliably free the resultset, and
 285                                 // would also leave the connection in unbuffered mode.
 286
 287                                 // Freeing result
 288                                 try {
 289                                         if ( $wrapper ) {
 290                                                 $wrapper->free();
 291                                         }
 292                                 } catch ( Exception $e2 ) {
 293                                         // Already in panic mode -> ignoring $e2 as $e has
 294                                         // higher priority
 295                                 }
 296
 297                                 // Putting database back in previous buffer mode
 298                                 try {
 299                                         if ( $this->buffer == WikiExporter::STREAM ) {
 300                                                 $this->db->bufferResults( $prev );
 301                                         }
 302                                 } catch ( Exception $e2 ) {
 303                                         // Already in panic mode -> ignoring $e2 as $e has
 304                                         // higher priority
 305                                 }
 306
 307                                 // Inform caller about problem
 308                                 wfProfileOut( __METHOD__ );
 309                                 throw $e;
 310                         }
 311                 # For page dumps...
 312                 } else {
 313                         $tables = array( 'page', 'revision' );
 314                         $opts = array( 'ORDER BY' => 'page_id ASC' );
 315                         $opts['USE INDEX'] = array();
 316                         $join = array();
 317                         if ( is_array( $this->history ) ) {
 318                                 # Time offset/limit for all pages/history...
 319                                 $revJoin = 'page_id=rev_page';
 320                                 # Set time order
 321                                 if ( $this->history['dir'] == 'asc' ) {
 322                                         $op = '>';
 323                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 324                                 } else {
 325                                         $op = '<';
 326                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 327                                 }
 328                                 # Set offset
 329                                 if ( !empty( $this->history['offset'] ) ) {
 330                                         $revJoin .= " AND rev_timestamp $op " .
 331                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 332                                 }
 333                                 $join['revision'] = array( 'INNER JOIN', $revJoin );
 334                                 # Set query limit
 335                                 if ( !empty( $this->history['limit'] ) ) {
 336                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 337                                 }
 338                         } elseif ( $this->history & WikiExporter::FULL ) {
 339                                 # Full history dumps...
 340                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 341                         } elseif ( $this->history & WikiExporter::CURRENT ) {
 342                                 # Latest revision dumps...
 343                                 if ( $this->list_authors && $cond != '' ) { // List authors, if so desired
 344                                         $this->do_list_authors( $cond );
 345                                 }
 346                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 347                         } elseif ( $this->history & WikiExporter::STABLE ) {
 348                                 # "Stable" revision dumps...
 349                                 # Default JOIN, to be overridden...
 350                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 351                                 # One, and only one hook should set this, and return false
 352                                 if ( wfRunHooks( 'WikiExporter::dumpStableQuery', array( &$tables, &$opts, &$join ) ) ) {
 353                                         wfProfileOut( __METHOD__ );
 354                                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 355                                 }
 356                         } elseif ( $this->history & WikiExporter::RANGE ) {
 357                                 # Dump of revisions within a specified range
 358                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 359                                 $opts['ORDER BY'] = array( 'rev_page ASC', 'rev_id ASC' );
 360                         } else {
 361                                 # Unknown history specification parameter?
 362                                 wfProfileOut( __METHOD__ );
 363                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 364                         }
 365                         # Query optimization hacks
 366                         if ( $cond == '' ) {
 367                                 $opts[] = 'STRAIGHT_JOIN';
 368                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 369                         }
 370                         # Build text join options
 371                         if ( $this->text != WikiExporter::STUB ) { // 1-pass
 372                                 $tables[] = 'text';
 373                                 $join['text'] = array( 'INNER JOIN', 'rev_text_id=old_id' );
 374                         }
 375
 376                         if ( $this->buffer == WikiExporter::STREAM ) {
 377                                 $prev = $this->db->bufferResults( false );
 378                         }
 379
 380                         $wrapper = null; // Assuring $wrapper is not undefined, if exception occurs early
 381                         try {
 382                                 wfRunHooks( 'ModifyExportQuery',
 383                                                 array( $this->db, &$tables, &$cond, &$opts, &$join ) );
 384
 385                                 # Do the query!
 386                                 $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 387                                 $wrapper = $this->db->resultObject( $result );
 388                                 # Output dump results
 389                                 $this->outputPageStream( $wrapper );
 390
 391                                 if ( $this->buffer == WikiExporter::STREAM ) {
 392                                         $this->db->bufferResults( $prev );
 393                                 }
 394                         } catch ( Exception $e ) {
 395                                 // Throwing the exception does not reliably free the resultset, and
 396                                 // would also leave the connection in unbuffered mode.
 397
 398                                 // Freeing result
 399                                 try {
 400                                         if ( $wrapper ) {
 401                                                 $wrapper->free();
 402                                         }
 403                                 } catch ( Exception $e2 ) {
 404                                         // Already in panic mode -> ignoring $e2 as $e has
 405                                         // higher priority
 406                                 }
 407
 408                                 // Putting database back in previous buffer mode
 409                                 try {
 410                                         if ( $this->buffer == WikiExporter::STREAM ) {
 411                                                 $this->db->bufferResults( $prev );
 412                                         }
 413                                 } catch ( Exception $e2 ) {
 414                                         // Already in panic mode -> ignoring $e2 as $e has
 415                                         // higher priority
 416                                 }
 417
 418                                 // Inform caller about problem
 419                                 throw $e;
 420                         }
 421                 }
 422                 wfProfileOut( __METHOD__ );
 423         }
 424
 425         /**
 426          * Runs through a query result set dumping page and revision records.
 427          * The result set should be sorted/grouped by page to avoid duplicate
 428          * page records in the output.
 429          *
 430          * Should be safe for
 431          * streaming (non-buffered) queries, as long as it was made on a
 432          * separate database connection not managed by LoadBalancer; some
 433          * blob storage types will make queries to pull source data.
 434          *
 435          * @param ResultWrapper $resultset
 436          */
 437         protected function outputPageStream( $resultset ) {
 438                 $last = null;
 439                 foreach ( $resultset as $row ) {
 440                         if ( $last === null ||
 441                                 $last->page_namespace != $row->page_namespace ||
 442                                 $last->page_title != $row->page_title ) {
 443                                 if ( $last !== null ) {
 444                                         $output = '';
 445                                         if ( $this->dumpUploads ) {
 446                                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 447                                         }
 448                                         $output .= $this->writer->closePage();
 449                                         $this->sink->writeClosePage( $output );
 450                                 }
 451                                 $output = $this->writer->openPage( $row );
 452                                 $this->sink->writeOpenPage( $row, $output );
 453                                 $last = $row;
 454                         }
 455                         $output = $this->writer->writeRevision( $row );
 456                         $this->sink->writeRevision( $row, $output );
 457                 }
 458                 if ( $last !== null ) {
 459                         $output = '';
 460                         if ( $this->dumpUploads ) {
 461                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 462                         }
 463                         $output .= $this->author_list;
 464                         $output .= $this->writer->closePage();
 465                         $this->sink->writeClosePage( $output );
 466                 }
 467         }
 468
 469         /**
 470          * @param array $resultset
 471          */
 472         protected function outputLogStream( $resultset ) {
 473                 foreach ( $resultset as $row ) {
 474                         $output = $this->writer->writeLogItem( $row );
 475                         $this->sink->writeLogItem( $row, $output );
 476                 }
 477         }
 478 }
 479
 480 /**
 481  * @ingroup Dump
 482  */
 483 class XmlDumpWriter {
 484         /**
 485          * Returns the export schema version.
 486          * @deprecated since 1.20; use WikiExporter::schemaVersion() instead
 487          * @return string
 488          */
 489         function schemaVersion() {
 490                 wfDeprecated( __METHOD__, '1.20' );
 491                 return WikiExporter::schemaVersion();
 492         }
 493
 494         /**
 495          * Opens the XML output stream's root "<mediawiki>" element.
 496          * This does not include an xml directive, so is safe to include
 497          * as a subelement in a larger XML stream. Namespace and XML Schema
 498          * references are included.
 499          *
 500          * Output will be encoded in UTF-8.
 501          *
 502          * @return string
 503          */
 504         function openStream() {
 505                 global $wgLanguageCode;
 506                 $ver = WikiExporter::schemaVersion();
 507                 return Xml::element( 'mediawiki', array(
 508                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 509                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 510                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 511                                 #TODO: how do we get a new version up there?
 512                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 513                         'version'            => $ver,
 514                         'xml:lang'           => $wgLanguageCode ),
 515                         null ) .
 516                         "\n" .
 517                         $this->siteInfo();
 518         }
 519
 520         /**
 521          * @return string
 522          */
 523         function siteInfo() {
 524                 $info = array(
 525                         $this->sitename(),
 526                         $this->homelink(),
 527                         $this->generator(),
 528                         $this->caseSetting(),
 529                         $this->namespaces() );
 530                 return "  <siteinfo>\n    " .
 531                         implode( "\n    ", $info ) .
 532                         "\n  </siteinfo>\n";
 533         }
 534
 535         /**
 536          * @return string
 537          */
 538         function sitename() {
 539                 global $wgSitename;
 540                 return Xml::element( 'sitename', array(), $wgSitename );
 541         }
 542
 543         /**
 544          * @return string
 545          */
 546         function generator() {
 547                 global $wgVersion;
 548                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 549         }
 550
 551         /**
 552          * @return string
 553          */
 554         function homelink() {
 555                 return Xml::element( 'base', array(), Title::newMainPage()->getCanonicalURL() );
 556         }
 557
 558         /**
 559          * @return string
 560          */
 561         function caseSetting() {
 562                 global $wgCapitalLinks;
 563                 // "case-insensitive" option is reserved for future
 564                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 565                 return Xml::element( 'case', array(), $sensitivity );
 566         }
 567
 568         /**
 569          * @return string
 570          */
 571         function namespaces() {
 572                 global $wgContLang;
 573                 $spaces = "<namespaces>\n";
 574                 foreach ( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 575                         $spaces .= '      ' .
 576                                 Xml::element( 'namespace',
 577                                         array(
 578                                                 'key' => $ns,
 579                                                 'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 580                                         ), $title ) . "\n";
 581                 }
 582                 $spaces .= "    </namespaces>";
 583                 return $spaces;
 584         }
 585
 586         /**
 587          * Closes the output stream with the closing root element.
 588          * Call when finished dumping things.
 589          *
 590          * @return string
 591          */
 592         function closeStream() {
 593                 return "</mediawiki>\n";
 594         }
 595
 596         /**
 597          * Opens a "<page>" section on the output stream, with data
 598          * from the given database row.
 599          *
 600          * @param object $row
 601          * @return string
 602          */
 603         public function openPage( $row ) {
 604                 $out = "  <page>\n";
 605                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 606                 $out .= '    ' . Xml::elementClean( 'title', array(), self::canonicalTitle( $title ) ) . "\n";
 607                 $out .= '    ' . Xml::element( 'ns', array(), strval( $row->page_namespace ) ) . "\n";
 608                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 609                 if ( $row->page_is_redirect ) {
 610                         $page = WikiPage::factory( $title );
 611                         $redirect = $page->getRedirectTarget();
 612                         if ( $redirect instanceof Title && $redirect->isValidRedirectTarget() ) {
 613                                 $out .= '    ';
 614                                 $out .= Xml::element( 'redirect', array( 'title' => self::canonicalTitle( $redirect ) ) );
 615                                 $out .= "\n";
 616                         }
 617                 }
 618
 619                 if ( $row->page_restrictions != '' ) {
 620                         $out .= '    ' . Xml::element( 'restrictions', array(),
 621                                 strval( $row->page_restrictions ) ) . "\n";
 622                 }
 623
 624                 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 625
 626                 return $out;
 627         }
 628
 629         /**
 630          * Closes a "<page>" section on the output stream.
 631          *
 632          * @access private
 633          * @return string
 634          */
 635         function closePage() {
 636                 return "  </page>\n";
 637         }
 638
 639         /**
 640          * Dumps a "<revision>" section on the output stream, with
 641          * data filled in from the given database row.
 642          *
 643          * @param object $row
 644          * @return string
 645          * @access private
 646          */
 647         function writeRevision( $row ) {
 648                 wfProfileIn( __METHOD__ );
 649
 650                 $out = "    <revision>\n";
 651                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 652                 if ( isset( $row->rev_parent_id ) && $row->rev_parent_id ) {
 653                         $out .= "      " . Xml::element( 'parentid', null, strval( $row->rev_parent_id ) ) . "\n";
 654                 }
 655
 656                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 657
 658                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_USER ) ) {
 659                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 660                 } else {
 661                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 662                 }
 663
 664                 if ( isset( $row->rev_minor_edit ) && $row->rev_minor_edit ) {
 665                         $out .= "      <minor/>\n";
 666                 }
 667                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_COMMENT ) ) {
 668                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 669                 } elseif ( $row->rev_comment != '' ) {
 670                         $out .= "      " . Xml::elementClean( 'comment', array(), strval( $row->rev_comment ) ) . "\n";
 671                 }
 672
 673                 if ( isset( $row->rev_content_model ) && !is_null( $row->rev_content_model ) ) {
 674                         $content_model = strval( $row->rev_content_model );
 675                 } else {
 676                         // probably using $wgContentHandlerUseDB = false;
 677                         $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 678                         $content_model = ContentHandler::getDefaultModelFor( $title );
 679                 }
 680
 681                 $content_handler = ContentHandler::getForModelID( $content_model );
 682
 683                 if ( isset( $row->rev_content_format ) && !is_null( $row->rev_content_format ) ) {
 684                         $content_format = strval( $row->rev_content_format );
 685                 } else {
 686                         // probably using $wgContentHandlerUseDB = false;
 687                         $content_format = $content_handler->getDefaultFormat();
 688                 }
 689
 690                 $text = '';
 691                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_TEXT ) ) {
 692                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 693                 } elseif ( isset( $row->old_text ) ) {
 694                         // Raw text from the database may have invalid chars
 695                         $text = strval( Revision::getRevisionText( $row ) );
 696                         $text = $content_handler->exportTransform( $text, $content_format );
 697                         $out .= "      " . Xml::elementClean( 'text',
 698                                 array( 'xml:space' => 'preserve', 'bytes' => intval( $row->rev_len ) ),
 699                                 strval( $text ) ) . "\n";
 700                 } else {
 701                         // Stub output
 702                         $out .= "      " . Xml::element( 'text',
 703                                 array( 'id' => $row->rev_text_id, 'bytes' => intval( $row->rev_len ) ),
 704                                 "" ) . "\n";
 705                 }
 706
 707                 if ( isset( $row->rev_sha1 )
 708                         && $row->rev_sha1
 709                         && !( $row->rev_deleted & Revision::DELETED_TEXT )
 710                 ) {
 711                         $out .= "      " . Xml::element( 'sha1', null, strval( $row->rev_sha1 ) ) . "\n";
 712                 } else {
 713                         $out .= "      <sha1/>\n";
 714                 }
 715
 716                 $out .= "      " . Xml::element( 'model', null, strval( $content_model ) ) . "\n";
 717                 $out .= "      " . Xml::element( 'format', null, strval( $content_format ) ) . "\n";
 718
 719                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 720
 721                 $out .= "    </revision>\n";
 722
 723                 wfProfileOut( __METHOD__ );
 724                 return $out;
 725         }
 726
 727         /**
 728          * Dumps a "<logitem>" section on the output stream, with
 729          * data filled in from the given database row.
 730          *
 731          * @param object $row
 732          * @return string
 733          * @access private
 734          */
 735         function writeLogItem( $row ) {
 736                 wfProfileIn( __METHOD__ );
 737
 738                 $out = "  <logitem>\n";
 739                 $out .= "    " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 740
 741                 $out .= $this->writeTimestamp( $row->log_timestamp, "    " );
 742
 743                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 744                         $out .= "    " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 745                 } else {
 746                         $out .= $this->writeContributor( $row->log_user, $row->user_name, "    " );
 747                 }
 748
 749                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 750                         $out .= "    " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 751                 } elseif ( $row->log_comment != '' ) {
 752                         $out .= "    " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 753                 }
 754
 755                 $out .= "    " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 756                 $out .= "    " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 757
 758                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 759                         $out .= "    " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 760                 } else {
 761                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 762                         $out .= "    " . Xml::elementClean( 'logtitle', null, self::canonicalTitle( $title ) ) . "\n";
 763                         $out .= "    " . Xml::elementClean( 'params',
 764                                 array( 'xml:space' => 'preserve' ),
 765                                 strval( $row->log_params ) ) . "\n";
 766                 }
 767
 768                 $out .= "  </logitem>\n";
 769
 770                 wfProfileOut( __METHOD__ );
 771                 return $out;
 772         }
 773
 774         /**
 775          * @param string $timestamp
 776          * @param string $indent Default to six spaces
 777          * @return string
 778          */
 779         function writeTimestamp( $timestamp, $indent = "      " ) {
 780                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 781                 return $indent . Xml::element( 'timestamp', null, $ts ) . "\n";
 782         }
 783
 784         /**
 785          * @param int $id
 786          * @param string $text
 787          * @param string $indent Default to six spaces
 788          * @return string
 789          */
 790         function writeContributor( $id, $text, $indent = "      " ) {
 791                 $out = $indent . "<contributor>\n";
 792                 if ( $id || !IP::isValid( $text ) ) {
 793                         $out .= $indent . "  " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 794                         $out .= $indent . "  " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 795                 } else {
 796                         $out .= $indent . "  " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 797                 }
 798                 $out .= $indent . "</contributor>\n";
 799                 return $out;
 800         }
 801
 802         /**
 803          * Warning! This data is potentially inconsistent. :(
 804          * @param object $row
 805          * @param bool $dumpContents
 806          * @return string
 807          */
 808         function writeUploads( $row, $dumpContents = false ) {
 809                 if ( $row->page_namespace == NS_FILE ) {
 810                         $img = wfLocalFile( $row->page_title );
 811                         if ( $img && $img->exists() ) {
 812                                 $out = '';
 813                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 814                                         $out .= $this->writeUpload( $ver, $dumpContents );
 815                                 }
 816                                 $out .= $this->writeUpload( $img, $dumpContents );
 817                                 return $out;
 818                         }
 819                 }
 820                 return '';
 821         }
 822
 823         /**
 824          * @param File $file
 825          * @param bool $dumpContents
 826          * @return string
 827          */
 828         function writeUpload( $file, $dumpContents = false ) {
 829                 if ( $file->isOld() ) {
 830                         $archiveName = "      " .
 831                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 832                 } else {
 833                         $archiveName = '';
 834                 }
 835                 if ( $dumpContents ) {
 836                         $be = $file->getRepo()->getBackend();
 837                         # Dump file as base64
 838                         # Uses only XML-safe characters, so does not need escaping
 839                         # @todo Too bad this loads the contents into memory (script might swap)
 840                         $contents = '      <contents encoding="base64">' .
 841                                 chunk_split( base64_encode(
 842                                         $be->getFileContents( array( 'src' => $file->getPath() ) ) ) ) .
 843                                 "      </contents>\n";
 844                 } else {
 845                         $contents = '';
 846                 }
 847                 if ( $file->isDeleted( File::DELETED_COMMENT ) ) {
 848                         $comment = Xml::element( 'comment', array( 'deleted' => 'deleted' ) );
 849                 } else {
 850                         $comment = Xml::elementClean( 'comment', null, $file->getDescription() );
 851                 }
 852                 return "    <upload>\n" .
 853                         $this->writeTimestamp( $file->getTimestamp() ) .
 854                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 855                         "      " . $comment . "\n" .
 856                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 857                         $archiveName .
 858                         "      " . Xml::element( 'src', null, $file->getCanonicalURL() ) . "\n" .
 859                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 860                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 861                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 862                         $contents .
 863                         "    </upload>\n";
 864         }
 865
 866         /**
 867          * Return prefixed text form of title, but using the content language's
 868          * canonical namespace. This skips any special-casing such as gendered
 869          * user namespaces -- which while useful, are not yet listed in the
 870          * XML "<siteinfo>" data so are unsafe in export.
 871          *
 872          * @param Title $title
 873          * @return string
 874          * @since 1.18
 875          */
 876         public static function canonicalTitle( Title $title ) {
 877                 if ( $title->isExternal() ) {
 878                         return $title->getPrefixedText();
 879                 }
 880
 881                 global $wgContLang;
 882                 $prefix = str_replace( '_', ' ', $wgContLang->getNsText( $title->getNamespace() ) );
 883
 884                 if ( $prefix !== '' ) {
 885                         $prefix .= ':';
 886                 }
 887
 888                 return $prefix . $title->getText();
 889         }
 890 }
 891
 892 /**
 893  * Base class for output stream; prints to stdout or buffer or wherever.
 894  * @ingroup Dump
 895  */
 896 class DumpOutput {
 897
 898         /**
 899          * @param string $string
 900          */
 901         function writeOpenStream( $string ) {
 902                 $this->write( $string );
 903         }
 904
 905         /**
 906          * @param string $string
 907          */
 908         function writeCloseStream( $string ) {
 909                 $this->write( $string );
 910         }
 911
 912         /**
 913          * @param object $page
 914          * @param string $string
 915          */
 916         function writeOpenPage( $page, $string ) {
 917                 $this->write( $string );
 918         }
 919
 920         /**
 921          * @param string $string
 922          */
 923         function writeClosePage( $string ) {
 924                 $this->write( $string );
 925         }
 926
 927         /**
 928          * @param object $rev
 929          * @param string $string
 930          */
 931         function writeRevision( $rev, $string ) {
 932                 $this->write( $string );
 933         }
 934
 935         /**
 936          * @param object $rev
 937          * @param string $string
 938          */
 939         function writeLogItem( $rev, $string ) {
 940                 $this->write( $string );
 941         }
 942
 943         /**
 944          * Override to write to a different stream type.
 945          * @param string $string
 946          * @return bool
 947          */
 948         function write( $string ) {
 949                 print $string;
 950         }
 951
 952         /**
 953          * Close the old file, move it to a specified name,
 954          * and reopen new file with the old name. Use this
 955          * for writing out a file in multiple pieces
 956          * at specified checkpoints (e.g. every n hours).
 957          * @param string|array $newname File name. May be a string or an array with one element
 958          */
 959         function closeRenameAndReopen( $newname ) {
 960         }
 961
 962         /**
 963          * Close the old file, and move it to a specified name.
 964          * Use this for the last piece of a file written out
 965          * at specified checkpoints (e.g. every n hours).
 966          * @param string|array $newname File name. May be a string or an array with one element
 967          * @param bool $open If true, a new file with the old filename will be opened
 968          *   again for writing (default: false)
 969          */
 970         function closeAndRename( $newname, $open = false ) {
 971         }
 972
 973         /**
 974          * Returns the name of the file or files which are
 975          * being written to, if there are any.
 976          * @return null
 977          */
 978         function getFilenames() {
 979                 return null;
 980         }
 981 }
 982
 983 /**
 984  * Stream outputter to send data to a file.
 985  * @ingroup Dump
 986  */
 987 class DumpFileOutput extends DumpOutput {
 988         protected $handle = false, $filename;
 989
 990         /**
 991          * @param string $file
 992          */
 993         function __construct( $file ) {
 994                 $this->handle = fopen( $file, "wt" );
 995                 $this->filename = $file;
 996         }
 997
 998         /**
 999          * @param string $string
1000          */
1001         function writeCloseStream( $string ) {
1002                 parent::writeCloseStream( $string );
1003                 if ( $this->handle ) {
1004                         fclose( $this->handle );
1005                         $this->handle = false;
1006                 }
1007         }
1008
1009         /**
1010          * @param string $string
1011          */
1012         function write( $string ) {
1013                 fputs( $this->handle, $string );
1014         }
1015
1016         /**
1017          * @param string $newname
1018          */
1019         function closeRenameAndReopen( $newname ) {
1020                 $this->closeAndRename( $newname, true );
1021         }
1022
1023         /**
1024          * @param string $newname
1025          * @throws MWException
1026          */
1027         function renameOrException( $newname ) {
1028                         if ( !rename( $this->filename, $newname ) ) {
1029                                 throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" );
1030                         }
1031         }
1032
1033         /**
1034          * @param array $newname
1035          * @return string
1036          * @throws MWException
1037          */
1038         function checkRenameArgCount( $newname ) {
1039                 if ( is_array( $newname ) ) {
1040                         if ( count( $newname ) > 1 ) {
1041                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
1042                         } else {
1043                                 $newname = $newname[0];
1044                         }
1045                 }
1046                 return $newname;
1047         }
1048
1049         /**
1050          * @param string $newname
1051          * @param bool $open
1052          */
1053         function closeAndRename( $newname, $open = false ) {
1054                 $newname = $this->checkRenameArgCount( $newname );
1055                 if ( $newname ) {
1056                         if ( $this->handle ) {
1057                                 fclose( $this->handle );
1058                                 $this->handle = false;
1059                         }
1060                         $this->renameOrException( $newname );
1061                         if ( $open ) {
1062                                 $this->handle = fopen( $this->filename, "wt" );
1063                         }
1064                 }
1065         }
1066
1067         /**
1068          * @return string|null
1069          */
1070         function getFilenames() {
1071                 return $this->filename;
1072         }
1073 }
1074
1075 /**
1076  * Stream outputter to send data to a file via some filter program.
1077  * Even if compression is available in a library, using a separate
1078  * program can allow us to make use of a multi-processor system.
1079  * @ingroup Dump
1080  */
1081 class DumpPipeOutput extends DumpFileOutput {
1082         protected $command, $filename;
1083         protected $procOpenResource = false;
1084
1085         /**
1086          * @param string $command
1087          * @param string $file
1088          */
1089         function __construct( $command, $file = null ) {
1090                 if ( !is_null( $file ) ) {
1091                         $command .= " > " . wfEscapeShellArg( $file );
1092                 }
1093
1094                 $this->startCommand( $command );
1095                 $this->command = $command;
1096                 $this->filename = $file;
1097         }
1098
1099         /**
1100          * @param string $string
1101          */
1102         function writeCloseStream( $string ) {
1103                 parent::writeCloseStream( $string );
1104                 if ( $this->procOpenResource ) {
1105                         proc_close( $this->procOpenResource );
1106                         $this->procOpenResource = false;
1107                 }
1108         }
1109
1110         /**
1111          * @param string $command
1112          */
1113         function startCommand( $command ) {
1114                 $spec = array(
1115                         0 => array( "pipe", "r" ),
1116                 );
1117                 $pipes = array();
1118                 $this->procOpenResource = proc_open( $command, $spec, $pipes );
1119                 $this->handle = $pipes[0];
1120         }
1121
1122         /**
1123          * @param string $newname
1124          */
1125         function closeRenameAndReopen( $newname ) {
1126                 $this->closeAndRename( $newname, true );
1127         }
1128
1129         /**
1130          * @param string $newname
1131          * @param bool $open
1132          */
1133         function closeAndRename( $newname, $open = false ) {
1134                 $newname = $this->checkRenameArgCount( $newname );
1135                 if ( $newname ) {
1136                         if ( $this->handle ) {
1137                                 fclose( $this->handle );
1138                                 $this->handle = false;
1139                         }
1140                         if ( $this->procOpenResource ) {
1141                                 proc_close( $this->procOpenResource );
1142                                 $this->procOpenResource = false;
1143                         }
1144                         $this->renameOrException( $newname );
1145                         if ( $open ) {
1146                                 $command = $this->command;
1147                                 $command .= " > " . wfEscapeShellArg( $this->filename );
1148                                 $this->startCommand( $command );
1149                         }
1150                 }
1151         }
1152 }
1153
1154 /**
1155  * Sends dump output via the gzip compressor.
1156  * @ingroup Dump
1157  */
1158 class DumpGZipOutput extends DumpPipeOutput {
1159         /**
1160          * @param string $file
1161          */
1162         function __construct( $file ) {
1163                 parent::__construct( "gzip", $file );
1164         }
1165 }
1166
1167 /**
1168  * Sends dump output via the bgzip2 compressor.
1169  * @ingroup Dump
1170  */
1171 class DumpBZip2Output extends DumpPipeOutput {
1172         /**
1173          * @param string $file
1174          */
1175         function __construct( $file ) {
1176                 parent::__construct( "bzip2", $file );
1177         }
1178 }
1179
1180 /**
1181  * Sends dump output via the p7zip compressor.
1182  * @ingroup Dump
1183  */
1184 class Dump7ZipOutput extends DumpPipeOutput {
1185         /**
1186          * @param string $file
1187          */
1188         function __construct( $file ) {
1189                 $command = $this->setup7zCommand( $file );
1190                 parent::__construct( $command );
1191                 $this->filename = $file;
1192         }
1193
1194         /**
1195          * @param string $file
1196          * @return string
1197          */
1198         function setup7zCommand( $file ) {
1199                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
1200                 // Suppress annoying useless crap from p7zip
1201                 // Unfortunately this could suppress real error messages too
1202                 $command .= ' >' . wfGetNull() . ' 2>&1';
1203                 return $command;
1204         }
1205
1206         /**
1207          * @param string $newname
1208          * @param bool $open
1209          */
1210         function closeAndRename( $newname, $open = false ) {
1211                 $newname = $this->checkRenameArgCount( $newname );
1212                 if ( $newname ) {
1213                         fclose( $this->handle );
1214                         proc_close( $this->procOpenResource );
1215                         $this->renameOrException( $newname );
1216                         if ( $open ) {
1217                                 $command = $this->setup7zCommand( $this->filename );
1218                                 $this->startCommand( $command );
1219                         }
1220                 }
1221         }
1222 }
1223
1224 /**
1225  * Dump output filter class.
1226  * This just does output filtering and streaming; XML formatting is done
1227  * higher up, so be careful in what you do.
1228  * @ingroup Dump
1229  */
1230 class DumpFilter {
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          * @param object $page
1342          * @return bool
1343          */
1344         function pass( $page ) {
1345                 return !MWNamespace::isTalk( $page->page_namespace );
1346         }
1347 }
1348
1349 /**
1350  * Dump output filter to include or exclude pages in a given set of namespaces.
1351  * @ingroup Dump
1352  */
1353 class DumpNamespaceFilter extends DumpFilter {
1354         /** @var bool */
1355         protected $invert = false;
1356
1357         /** @var array */
1358         protected $namespaces = array();
1359
1360         /**
1361          * @param DumpOutput $sink
1362          * @param array $param
1363          * @throws MWException
1364          */
1365         function __construct( &$sink, $param ) {
1366                 parent::__construct( $sink );
1367
1368                 $constants = array(
1369                         "NS_MAIN"           => NS_MAIN,
1370                         "NS_TALK"           => NS_TALK,
1371                         "NS_USER"           => NS_USER,
1372                         "NS_USER_TALK"      => NS_USER_TALK,
1373                         "NS_PROJECT"        => NS_PROJECT,
1374                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
1375                         "NS_FILE"           => NS_FILE,
1376                         "NS_FILE_TALK"      => NS_FILE_TALK,
1377                         "NS_IMAGE"          => NS_IMAGE, // NS_IMAGE is an alias for NS_FILE
1378                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
1379                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
1380                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
1381                         "NS_TEMPLATE"       => NS_TEMPLATE,
1382                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
1383                         "NS_HELP"           => NS_HELP,
1384                         "NS_HELP_TALK"      => NS_HELP_TALK,
1385                         "NS_CATEGORY"       => NS_CATEGORY,
1386                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
1387
1388                 if ( $param { 0 } == '!' ) {
1389                         $this->invert = true;
1390                         $param = substr( $param, 1 );
1391                 }
1392
1393                 foreach ( explode( ',', $param ) as $key ) {
1394                         $key = trim( $key );
1395                         if ( isset( $constants[$key] ) ) {
1396                                 $ns = $constants[$key];
1397                                 $this->namespaces[$ns] = true;
1398                         } elseif ( is_numeric( $key ) ) {
1399                                 $ns = intval( $key );
1400                                 $this->namespaces[$ns] = true;
1401                         } else {
1402                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
1403                         }
1404                 }
1405         }
1406
1407         /**
1408          * @param object $page
1409          * @return bool
1410          */
1411         function pass( $page ) {
1412                 $match = isset( $this->namespaces[$page->page_namespace] );
1413                 return $this->invert xor $match;
1414         }
1415 }
1416
1417 /**
1418  * Dump output filter to include only the last revision in each page sequence.
1419  * @ingroup Dump
1420  */
1421 class DumpLatestFilter extends DumpFilter {
1422         protected $page;
1423
1424         protected $pageString;
1425
1426         protected $rev;
1427
1428         protected $revString;
1429
1430         /**
1431          * @param object $page
1432          * @param string $string
1433          */
1434         function writeOpenPage( $page, $string ) {
1435                 $this->page = $page;
1436                 $this->pageString = $string;
1437         }
1438
1439         /**
1440          * @param string $string
1441          */
1442         function writeClosePage( $string ) {
1443                 if ( $this->rev ) {
1444                         $this->sink->writeOpenPage( $this->page, $this->pageString );
1445                         $this->sink->writeRevision( $this->rev, $this->revString );
1446                         $this->sink->writeClosePage( $string );
1447                 }
1448                 $this->rev = null;
1449                 $this->revString = null;
1450                 $this->page = null;
1451                 $this->pageString = null;
1452         }
1453
1454         /**
1455          * @param object $rev
1456          * @param string $string
1457          */
1458         function writeRevision( $rev, $string ) {
1459                 if ( $rev->rev_id == $this->page->page_latest ) {
1460                         $this->rev = $rev;
1461                         $this->revString = $string;
1462                 }
1463         }
1464 }
1465
1466 /**
1467  * Base class for output stream; prints to stdout or buffer or wherever.
1468  * @ingroup Dump
1469  */
1470 class DumpMultiWriter {
1471
1472         /**
1473          * @param array $sinks
1474          */
1475         function __construct( $sinks ) {
1476                 $this->sinks = $sinks;
1477                 $this->count = count( $sinks );
1478         }
1479
1480         /**
1481          * @param string $string
1482          */
1483         function writeOpenStream( $string ) {
1484                 for ( $i = 0; $i < $this->count; $i++ ) {
1485                         $this->sinks[$i]->writeOpenStream( $string );
1486                 }
1487         }
1488
1489         /**
1490          * @param string $string
1491          */
1492         function writeCloseStream( $string ) {
1493                 for ( $i = 0; $i < $this->count; $i++ ) {
1494                         $this->sinks[$i]->writeCloseStream( $string );
1495                 }
1496         }
1497
1498         /**
1499          * @param object $page
1500          * @param string $string
1501          */
1502         function writeOpenPage( $page, $string ) {
1503                 for ( $i = 0; $i < $this->count; $i++ ) {
1504                         $this->sinks[$i]->writeOpenPage( $page, $string );
1505                 }
1506         }
1507
1508         /**
1509          * @param string $string
1510          */
1511         function writeClosePage( $string ) {
1512                 for ( $i = 0; $i < $this->count; $i++ ) {
1513                         $this->sinks[$i]->writeClosePage( $string );
1514                 }
1515         }
1516
1517         /**
1518          * @param object $rev
1519          * @param string $string
1520          */
1521         function writeRevision( $rev, $string ) {
1522                 for ( $i = 0; $i < $this->count; $i++ ) {
1523                         $this->sinks[$i]->writeRevision( $rev, $string );
1524                 }
1525         }
1526
1527         /**
1528          * @param array $newnames
1529          */
1530         function closeRenameAndReopen( $newnames ) {
1531                 $this->closeAndRename( $newnames, true );
1532         }
1533
1534         /**
1535          * @param array $newnames
1536          * @param bool $open
1537          */
1538         function closeAndRename( $newnames, $open = false ) {
1539                 for ( $i = 0; $i < $this->count; $i++ ) {
1540                         $this->sinks[$i]->closeAndRename( $newnames[$i], $open );
1541                 }
1542         }
1543
1544         /**
1545          * @return array
1546          */
1547         function getFilenames() {
1548                 $filenames = array();
1549                 for ( $i = 0; $i < $this->count; $i++ ) {
1550                         $filenames[] = $this->sinks[$i]->getFilenames();
1551                 }
1552                 return $filenames;
1553         }
1554 }
1555
1556 /**
1557  * @param string $string
1558  * @return string
1559  * @todo FIXME: Only used in OAI extension. Move over there.
1560  */
1561 function xmlsafe( $string ) {
1562         wfProfileIn( __FUNCTION__ );
1563
1564         /**
1565          * The page may contain old data which has not been properly normalized.
1566          * Invalid UTF-8 sequences or forbidden control characters will make our
1567          * XML output invalid, so be sure to strip them out.
1568          */
1569         $string = UtfNormal::cleanUp( $string );
1570
1571         $string = htmlspecialchars( $string );
1572         wfProfileOut( __FUNCTION__ );
1573         return $string;
1574 }