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         public $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         public $buffer;
  60
  61         /** @var int */
  62         public $text;
  63
  64         /** @var DumpOutput */
  65         public $sink;
  66
  67         /**
  68          * Returns the export schema version.
  69          * @return string
  70          */
  71         public static function schemaVersion() {
  72                 return "0.9";
  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                         $result = null; // Assuring $result 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                                 $this->outputLogStream( $result );
 279                                 if ( $this->buffer == WikiExporter::STREAM ) {
 280                                         $this->db->bufferResults( $prev );
 281                                 }
 282                         } catch ( Exception $e ) {
 283                                 // Throwing the exception does not reliably free the resultset, and
 284                                 // would also leave the connection in unbuffered mode.
 285
 286                                 // Freeing result
 287                                 try {
 288                                         if ( $result ) {
 289                                                 $result->free();
 290                                         }
 291                                 } catch ( Exception $e2 ) {
 292                                         // Already in panic mode -> ignoring $e2 as $e has
 293                                         // higher priority
 294                                 }
 295
 296                                 // Putting database back in previous buffer mode
 297                                 try {
 298                                         if ( $this->buffer == WikiExporter::STREAM ) {
 299                                                 $this->db->bufferResults( $prev );
 300                                         }
 301                                 } catch ( Exception $e2 ) {
 302                                         // Already in panic mode -> ignoring $e2 as $e has
 303                                         // higher priority
 304                                 }
 305
 306                                 // Inform caller about problem
 307                                 wfProfileOut( __METHOD__ );
 308                                 throw $e;
 309                         }
 310                 # For page dumps...
 311                 } else {
 312                         $tables = array( 'page', 'revision' );
 313                         $opts = array( 'ORDER BY' => 'page_id ASC' );
 314                         $opts['USE INDEX'] = array();
 315                         $join = array();
 316                         if ( is_array( $this->history ) ) {
 317                                 # Time offset/limit for all pages/history...
 318                                 $revJoin = 'page_id=rev_page';
 319                                 # Set time order
 320                                 if ( $this->history['dir'] == 'asc' ) {
 321                                         $op = '>';
 322                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 323                                 } else {
 324                                         $op = '<';
 325                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 326                                 }
 327                                 # Set offset
 328                                 if ( !empty( $this->history['offset'] ) ) {
 329                                         $revJoin .= " AND rev_timestamp $op " .
 330                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 331                                 }
 332                                 $join['revision'] = array( 'INNER JOIN', $revJoin );
 333                                 # Set query limit
 334                                 if ( !empty( $this->history['limit'] ) ) {
 335                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 336                                 }
 337                         } elseif ( $this->history & WikiExporter::FULL ) {
 338                                 # Full history dumps...
 339                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 340                         } elseif ( $this->history & WikiExporter::CURRENT ) {
 341                                 # Latest revision dumps...
 342                                 if ( $this->list_authors && $cond != '' ) { // List authors, if so desired
 343                                         $this->do_list_authors( $cond );
 344                                 }
 345                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 346                         } elseif ( $this->history & WikiExporter::STABLE ) {
 347                                 # "Stable" revision dumps...
 348                                 # Default JOIN, to be overridden...
 349                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 350                                 # One, and only one hook should set this, and return false
 351                                 if ( wfRunHooks( 'WikiExporter::dumpStableQuery', array( &$tables, &$opts, &$join ) ) ) {
 352                                         wfProfileOut( __METHOD__ );
 353                                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 354                                 }
 355                         } elseif ( $this->history & WikiExporter::RANGE ) {
 356                                 # Dump of revisions within a specified range
 357                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 358                                 $opts['ORDER BY'] = array( 'rev_page ASC', 'rev_id ASC' );
 359                         } else {
 360                                 # Unknown history specification parameter?
 361                                 wfProfileOut( __METHOD__ );
 362                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 363                         }
 364                         # Query optimization hacks
 365                         if ( $cond == '' ) {
 366                                 $opts[] = 'STRAIGHT_JOIN';
 367                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 368                         }
 369                         # Build text join options
 370                         if ( $this->text != WikiExporter::STUB ) { // 1-pass
 371                                 $tables[] = 'text';
 372                                 $join['text'] = array( 'INNER JOIN', 'rev_text_id=old_id' );
 373                         }
 374
 375                         if ( $this->buffer == WikiExporter::STREAM ) {
 376                                 $prev = $this->db->bufferResults( false );
 377                         }
 378
 379                         $result = null; // Assuring $result is not undefined, if exception occurs early
 380                         try {
 381                                 wfRunHooks( 'ModifyExportQuery',
 382                                                 array( $this->db, &$tables, &$cond, &$opts, &$join ) );
 383
 384                                 # Do the query!
 385                                 $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 386                                 # Output dump results
 387                                 $this->outputPageStream( $result );
 388
 389                                 if ( $this->buffer == WikiExporter::STREAM ) {
 390                                         $this->db->bufferResults( $prev );
 391                                 }
 392                         } catch ( Exception $e ) {
 393                                 // Throwing the exception does not reliably free the resultset, and
 394                                 // would also leave the connection in unbuffered mode.
 395
 396                                 // Freeing result
 397                                 try {
 398                                         if ( $result ) {
 399                                                 $result->free();
 400                                         }
 401                                 } catch ( Exception $e2 ) {
 402                                         // Already in panic mode -> ignoring $e2 as $e has
 403                                         // higher priority
 404                                 }
 405
 406                                 // Putting database back in previous buffer mode
 407                                 try {
 408                                         if ( $this->buffer == WikiExporter::STREAM ) {
 409                                                 $this->db->bufferResults( $prev );
 410                                         }
 411                                 } catch ( Exception $e2 ) {
 412                                         // Already in panic mode -> ignoring $e2 as $e has
 413                                         // higher priority
 414                                 }
 415
 416                                 // Inform caller about problem
 417                                 throw $e;
 418                         }
 419                 }
 420                 wfProfileOut( __METHOD__ );
 421         }
 422
 423         /**
 424          * Runs through a query result set dumping page and revision records.
 425          * The result set should be sorted/grouped by page to avoid duplicate
 426          * page records in the output.
 427          *
 428          * Should be safe for
 429          * streaming (non-buffered) queries, as long as it was made on a
 430          * separate database connection not managed by LoadBalancer; some
 431          * blob storage types will make queries to pull source data.
 432          *
 433          * @param ResultWrapper $resultset
 434          */
 435         protected function outputPageStream( $resultset ) {
 436                 $last = null;
 437                 foreach ( $resultset as $row ) {
 438                         if ( $last === null ||
 439                                 $last->page_namespace != $row->page_namespace ||
 440                                 $last->page_title != $row->page_title ) {
 441                                 if ( $last !== null ) {
 442                                         $output = '';
 443                                         if ( $this->dumpUploads ) {
 444                                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 445                                         }
 446                                         $output .= $this->writer->closePage();
 447                                         $this->sink->writeClosePage( $output );
 448                                 }
 449                                 $output = $this->writer->openPage( $row );
 450                                 $this->sink->writeOpenPage( $row, $output );
 451                                 $last = $row;
 452                         }
 453                         $output = $this->writer->writeRevision( $row );
 454                         $this->sink->writeRevision( $row, $output );
 455                 }
 456                 if ( $last !== null ) {
 457                         $output = '';
 458                         if ( $this->dumpUploads ) {
 459                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 460                         }
 461                         $output .= $this->author_list;
 462                         $output .= $this->writer->closePage();
 463                         $this->sink->writeClosePage( $output );
 464                 }
 465         }
 466
 467         /**
 468          * @param ResultWrapper $resultset
 469          */
 470         protected function outputLogStream( $resultset ) {
 471                 foreach ( $resultset as $row ) {
 472                         $output = $this->writer->writeLogItem( $row );
 473                         $this->sink->writeLogItem( $row, $output );
 474                 }
 475         }
 476 }
 477
 478 /**
 479  * @ingroup Dump
 480  */
 481 class XmlDumpWriter {
 482         /**
 483          * Returns the export schema version.
 484          * @deprecated since 1.20; use WikiExporter::schemaVersion() instead
 485          * @return string
 486          */
 487         function schemaVersion() {
 488                 wfDeprecated( __METHOD__, '1.20' );
 489                 return WikiExporter::schemaVersion();
 490         }
 491
 492         /**
 493          * Opens the XML output stream's root "<mediawiki>" element.
 494          * This does not include an xml directive, so is safe to include
 495          * as a subelement in a larger XML stream. Namespace and XML Schema
 496          * references are included.
 497          *
 498          * Output will be encoded in UTF-8.
 499          *
 500          * @return string
 501          */
 502         function openStream() {
 503                 global $wgLanguageCode;
 504                 $ver = WikiExporter::schemaVersion();
 505                 return Xml::element( 'mediawiki', array(
 506                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 507                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 508                         /*
 509                          * When a new version of the schema is created, it needs staging on mediawiki.org.
 510                          * This requires a change in the operations/mediawiki-config git repo.
 511                          *
 512                          * Create a changeset like https://gerrit.wikimedia.org/r/#/c/149643/ in which
 513                          * you copy in the new xsd file.
 514                          *
 515                          * After it is reviewed, merged and deployed (sync-docroot), the index.html needs purging.
 516                          * echo "http://www.mediawiki.org/xml/index.html" | mwscript purgeList.php --wiki=aawiki
 517                          */
 518                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 519                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 520                         'version'            => $ver,
 521                         'xml:lang'           => $wgLanguageCode ),
 522                         null ) .
 523                         "\n" .
 524                         $this->siteInfo();
 525         }
 526
 527         /**
 528          * @return string
 529          */
 530         function siteInfo() {
 531                 $info = array(
 532                         $this->sitename(),
 533                         $this->dbname(),
 534                         $this->homelink(),
 535                         $this->generator(),
 536                         $this->caseSetting(),
 537                         $this->namespaces() );
 538                 return "  <siteinfo>\n    " .
 539                         implode( "\n    ", $info ) .
 540                         "\n  </siteinfo>\n";
 541         }
 542
 543         /**
 544          * @return string
 545          */
 546         function sitename() {
 547                 global $wgSitename;
 548                 return Xml::element( 'sitename', array(), $wgSitename );
 549         }
 550
 551         /**
 552          * @return string
 553          */
 554         function dbname() {
 555                 global $wgDBname;
 556                 return Xml::element( 'dbname', array(), $wgDBname );
 557         }
 558
 559         /**
 560          * @return string
 561          */
 562         function generator() {
 563                 global $wgVersion;
 564                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 565         }
 566
 567         /**
 568          * @return string
 569          */
 570         function homelink() {
 571                 return Xml::element( 'base', array(), Title::newMainPage()->getCanonicalURL() );
 572         }
 573
 574         /**
 575          * @return string
 576          */
 577         function caseSetting() {
 578                 global $wgCapitalLinks;
 579                 // "case-insensitive" option is reserved for future
 580                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 581                 return Xml::element( 'case', array(), $sensitivity );
 582         }
 583
 584         /**
 585          * @return string
 586          */
 587         function namespaces() {
 588                 global $wgContLang;
 589                 $spaces = "<namespaces>\n";
 590                 foreach ( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 591                         $spaces .= '      ' .
 592                                 Xml::element( 'namespace',
 593                                         array(
 594                                                 'key' => $ns,
 595                                                 'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 596                                         ), $title ) . "\n";
 597                 }
 598                 $spaces .= "    </namespaces>";
 599                 return $spaces;
 600         }
 601
 602         /**
 603          * Closes the output stream with the closing root element.
 604          * Call when finished dumping things.
 605          *
 606          * @return string
 607          */
 608         function closeStream() {
 609                 return "</mediawiki>\n";
 610         }
 611
 612         /**
 613          * Opens a "<page>" section on the output stream, with data
 614          * from the given database row.
 615          *
 616          * @param object $row
 617          * @return string
 618          */
 619         public function openPage( $row ) {
 620                 $out = "  <page>\n";
 621                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 622                 $out .= '    ' . Xml::elementClean( 'title', array(), self::canonicalTitle( $title ) ) . "\n";
 623                 $out .= '    ' . Xml::element( 'ns', array(), strval( $row->page_namespace ) ) . "\n";
 624                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 625                 if ( $row->page_is_redirect ) {
 626                         $page = WikiPage::factory( $title );
 627                         $redirect = $page->getRedirectTarget();
 628                         if ( $redirect instanceof Title && $redirect->isValidRedirectTarget() ) {
 629                                 $out .= '    ';
 630                                 $out .= Xml::element( 'redirect', array( 'title' => self::canonicalTitle( $redirect ) ) );
 631                                 $out .= "\n";
 632                         }
 633                 }
 634
 635                 if ( $row->page_restrictions != '' ) {
 636                         $out .= '    ' . Xml::element( 'restrictions', array(),
 637                                 strval( $row->page_restrictions ) ) . "\n";
 638                 }
 639
 640                 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 641
 642                 return $out;
 643         }
 644
 645         /**
 646          * Closes a "<page>" section on the output stream.
 647          *
 648          * @access private
 649          * @return string
 650          */
 651         function closePage() {
 652                 return "  </page>\n";
 653         }
 654
 655         /**
 656          * Dumps a "<revision>" section on the output stream, with
 657          * data filled in from the given database row.
 658          *
 659          * @param object $row
 660          * @return string
 661          * @access private
 662          */
 663         function writeRevision( $row ) {
 664                 wfProfileIn( __METHOD__ );
 665
 666                 $out = "    <revision>\n";
 667                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 668                 if ( isset( $row->rev_parent_id ) && $row->rev_parent_id ) {
 669                         $out .= "      " . Xml::element( 'parentid', null, strval( $row->rev_parent_id ) ) . "\n";
 670                 }
 671
 672                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 673
 674                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_USER ) ) {
 675                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 676                 } else {
 677                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 678                 }
 679
 680                 if ( isset( $row->rev_minor_edit ) && $row->rev_minor_edit ) {
 681                         $out .= "      <minor/>\n";
 682                 }
 683                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_COMMENT ) ) {
 684                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 685                 } elseif ( $row->rev_comment != '' ) {
 686                         $out .= "      " . Xml::elementClean( 'comment', array(), strval( $row->rev_comment ) ) . "\n";
 687                 }
 688
 689                 if ( isset( $row->rev_content_model ) && !is_null( $row->rev_content_model ) ) {
 690                         $content_model = strval( $row->rev_content_model );
 691                 } else {
 692                         // probably using $wgContentHandlerUseDB = false;
 693                         $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 694                         $content_model = ContentHandler::getDefaultModelFor( $title );
 695                 }
 696
 697                 $content_handler = ContentHandler::getForModelID( $content_model );
 698
 699                 if ( isset( $row->rev_content_format ) && !is_null( $row->rev_content_format ) ) {
 700                         $content_format = strval( $row->rev_content_format );
 701                 } else {
 702                         // probably using $wgContentHandlerUseDB = false;
 703                         $content_format = $content_handler->getDefaultFormat();
 704                 }
 705
 706                 $text = '';
 707                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_TEXT ) ) {
 708                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 709                 } elseif ( isset( $row->old_text ) ) {
 710                         // Raw text from the database may have invalid chars
 711                         $text = strval( Revision::getRevisionText( $row ) );
 712                         $text = $content_handler->exportTransform( $text, $content_format );
 713                         $out .= "      " . Xml::elementClean( 'text',
 714                                 array( 'xml:space' => 'preserve', 'bytes' => intval( $row->rev_len ) ),
 715                                 strval( $text ) ) . "\n";
 716                 } else {
 717                         // Stub output
 718                         $out .= "      " . Xml::element( 'text',
 719                                 array( 'id' => $row->rev_text_id, 'bytes' => intval( $row->rev_len ) ),
 720                                 "" ) . "\n";
 721                 }
 722
 723                 if ( isset( $row->rev_sha1 )
 724                         && $row->rev_sha1
 725                         && !( $row->rev_deleted & Revision::DELETED_TEXT )
 726                 ) {
 727                         $out .= "      " . Xml::element( 'sha1', null, strval( $row->rev_sha1 ) ) . "\n";
 728                 } else {
 729                         $out .= "      <sha1/>\n";
 730                 }
 731
 732                 $out .= "      " . Xml::element( 'model', null, strval( $content_model ) ) . "\n";
 733                 $out .= "      " . Xml::element( 'format', null, strval( $content_format ) ) . "\n";
 734
 735                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 736
 737                 $out .= "    </revision>\n";
 738
 739                 wfProfileOut( __METHOD__ );
 740                 return $out;
 741         }
 742
 743         /**
 744          * Dumps a "<logitem>" section on the output stream, with
 745          * data filled in from the given database row.
 746          *
 747          * @param object $row
 748          * @return string
 749          * @access private
 750          */
 751         function writeLogItem( $row ) {
 752                 wfProfileIn( __METHOD__ );
 753
 754                 $out = "  <logitem>\n";
 755                 $out .= "    " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 756
 757                 $out .= $this->writeTimestamp( $row->log_timestamp, "    " );
 758
 759                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 760                         $out .= "    " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 761                 } else {
 762                         $out .= $this->writeContributor( $row->log_user, $row->user_name, "    " );
 763                 }
 764
 765                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 766                         $out .= "    " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 767                 } elseif ( $row->log_comment != '' ) {
 768                         $out .= "    " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 769                 }
 770
 771                 $out .= "    " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 772                 $out .= "    " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 773
 774                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 775                         $out .= "    " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 776                 } else {
 777                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 778                         $out .= "    " . Xml::elementClean( 'logtitle', null, self::canonicalTitle( $title ) ) . "\n";
 779                         $out .= "    " . Xml::elementClean( 'params',
 780                                 array( 'xml:space' => 'preserve' ),
 781                                 strval( $row->log_params ) ) . "\n";
 782                 }
 783
 784                 $out .= "  </logitem>\n";
 785
 786                 wfProfileOut( __METHOD__ );
 787                 return $out;
 788         }
 789
 790         /**
 791          * @param string $timestamp
 792          * @param string $indent Default to six spaces
 793          * @return string
 794          */
 795         function writeTimestamp( $timestamp, $indent = "      " ) {
 796                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 797                 return $indent . Xml::element( 'timestamp', null, $ts ) . "\n";
 798         }
 799
 800         /**
 801          * @param int $id
 802          * @param string $text
 803          * @param string $indent Default to six spaces
 804          * @return string
 805          */
 806         function writeContributor( $id, $text, $indent = "      " ) {
 807                 $out = $indent . "<contributor>\n";
 808                 if ( $id || !IP::isValid( $text ) ) {
 809                         $out .= $indent . "  " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 810                         $out .= $indent . "  " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 811                 } else {
 812                         $out .= $indent . "  " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 813                 }
 814                 $out .= $indent . "</contributor>\n";
 815                 return $out;
 816         }
 817
 818         /**
 819          * Warning! This data is potentially inconsistent. :(
 820          * @param object $row
 821          * @param bool $dumpContents
 822          * @return string
 823          */
 824         function writeUploads( $row, $dumpContents = false ) {
 825                 if ( $row->page_namespace == NS_FILE ) {
 826                         $img = wfLocalFile( $row->page_title );
 827                         if ( $img && $img->exists() ) {
 828                                 $out = '';
 829                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 830                                         $out .= $this->writeUpload( $ver, $dumpContents );
 831                                 }
 832                                 $out .= $this->writeUpload( $img, $dumpContents );
 833                                 return $out;
 834                         }
 835                 }
 836                 return '';
 837         }
 838
 839         /**
 840          * @param File $file
 841          * @param bool $dumpContents
 842          * @return string
 843          */
 844         function writeUpload( $file, $dumpContents = false ) {
 845                 if ( $file->isOld() ) {
 846                         $archiveName = "      " .
 847                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 848                 } else {
 849                         $archiveName = '';
 850                 }
 851                 if ( $dumpContents ) {
 852                         $be = $file->getRepo()->getBackend();
 853                         # Dump file as base64
 854                         # Uses only XML-safe characters, so does not need escaping
 855                         # @todo Too bad this loads the contents into memory (script might swap)
 856                         $contents = '      <contents encoding="base64">' .
 857                                 chunk_split( base64_encode(
 858                                         $be->getFileContents( array( 'src' => $file->getPath() ) ) ) ) .
 859                                 "      </contents>\n";
 860                 } else {
 861                         $contents = '';
 862                 }
 863                 if ( $file->isDeleted( File::DELETED_COMMENT ) ) {
 864                         $comment = Xml::element( 'comment', array( 'deleted' => 'deleted' ) );
 865                 } else {
 866                         $comment = Xml::elementClean( 'comment', null, $file->getDescription() );
 867                 }
 868                 return "    <upload>\n" .
 869                         $this->writeTimestamp( $file->getTimestamp() ) .
 870                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 871                         "      " . $comment . "\n" .
 872                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 873                         $archiveName .
 874                         "      " . Xml::element( 'src', null, $file->getCanonicalURL() ) . "\n" .
 875                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 876                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 877                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 878                         $contents .
 879                         "    </upload>\n";
 880         }
 881
 882         /**
 883          * Return prefixed text form of title, but using the content language's
 884          * canonical namespace. This skips any special-casing such as gendered
 885          * user namespaces -- which while useful, are not yet listed in the
 886          * XML "<siteinfo>" data so are unsafe in export.
 887          *
 888          * @param Title $title
 889          * @return string
 890          * @since 1.18
 891          */
 892         public static function canonicalTitle( Title $title ) {
 893                 if ( $title->isExternal() ) {
 894                         return $title->getPrefixedText();
 895                 }
 896
 897                 global $wgContLang;
 898                 $prefix = str_replace( '_', ' ', $wgContLang->getNsText( $title->getNamespace() ) );
 899
 900                 if ( $prefix !== '' ) {
 901                         $prefix .= ':';
 902                 }
 903
 904                 return $prefix . $title->getText();
 905         }
 906 }
 907
 908 /**
 909  * Base class for output stream; prints to stdout or buffer or wherever.
 910  * @ingroup Dump
 911  */
 912 class DumpOutput {
 913
 914         /**
 915          * @param string $string
 916          */
 917         function writeOpenStream( $string ) {
 918                 $this->write( $string );
 919         }
 920
 921         /**
 922          * @param string $string
 923          */
 924         function writeCloseStream( $string ) {
 925                 $this->write( $string );
 926         }
 927
 928         /**
 929          * @param object $page
 930          * @param string $string
 931          */
 932         function writeOpenPage( $page, $string ) {
 933                 $this->write( $string );
 934         }
 935
 936         /**
 937          * @param string $string
 938          */
 939         function writeClosePage( $string ) {
 940                 $this->write( $string );
 941         }
 942
 943         /**
 944          * @param object $rev
 945          * @param string $string
 946          */
 947         function writeRevision( $rev, $string ) {
 948                 $this->write( $string );
 949         }
 950
 951         /**
 952          * @param object $rev
 953          * @param string $string
 954          */
 955         function writeLogItem( $rev, $string ) {
 956                 $this->write( $string );
 957         }
 958
 959         /**
 960          * Override to write to a different stream type.
 961          * @param string $string
 962          * @return bool
 963          */
 964         function write( $string ) {
 965                 print $string;
 966         }
 967
 968         /**
 969          * Close the old file, move it to a specified name,
 970          * and reopen new file with the old name. Use this
 971          * for writing out a file in multiple pieces
 972          * at specified checkpoints (e.g. every n hours).
 973          * @param string|array $newname File name. May be a string or an array with one element
 974          */
 975         function closeRenameAndReopen( $newname ) {
 976         }
 977
 978         /**
 979          * Close the old file, and move it to a specified name.
 980          * Use this for the last piece of a file written out
 981          * at specified checkpoints (e.g. every n hours).
 982          * @param string|array $newname File name. May be a string or an array with one element
 983          * @param bool $open If true, a new file with the old filename will be opened
 984          *   again for writing (default: false)
 985          */
 986         function closeAndRename( $newname, $open = false ) {
 987         }
 988
 989         /**
 990          * Returns the name of the file or files which are
 991          * being written to, if there are any.
 992          * @return null
 993          */
 994         function getFilenames() {
 995                 return null;
 996         }
 997 }
 998
 999 /**
1000  * Stream outputter to send data to a file.
1001  * @ingroup Dump
1002  */
1003 class DumpFileOutput extends DumpOutput {
1004         protected $handle = false, $filename;
1005
1006         /**
1007          * @param string $file
1008          */
1009         function __construct( $file ) {
1010                 $this->handle = fopen( $file, "wt" );
1011                 $this->filename = $file;
1012         }
1013
1014         /**
1015          * @param string $string
1016          */
1017         function writeCloseStream( $string ) {
1018                 parent::writeCloseStream( $string );
1019                 if ( $this->handle ) {
1020                         fclose( $this->handle );
1021                         $this->handle = false;
1022                 }
1023         }
1024
1025         /**
1026          * @param string $string
1027          */
1028         function write( $string ) {
1029                 fputs( $this->handle, $string );
1030         }
1031
1032         /**
1033          * @param string $newname
1034          */
1035         function closeRenameAndReopen( $newname ) {
1036                 $this->closeAndRename( $newname, true );
1037         }
1038
1039         /**
1040          * @param string $newname
1041          * @throws MWException
1042          */
1043         function renameOrException( $newname ) {
1044                         if ( !rename( $this->filename, $newname ) ) {
1045                                 throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" );
1046                         }
1047         }
1048
1049         /**
1050          * @param array $newname
1051          * @return string
1052          * @throws MWException
1053          */
1054         function checkRenameArgCount( $newname ) {
1055                 if ( is_array( $newname ) ) {
1056                         if ( count( $newname ) > 1 ) {
1057                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
1058                         } else {
1059                                 $newname = $newname[0];
1060                         }
1061                 }
1062                 return $newname;
1063         }
1064
1065         /**
1066          * @param string $newname
1067          * @param bool $open
1068          */
1069         function closeAndRename( $newname, $open = false ) {
1070                 $newname = $this->checkRenameArgCount( $newname );
1071                 if ( $newname ) {
1072                         if ( $this->handle ) {
1073                                 fclose( $this->handle );
1074                                 $this->handle = false;
1075                         }
1076                         $this->renameOrException( $newname );
1077                         if ( $open ) {
1078                                 $this->handle = fopen( $this->filename, "wt" );
1079                         }
1080                 }
1081         }
1082
1083         /**
1084          * @return string|null
1085          */
1086         function getFilenames() {
1087                 return $this->filename;
1088         }
1089 }
1090
1091 /**
1092  * Stream outputter to send data to a file via some filter program.
1093  * Even if compression is available in a library, using a separate
1094  * program can allow us to make use of a multi-processor system.
1095  * @ingroup Dump
1096  */
1097 class DumpPipeOutput extends DumpFileOutput {
1098         protected $command, $filename;
1099         protected $procOpenResource = false;
1100
1101         /**
1102          * @param string $command
1103          * @param string $file
1104          */
1105         function __construct( $command, $file = null ) {
1106                 if ( !is_null( $file ) ) {
1107                         $command .= " > " . wfEscapeShellArg( $file );
1108                 }
1109
1110                 $this->startCommand( $command );
1111                 $this->command = $command;
1112                 $this->filename = $file;
1113         }
1114
1115         /**
1116          * @param string $string
1117          */
1118         function writeCloseStream( $string ) {
1119                 parent::writeCloseStream( $string );
1120                 if ( $this->procOpenResource ) {
1121                         proc_close( $this->procOpenResource );
1122                         $this->procOpenResource = false;
1123                 }
1124         }
1125
1126         /**
1127          * @param string $command
1128          */
1129         function startCommand( $command ) {
1130                 $spec = array(
1131                         0 => array( "pipe", "r" ),
1132                 );
1133                 $pipes = array();
1134                 $this->procOpenResource = proc_open( $command, $spec, $pipes );
1135                 $this->handle = $pipes[0];
1136         }
1137
1138         /**
1139          * @param string $newname
1140          */
1141         function closeRenameAndReopen( $newname ) {
1142                 $this->closeAndRename( $newname, true );
1143         }
1144
1145         /**
1146          * @param string $newname
1147          * @param bool $open
1148          */
1149         function closeAndRename( $newname, $open = false ) {
1150                 $newname = $this->checkRenameArgCount( $newname );
1151                 if ( $newname ) {
1152                         if ( $this->handle ) {
1153                                 fclose( $this->handle );
1154                                 $this->handle = false;
1155                         }
1156                         if ( $this->procOpenResource ) {
1157                                 proc_close( $this->procOpenResource );
1158                                 $this->procOpenResource = false;
1159                         }
1160                         $this->renameOrException( $newname );
1161                         if ( $open ) {
1162                                 $command = $this->command;
1163                                 $command .= " > " . wfEscapeShellArg( $this->filename );
1164                                 $this->startCommand( $command );
1165                         }
1166                 }
1167         }
1168 }
1169
1170 /**
1171  * Sends dump output via the gzip compressor.
1172  * @ingroup Dump
1173  */
1174 class DumpGZipOutput extends DumpPipeOutput {
1175         /**
1176          * @param string $file
1177          */
1178         function __construct( $file ) {
1179                 parent::__construct( "gzip", $file );
1180         }
1181 }
1182
1183 /**
1184  * Sends dump output via the bgzip2 compressor.
1185  * @ingroup Dump
1186  */
1187 class DumpBZip2Output extends DumpPipeOutput {
1188         /**
1189          * @param string $file
1190          */
1191         function __construct( $file ) {
1192                 parent::__construct( "bzip2", $file );
1193         }
1194 }
1195
1196 /**
1197  * Sends dump output via the p7zip compressor.
1198  * @ingroup Dump
1199  */
1200 class Dump7ZipOutput extends DumpPipeOutput {
1201         /**
1202          * @param string $file
1203          */
1204         function __construct( $file ) {
1205                 $command = $this->setup7zCommand( $file );
1206                 parent::__construct( $command );
1207                 $this->filename = $file;
1208         }
1209
1210         /**
1211          * @param string $file
1212          * @return string
1213          */
1214         function setup7zCommand( $file ) {
1215                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
1216                 // Suppress annoying useless crap from p7zip
1217                 // Unfortunately this could suppress real error messages too
1218                 $command .= ' >' . wfGetNull() . ' 2>&1';
1219                 return $command;
1220         }
1221
1222         /**
1223          * @param string $newname
1224          * @param bool $open
1225          */
1226         function closeAndRename( $newname, $open = false ) {
1227                 $newname = $this->checkRenameArgCount( $newname );
1228                 if ( $newname ) {
1229                         fclose( $this->handle );
1230                         proc_close( $this->procOpenResource );
1231                         $this->renameOrException( $newname );
1232                         if ( $open ) {
1233                                 $command = $this->setup7zCommand( $this->filename );
1234                                 $this->startCommand( $command );
1235                         }
1236                 }
1237         }
1238 }
1239
1240 /**
1241  * Dump output filter class.
1242  * This just does output filtering and streaming; XML formatting is done
1243  * higher up, so be careful in what you do.
1244  * @ingroup Dump
1245  */
1246 class DumpFilter {
1247         /**
1248          * @var DumpOutput
1249          * FIXME will need to be made protected whenever legacy code
1250          * is updated.
1251          */
1252         public $sink;
1253
1254         /**
1255          * @var bool
1256          */
1257         protected $sendingThisPage;
1258
1259         /**
1260          * @param DumpOutput $sink
1261          */
1262         function __construct( &$sink ) {
1263                 $this->sink =& $sink;
1264         }
1265
1266         /**
1267          * @param string $string
1268          */
1269         function writeOpenStream( $string ) {
1270                 $this->sink->writeOpenStream( $string );
1271         }
1272
1273         /**
1274          * @param string $string
1275          */
1276         function writeCloseStream( $string ) {
1277                 $this->sink->writeCloseStream( $string );
1278         }
1279
1280         /**
1281          * @param object $page
1282          * @param string $string
1283          */
1284         function writeOpenPage( $page, $string ) {
1285                 $this->sendingThisPage = $this->pass( $page, $string );
1286                 if ( $this->sendingThisPage ) {
1287                         $this->sink->writeOpenPage( $page, $string );
1288                 }
1289         }
1290
1291         /**
1292          * @param string $string
1293          */
1294         function writeClosePage( $string ) {
1295                 if ( $this->sendingThisPage ) {
1296                         $this->sink->writeClosePage( $string );
1297                         $this->sendingThisPage = false;
1298                 }
1299         }
1300
1301         /**
1302          * @param object $rev
1303          * @param string $string
1304          */
1305         function writeRevision( $rev, $string ) {
1306                 if ( $this->sendingThisPage ) {
1307                         $this->sink->writeRevision( $rev, $string );
1308                 }
1309         }
1310
1311         /**
1312          * @param object $rev
1313          * @param string $string
1314          */
1315         function writeLogItem( $rev, $string ) {
1316                 $this->sink->writeRevision( $rev, $string );
1317         }
1318
1319         /**
1320          * @param string $newname
1321          */
1322         function closeRenameAndReopen( $newname ) {
1323                 $this->sink->closeRenameAndReopen( $newname );
1324         }
1325
1326         /**
1327          * @param string $newname
1328          * @param bool $open
1329          */
1330         function closeAndRename( $newname, $open = false ) {
1331                 $this->sink->closeAndRename( $newname, $open );
1332         }
1333
1334         /**
1335          * @return array
1336          */
1337         function getFilenames() {
1338                 return $this->sink->getFilenames();
1339         }
1340
1341         /**
1342          * Override for page-based filter types.
1343          * @param object $page
1344          * @return bool
1345          */
1346         function pass( $page ) {
1347                 return true;
1348         }
1349 }
1350
1351 /**
1352  * Simple dump output filter to exclude all talk pages.
1353  * @ingroup Dump
1354  */
1355 class DumpNotalkFilter extends DumpFilter {
1356         /**
1357          * @param object $page
1358          * @return bool
1359          */
1360         function pass( $page ) {
1361                 return !MWNamespace::isTalk( $page->page_namespace );
1362         }
1363 }
1364
1365 /**
1366  * Dump output filter to include or exclude pages in a given set of namespaces.
1367  * @ingroup Dump
1368  */
1369 class DumpNamespaceFilter extends DumpFilter {
1370         /** @var bool */
1371         public $invert = false;
1372
1373         /** @var array */
1374         public $namespaces = array();
1375
1376         /**
1377          * @param DumpOutput $sink
1378          * @param array $param
1379          * @throws MWException
1380          */
1381         function __construct( &$sink, $param ) {
1382                 parent::__construct( $sink );
1383
1384                 $constants = array(
1385                         "NS_MAIN"           => NS_MAIN,
1386                         "NS_TALK"           => NS_TALK,
1387                         "NS_USER"           => NS_USER,
1388                         "NS_USER_TALK"      => NS_USER_TALK,
1389                         "NS_PROJECT"        => NS_PROJECT,
1390                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
1391                         "NS_FILE"           => NS_FILE,
1392                         "NS_FILE_TALK"      => NS_FILE_TALK,
1393                         "NS_IMAGE"          => NS_IMAGE, // NS_IMAGE is an alias for NS_FILE
1394                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
1395                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
1396                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
1397                         "NS_TEMPLATE"       => NS_TEMPLATE,
1398                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
1399                         "NS_HELP"           => NS_HELP,
1400                         "NS_HELP_TALK"      => NS_HELP_TALK,
1401                         "NS_CATEGORY"       => NS_CATEGORY,
1402                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
1403
1404                 if ( $param { 0 } == '!' ) {
1405                         $this->invert = true;
1406                         $param = substr( $param, 1 );
1407                 }
1408
1409                 foreach ( explode( ',', $param ) as $key ) {
1410                         $key = trim( $key );
1411                         if ( isset( $constants[$key] ) ) {
1412                                 $ns = $constants[$key];
1413                                 $this->namespaces[$ns] = true;
1414                         } elseif ( is_numeric( $key ) ) {
1415                                 $ns = intval( $key );
1416                                 $this->namespaces[$ns] = true;
1417                         } else {
1418                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
1419                         }
1420                 }
1421         }
1422
1423         /**
1424          * @param object $page
1425          * @return bool
1426          */
1427         function pass( $page ) {
1428                 $match = isset( $this->namespaces[$page->page_namespace] );
1429                 return $this->invert xor $match;
1430         }
1431 }
1432
1433 /**
1434  * Dump output filter to include only the last revision in each page sequence.
1435  * @ingroup Dump
1436  */
1437 class DumpLatestFilter extends DumpFilter {
1438         public $page;
1439
1440         public $pageString;
1441
1442         public $rev;
1443
1444         public $revString;
1445
1446         /**
1447          * @param object $page
1448          * @param string $string
1449          */
1450         function writeOpenPage( $page, $string ) {
1451                 $this->page = $page;
1452                 $this->pageString = $string;
1453         }
1454
1455         /**
1456          * @param string $string
1457          */
1458         function writeClosePage( $string ) {
1459                 if ( $this->rev ) {
1460                         $this->sink->writeOpenPage( $this->page, $this->pageString );
1461                         $this->sink->writeRevision( $this->rev, $this->revString );
1462                         $this->sink->writeClosePage( $string );
1463                 }
1464                 $this->rev = null;
1465                 $this->revString = null;
1466                 $this->page = null;
1467                 $this->pageString = null;
1468         }
1469
1470         /**
1471          * @param object $rev
1472          * @param string $string
1473          */
1474         function writeRevision( $rev, $string ) {
1475                 if ( $rev->rev_id == $this->page->page_latest ) {
1476                         $this->rev = $rev;
1477                         $this->revString = $string;
1478                 }
1479         }
1480 }
1481
1482 /**
1483  * Base class for output stream; prints to stdout or buffer or wherever.
1484  * @ingroup Dump
1485  */
1486 class DumpMultiWriter {
1487
1488         /**
1489          * @param array $sinks
1490          */
1491         function __construct( $sinks ) {
1492                 $this->sinks = $sinks;
1493                 $this->count = count( $sinks );
1494         }
1495
1496         /**
1497          * @param string $string
1498          */
1499         function writeOpenStream( $string ) {
1500                 for ( $i = 0; $i < $this->count; $i++ ) {
1501                         $this->sinks[$i]->writeOpenStream( $string );
1502                 }
1503         }
1504
1505         /**
1506          * @param string $string
1507          */
1508         function writeCloseStream( $string ) {
1509                 for ( $i = 0; $i < $this->count; $i++ ) {
1510                         $this->sinks[$i]->writeCloseStream( $string );
1511                 }
1512         }
1513
1514         /**
1515          * @param object $page
1516          * @param string $string
1517          */
1518         function writeOpenPage( $page, $string ) {
1519                 for ( $i = 0; $i < $this->count; $i++ ) {
1520                         $this->sinks[$i]->writeOpenPage( $page, $string );
1521                 }
1522         }
1523
1524         /**
1525          * @param string $string
1526          */
1527         function writeClosePage( $string ) {
1528                 for ( $i = 0; $i < $this->count; $i++ ) {
1529                         $this->sinks[$i]->writeClosePage( $string );
1530                 }
1531         }
1532
1533         /**
1534          * @param object $rev
1535          * @param string $string
1536          */
1537         function writeRevision( $rev, $string ) {
1538                 for ( $i = 0; $i < $this->count; $i++ ) {
1539                         $this->sinks[$i]->writeRevision( $rev, $string );
1540                 }
1541         }
1542
1543         /**
1544          * @param array $newnames
1545          */
1546         function closeRenameAndReopen( $newnames ) {
1547                 $this->closeAndRename( $newnames, true );
1548         }
1549
1550         /**
1551          * @param array $newnames
1552          * @param bool $open
1553          */
1554         function closeAndRename( $newnames, $open = false ) {
1555                 for ( $i = 0; $i < $this->count; $i++ ) {
1556                         $this->sinks[$i]->closeAndRename( $newnames[$i], $open );
1557                 }
1558         }
1559
1560         /**
1561          * @return array
1562          */
1563         function getFilenames() {
1564                 $filenames = array();
1565                 for ( $i = 0; $i < $this->count; $i++ ) {
1566                         $filenames[] = $this->sinks[$i]->getFilenames();
1567                 }
1568                 return $filenames;
1569         }
1570 }