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 "0.10";
  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                 $this->author_list = "<contributors>";
 217                 // rev_deleted
 218
 219                 $res = $this->db->select(
 220                         array( 'page', 'revision' ),
 221                         array( 'DISTINCT rev_user_text', 'rev_user' ),
 222                         array(
 223                                 $this->db->bitAnd( 'rev_deleted', Revision::DELETED_USER ) . ' = 0',
 224                                 $cond,
 225                                 'page_id = rev_id',
 226                         ),
 227                         __METHOD__
 228                 );
 229
 230                 foreach ( $res as $row ) {
 231                         $this->author_list .= "<contributor>" .
 232                                 "<username>" .
 233                                 htmlentities( $row->rev_user_text ) .
 234                                 "</username>" .
 235                                 "<id>" .
 236                                 $row->rev_user .
 237                                 "</id>" .
 238                                 "</contributor>";
 239                 }
 240                 $this->author_list .= "</contributors>";
 241         }
 242
 243         /**
 244          * @param string $cond
 245          * @throws MWException
 246          * @throws Exception
 247          */
 248         protected function dumpFrom( $cond = '' ) {
 249                 # For logging dumps...
 250                 if ( $this->history & self::LOGS ) {
 251                         $where = array( 'user_id = log_user' );
 252                         # Hide private logs
 253                         $hideLogs = LogEventsList::getExcludeClause( $this->db );
 254                         if ( $hideLogs ) {
 255                                 $where[] = $hideLogs;
 256                         }
 257                         # Add on any caller specified conditions
 258                         if ( $cond ) {
 259                                 $where[] = $cond;
 260                         }
 261                         # Get logging table name for logging.* clause
 262                         $logging = $this->db->tableName( 'logging' );
 263
 264                         if ( $this->buffer == WikiExporter::STREAM ) {
 265                                 $prev = $this->db->bufferResults( false );
 266                         }
 267                         $result = null; // Assuring $result is not undefined, if exception occurs early
 268                         try {
 269                                 $result = $this->db->select( array( 'logging', 'user' ),
 270                                         array( "{$logging}.*", 'user_name' ), // grab the user name
 271                                         $where,
 272                                         __METHOD__,
 273                                         array( 'ORDER BY' => 'log_id', 'USE INDEX' => array( 'logging' => 'PRIMARY' ) )
 274                                 );
 275                                 $this->outputLogStream( $result );
 276                                 if ( $this->buffer == WikiExporter::STREAM ) {
 277                                         $this->db->bufferResults( $prev );
 278                                 }
 279                         } catch ( Exception $e ) {
 280                                 // Throwing the exception does not reliably free the resultset, and
 281                                 // would also leave the connection in unbuffered mode.
 282
 283                                 // Freeing result
 284                                 try {
 285                                         if ( $result ) {
 286                                                 $result->free();
 287                                         }
 288                                 } catch ( Exception $e2 ) {
 289                                         // Already in panic mode -> ignoring $e2 as $e has
 290                                         // higher priority
 291                                 }
 292
 293                                 // Putting database back in previous buffer mode
 294                                 try {
 295                                         if ( $this->buffer == WikiExporter::STREAM ) {
 296                                                 $this->db->bufferResults( $prev );
 297                                         }
 298                                 } catch ( Exception $e2 ) {
 299                                         // Already in panic mode -> ignoring $e2 as $e has
 300                                         // higher priority
 301                                 }
 302
 303                                 // Inform caller about problem
 304                                 throw $e;
 305                         }
 306                 # For page dumps...
 307                 } else {
 308                         $tables = array( 'page', 'revision' );
 309                         $opts = array( 'ORDER BY' => 'page_id ASC' );
 310                         $opts['USE INDEX'] = array();
 311                         $join = array();
 312                         if ( is_array( $this->history ) ) {
 313                                 # Time offset/limit for all pages/history...
 314                                 $revJoin = 'page_id=rev_page';
 315                                 # Set time order
 316                                 if ( $this->history['dir'] == 'asc' ) {
 317                                         $op = '>';
 318                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 319                                 } else {
 320                                         $op = '<';
 321                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 322                                 }
 323                                 # Set offset
 324                                 if ( !empty( $this->history['offset'] ) ) {
 325                                         $revJoin .= " AND rev_timestamp $op " .
 326                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 327                                 }
 328                                 $join['revision'] = array( 'INNER JOIN', $revJoin );
 329                                 # Set query limit
 330                                 if ( !empty( $this->history['limit'] ) ) {
 331                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 332                                 }
 333                         } elseif ( $this->history & WikiExporter::FULL ) {
 334                                 # Full history dumps...
 335                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 336                         } elseif ( $this->history & WikiExporter::CURRENT ) {
 337                                 # Latest revision dumps...
 338                                 if ( $this->list_authors && $cond != '' ) { // List authors, if so desired
 339                                         $this->do_list_authors( $cond );
 340                                 }
 341                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 342                         } elseif ( $this->history & WikiExporter::STABLE ) {
 343                                 # "Stable" revision dumps...
 344                                 # Default JOIN, to be overridden...
 345                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 346                                 # One, and only one hook should set this, and return false
 347                                 if ( Hooks::run( 'WikiExporter::dumpStableQuery', array( &$tables, &$opts, &$join ) ) ) {
 348                                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 349                                 }
 350                         } elseif ( $this->history & WikiExporter::RANGE ) {
 351                                 # Dump of revisions within a specified range
 352                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 353                                 $opts['ORDER BY'] = array( 'rev_page ASC', 'rev_id ASC' );
 354                         } else {
 355                                 # Unknown history specification parameter?
 356                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 357                         }
 358                         # Query optimization hacks
 359                         if ( $cond == '' ) {
 360                                 $opts[] = 'STRAIGHT_JOIN';
 361                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 362                         }
 363                         # Build text join options
 364                         if ( $this->text != WikiExporter::STUB ) { // 1-pass
 365                                 $tables[] = 'text';
 366                                 $join['text'] = array( 'INNER JOIN', 'rev_text_id=old_id' );
 367                         }
 368
 369                         if ( $this->buffer == WikiExporter::STREAM ) {
 370                                 $prev = $this->db->bufferResults( false );
 371                         }
 372
 373                         $result = null; // Assuring $result is not undefined, if exception occurs early
 374                         try {
 375                                 Hooks::run( 'ModifyExportQuery',
 376                                                 array( $this->db, &$tables, &$cond, &$opts, &$join ) );
 377
 378                                 # Do the query!
 379                                 $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 380                                 # Output dump results
 381                                 $this->outputPageStream( $result );
 382
 383                                 if ( $this->buffer == WikiExporter::STREAM ) {
 384                                         $this->db->bufferResults( $prev );
 385                                 }
 386                         } catch ( Exception $e ) {
 387                                 // Throwing the exception does not reliably free the resultset, and
 388                                 // would also leave the connection in unbuffered mode.
 389
 390                                 // Freeing result
 391                                 try {
 392                                         if ( $result ) {
 393                                                 $result->free();
 394                                         }
 395                                 } catch ( Exception $e2 ) {
 396                                         // Already in panic mode -> ignoring $e2 as $e has
 397                                         // higher priority
 398                                 }
 399
 400                                 // Putting database back in previous buffer mode
 401                                 try {
 402                                         if ( $this->buffer == WikiExporter::STREAM ) {
 403                                                 $this->db->bufferResults( $prev );
 404                                         }
 405                                 } catch ( Exception $e2 ) {
 406                                         // Already in panic mode -> ignoring $e2 as $e has
 407                                         // higher priority
 408                                 }
 409
 410                                 // Inform caller about problem
 411                                 throw $e;
 412                         }
 413                 }
 414         }
 415
 416         /**
 417          * Runs through a query result set dumping page and revision records.
 418          * The result set should be sorted/grouped by page to avoid duplicate
 419          * page records in the output.
 420          *
 421          * Should be safe for
 422          * streaming (non-buffered) queries, as long as it was made on a
 423          * separate database connection not managed by LoadBalancer; some
 424          * blob storage types will make queries to pull source data.
 425          *
 426          * @param ResultWrapper $resultset
 427          */
 428         protected function outputPageStream( $resultset ) {
 429                 $last = null;
 430                 foreach ( $resultset as $row ) {
 431                         if ( $last === null ||
 432                                 $last->page_namespace != $row->page_namespace ||
 433                                 $last->page_title != $row->page_title ) {
 434                                 if ( $last !== null ) {
 435                                         $output = '';
 436                                         if ( $this->dumpUploads ) {
 437                                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 438                                         }
 439                                         $output .= $this->writer->closePage();
 440                                         $this->sink->writeClosePage( $output );
 441                                 }
 442                                 $output = $this->writer->openPage( $row );
 443                                 $this->sink->writeOpenPage( $row, $output );
 444                                 $last = $row;
 445                         }
 446                         $output = $this->writer->writeRevision( $row );
 447                         $this->sink->writeRevision( $row, $output );
 448                 }
 449                 if ( $last !== null ) {
 450                         $output = '';
 451                         if ( $this->dumpUploads ) {
 452                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 453                         }
 454                         $output .= $this->author_list;
 455                         $output .= $this->writer->closePage();
 456                         $this->sink->writeClosePage( $output );
 457                 }
 458         }
 459
 460         /**
 461          * @param ResultWrapper $resultset
 462          */
 463         protected function outputLogStream( $resultset ) {
 464                 foreach ( $resultset as $row ) {
 465                         $output = $this->writer->writeLogItem( $row );
 466                         $this->sink->writeLogItem( $row, $output );
 467                 }
 468         }
 469 }
 470
 471 /**
 472  * @ingroup Dump
 473  */
 474 class XmlDumpWriter {
 475         /**
 476          * Opens the XML output stream's root "<mediawiki>" element.
 477          * This does not include an xml directive, so is safe to include
 478          * as a subelement in a larger XML stream. Namespace and XML Schema
 479          * references are included.
 480          *
 481          * Output will be encoded in UTF-8.
 482          *
 483          * @return string
 484          */
 485         function openStream() {
 486                 global $wgLanguageCode;
 487                 $ver = WikiExporter::schemaVersion();
 488                 return Xml::element( 'mediawiki', array(
 489                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 490                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 491                         /*
 492                          * When a new version of the schema is created, it needs staging on mediawiki.org.
 493                          * This requires a change in the operations/mediawiki-config git repo.
 494                          *
 495                          * Create a changeset like https://gerrit.wikimedia.org/r/#/c/149643/ in which
 496                          * you copy in the new xsd file.
 497                          *
 498                          * After it is reviewed, merged and deployed (sync-docroot), the index.html needs purging.
 499                          * echo "http://www.mediawiki.org/xml/index.html" | mwscript purgeList.php --wiki=aawiki
 500                          */
 501                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 502                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 503                         'version'            => $ver,
 504                         'xml:lang'           => $wgLanguageCode ),
 505                         null ) .
 506                         "\n" .
 507                         $this->siteInfo();
 508         }
 509
 510         /**
 511          * @return string
 512          */
 513         function siteInfo() {
 514                 $info = array(
 515                         $this->sitename(),
 516                         $this->dbname(),
 517                         $this->homelink(),
 518                         $this->generator(),
 519                         $this->caseSetting(),
 520                         $this->namespaces() );
 521                 return "  <siteinfo>\n    " .
 522                         implode( "\n    ", $info ) .
 523                         "\n  </siteinfo>\n";
 524         }
 525
 526         /**
 527          * @return string
 528          */
 529         function sitename() {
 530                 global $wgSitename;
 531                 return Xml::element( 'sitename', array(), $wgSitename );
 532         }
 533
 534         /**
 535          * @return string
 536          */
 537         function dbname() {
 538                 global $wgDBname;
 539                 return Xml::element( 'dbname', array(), $wgDBname );
 540         }
 541
 542         /**
 543          * @return string
 544          */
 545         function generator() {
 546                 global $wgVersion;
 547                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 548         }
 549
 550         /**
 551          * @return string
 552          */
 553         function homelink() {
 554                 return Xml::element( 'base', array(), Title::newMainPage()->getCanonicalURL() );
 555         }
 556
 557         /**
 558          * @return string
 559          */
 560         function caseSetting() {
 561                 global $wgCapitalLinks;
 562                 // "case-insensitive" option is reserved for future
 563                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 564                 return Xml::element( 'case', array(), $sensitivity );
 565         }
 566
 567         /**
 568          * @return string
 569          */
 570         function namespaces() {
 571                 global $wgContLang;
 572                 $spaces = "<namespaces>\n";
 573                 foreach ( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 574                         $spaces .= '      ' .
 575                                 Xml::element( 'namespace',
 576                                         array(
 577                                                 'key' => $ns,
 578                                                 'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 579                                         ), $title ) . "\n";
 580                 }
 581                 $spaces .= "    </namespaces>";
 582                 return $spaces;
 583         }
 584
 585         /**
 586          * Closes the output stream with the closing root element.
 587          * Call when finished dumping things.
 588          *
 589          * @return string
 590          */
 591         function closeStream() {
 592                 return "</mediawiki>\n";
 593         }
 594
 595         /**
 596          * Opens a "<page>" section on the output stream, with data
 597          * from the given database row.
 598          *
 599          * @param object $row
 600          * @return string
 601          */
 602         public function openPage( $row ) {
 603                 $out = "  <page>\n";
 604                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 605                 $out .= '    ' . Xml::elementClean( 'title', array(), self::canonicalTitle( $title ) ) . "\n";
 606                 $out .= '    ' . Xml::element( 'ns', array(), strval( $row->page_namespace ) ) . "\n";
 607                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 608                 if ( $row->page_is_redirect ) {
 609                         $page = WikiPage::factory( $title );
 610                         $redirect = $page->getRedirectTarget();
 611                         if ( $redirect instanceof Title && $redirect->isValidRedirectTarget() ) {
 612                                 $out .= '    ';
 613                                 $out .= Xml::element( 'redirect', array( 'title' => self::canonicalTitle( $redirect ) ) );
 614                                 $out .= "\n";
 615                         }
 616                 }
 617
 618                 if ( $row->page_restrictions != '' ) {
 619                         $out .= '    ' . Xml::element( 'restrictions', array(),
 620                                 strval( $row->page_restrictions ) ) . "\n";
 621                 }
 622
 623                 Hooks::run( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 624
 625                 return $out;
 626         }
 627
 628         /**
 629          * Closes a "<page>" section on the output stream.
 630          *
 631          * @access private
 632          * @return string
 633          */
 634         function closePage() {
 635                 return "  </page>\n";
 636         }
 637
 638         /**
 639          * Dumps a "<revision>" section on the output stream, with
 640          * data filled in from the given database row.
 641          *
 642          * @param object $row
 643          * @return string
 644          * @access private
 645          */
 646         function writeRevision( $row ) {
 647
 648                 $out = "    <revision>\n";
 649                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 650                 if ( isset( $row->rev_parent_id ) && $row->rev_parent_id ) {
 651                         $out .= "      " . Xml::element( 'parentid', null, strval( $row->rev_parent_id ) ) . "\n";
 652                 }
 653
 654                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 655
 656                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_USER ) ) {
 657                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 658                 } else {
 659                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 660                 }
 661
 662                 if ( isset( $row->rev_minor_edit ) && $row->rev_minor_edit ) {
 663                         $out .= "      <minor/>\n";
 664                 }
 665                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_COMMENT ) ) {
 666                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 667                 } elseif ( $row->rev_comment != '' ) {
 668                         $out .= "      " . Xml::elementClean( 'comment', array(), strval( $row->rev_comment ) ) . "\n";
 669                 }
 670
 671                 if ( isset( $row->rev_content_model ) && !is_null( $row->rev_content_model ) ) {
 672                         $content_model = strval( $row->rev_content_model );
 673                 } else {
 674                         // probably using $wgContentHandlerUseDB = false;
 675                         $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 676                         $content_model = ContentHandler::getDefaultModelFor( $title );
 677                 }
 678
 679                 $content_handler = ContentHandler::getForModelID( $content_model );
 680
 681                 if ( isset( $row->rev_content_format ) && !is_null( $row->rev_content_format ) ) {
 682                         $content_format = strval( $row->rev_content_format );
 683                 } else {
 684                         // probably using $wgContentHandlerUseDB = false;
 685                         $content_format = $content_handler->getDefaultFormat();
 686                 }
 687
 688                 $out .= "      " . Xml::element( 'model', null, strval( $content_model ) ) . "\n";
 689                 $out .= "      " . Xml::element( 'format', null, strval( $content_format ) ) . "\n";
 690
 691                 $text = '';
 692                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_TEXT ) ) {
 693                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 694                 } elseif ( isset( $row->old_text ) ) {
 695                         // Raw text from the database may have invalid chars
 696                         $text = strval( Revision::getRevisionText( $row ) );
 697                         $text = $content_handler->exportTransform( $text, $content_format );
 698                         $out .= "      " . Xml::elementClean( 'text',
 699                                 array( 'xml:space' => 'preserve', 'bytes' => intval( $row->rev_len ) ),
 700                                 strval( $text ) ) . "\n";
 701                 } else {
 702                         // Stub output
 703                         $out .= "      " . Xml::element( 'text',
 704                                 array( 'id' => $row->rev_text_id, 'bytes' => intval( $row->rev_len ) ),
 705                                 "" ) . "\n";
 706                 }
 707
 708                 if ( isset( $row->rev_sha1 )
 709                         && $row->rev_sha1
 710                         && !( $row->rev_deleted & Revision::DELETED_TEXT )
 711                 ) {
 712                         $out .= "      " . Xml::element( 'sha1', null, strval( $row->rev_sha1 ) ) . "\n";
 713                 } else {
 714                         $out .= "      <sha1/>\n";
 715                 }
 716
 717                 Hooks::run( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 718
 719                 $out .= "    </revision>\n";
 720
 721                 return $out;
 722         }
 723
 724         /**
 725          * Dumps a "<logitem>" section on the output stream, with
 726          * data filled in from the given database row.
 727          *
 728          * @param object $row
 729          * @return string
 730          * @access private
 731          */
 732         function writeLogItem( $row ) {
 733
 734                 $out = "  <logitem>\n";
 735                 $out .= "    " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 736
 737                 $out .= $this->writeTimestamp( $row->log_timestamp, "    " );
 738
 739                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 740                         $out .= "    " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 741                 } else {
 742                         $out .= $this->writeContributor( $row->log_user, $row->user_name, "    " );
 743                 }
 744
 745                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 746                         $out .= "    " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 747                 } elseif ( $row->log_comment != '' ) {
 748                         $out .= "    " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 749                 }
 750
 751                 $out .= "    " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 752                 $out .= "    " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 753
 754                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 755                         $out .= "    " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 756                 } else {
 757                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 758                         $out .= "    " . Xml::elementClean( 'logtitle', null, self::canonicalTitle( $title ) ) . "\n";
 759                         $out .= "    " . Xml::elementClean( 'params',
 760                                 array( 'xml:space' => 'preserve' ),
 761                                 strval( $row->log_params ) ) . "\n";
 762                 }
 763
 764                 $out .= "  </logitem>\n";
 765
 766                 return $out;
 767         }
 768
 769         /**
 770          * @param string $timestamp
 771          * @param string $indent Default to six spaces
 772          * @return string
 773          */
 774         function writeTimestamp( $timestamp, $indent = "      " ) {
 775                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 776                 return $indent . Xml::element( 'timestamp', null, $ts ) . "\n";
 777         }
 778
 779         /**
 780          * @param int $id
 781          * @param string $text
 782          * @param string $indent Default to six spaces
 783          * @return string
 784          */
 785         function writeContributor( $id, $text, $indent = "      " ) {
 786                 $out = $indent . "<contributor>\n";
 787                 if ( $id || !IP::isValid( $text ) ) {
 788                         $out .= $indent . "  " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 789                         $out .= $indent . "  " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 790                 } else {
 791                         $out .= $indent . "  " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 792                 }
 793                 $out .= $indent . "</contributor>\n";
 794                 return $out;
 795         }
 796
 797         /**
 798          * Warning! This data is potentially inconsistent. :(
 799          * @param object $row
 800          * @param bool $dumpContents
 801          * @return string
 802          */
 803         function writeUploads( $row, $dumpContents = false ) {
 804                 if ( $row->page_namespace == NS_FILE ) {
 805                         $img = wfLocalFile( $row->page_title );
 806                         if ( $img && $img->exists() ) {
 807                                 $out = '';
 808                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 809                                         $out .= $this->writeUpload( $ver, $dumpContents );
 810                                 }
 811                                 $out .= $this->writeUpload( $img, $dumpContents );
 812                                 return $out;
 813                         }
 814                 }
 815                 return '';
 816         }
 817
 818         /**
 819          * @param File $file
 820          * @param bool $dumpContents
 821          * @return string
 822          */
 823         function writeUpload( $file, $dumpContents = false ) {
 824                 if ( $file->isOld() ) {
 825                         $archiveName = "      " .
 826                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 827                 } else {
 828                         $archiveName = '';
 829                 }
 830                 if ( $dumpContents ) {
 831                         $be = $file->getRepo()->getBackend();
 832                         # Dump file as base64
 833                         # Uses only XML-safe characters, so does not need escaping
 834                         # @todo Too bad this loads the contents into memory (script might swap)
 835                         $contents = '      <contents encoding="base64">' .
 836                                 chunk_split( base64_encode(
 837                                         $be->getFileContents( array( 'src' => $file->getPath() ) ) ) ) .
 838                                 "      </contents>\n";
 839                 } else {
 840                         $contents = '';
 841                 }
 842                 if ( $file->isDeleted( File::DELETED_COMMENT ) ) {
 843                         $comment = Xml::element( 'comment', array( 'deleted' => 'deleted' ) );
 844                 } else {
 845                         $comment = Xml::elementClean( 'comment', null, $file->getDescription() );
 846                 }
 847                 return "    <upload>\n" .
 848                         $this->writeTimestamp( $file->getTimestamp() ) .
 849                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 850                         "      " . $comment . "\n" .
 851                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 852                         $archiveName .
 853                         "      " . Xml::element( 'src', null, $file->getCanonicalURL() ) . "\n" .
 854                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 855                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 856                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 857                         $contents .
 858                         "    </upload>\n";
 859         }
 860
 861         /**
 862          * Return prefixed text form of title, but using the content language's
 863          * canonical namespace. This skips any special-casing such as gendered
 864          * user namespaces -- which while useful, are not yet listed in the
 865          * XML "<siteinfo>" data so are unsafe in export.
 866          *
 867          * @param Title $title
 868          * @return string
 869          * @since 1.18
 870          */
 871         public static function canonicalTitle( Title $title ) {
 872                 if ( $title->isExternal() ) {
 873                         return $title->getPrefixedText();
 874                 }
 875
 876                 global $wgContLang;
 877                 $prefix = str_replace( '_', ' ', $wgContLang->getNsText( $title->getNamespace() ) );
 878
 879                 if ( $prefix !== '' ) {
 880                         $prefix .= ':';
 881                 }
 882
 883                 return $prefix . $title->getText();
 884         }
 885 }
 886
 887 /**
 888  * Base class for output stream; prints to stdout or buffer or wherever.
 889  * @ingroup Dump
 890  */
 891 class DumpOutput {
 892
 893         /**
 894          * @param string $string
 895          */
 896         function writeOpenStream( $string ) {
 897                 $this->write( $string );
 898         }
 899
 900         /**
 901          * @param string $string
 902          */
 903         function writeCloseStream( $string ) {
 904                 $this->write( $string );
 905         }
 906
 907         /**
 908          * @param object $page
 909          * @param string $string
 910          */
 911         function writeOpenPage( $page, $string ) {
 912                 $this->write( $string );
 913         }
 914
 915         /**
 916          * @param string $string
 917          */
 918         function writeClosePage( $string ) {
 919                 $this->write( $string );
 920         }
 921
 922         /**
 923          * @param object $rev
 924          * @param string $string
 925          */
 926         function writeRevision( $rev, $string ) {
 927                 $this->write( $string );
 928         }
 929
 930         /**
 931          * @param object $rev
 932          * @param string $string
 933          */
 934         function writeLogItem( $rev, $string ) {
 935                 $this->write( $string );
 936         }
 937
 938         /**
 939          * Override to write to a different stream type.
 940          * @param string $string
 941          * @return bool
 942          */
 943         function write( $string ) {
 944                 print $string;
 945         }
 946
 947         /**
 948          * Close the old file, move it to a specified name,
 949          * and reopen new file with the old name. Use this
 950          * for writing out a file in multiple pieces
 951          * at specified checkpoints (e.g. every n hours).
 952          * @param string|array $newname File name. May be a string or an array with one element
 953          */
 954         function closeRenameAndReopen( $newname ) {
 955         }
 956
 957         /**
 958          * Close the old file, and move it to a specified name.
 959          * Use this for the last piece of a file written out
 960          * at specified checkpoints (e.g. every n hours).
 961          * @param string|array $newname File name. May be a string or an array with one element
 962          * @param bool $open If true, a new file with the old filename will be opened
 963          *   again for writing (default: false)
 964          */
 965         function closeAndRename( $newname, $open = false ) {
 966         }
 967
 968         /**
 969          * Returns the name of the file or files which are
 970          * being written to, if there are any.
 971          * @return null
 972          */
 973         function getFilenames() {
 974                 return null;
 975         }
 976 }
 977
 978 /**
 979  * Stream outputter to send data to a file.
 980  * @ingroup Dump
 981  */
 982 class DumpFileOutput extends DumpOutput {
 983         protected $handle = false, $filename;
 984
 985         /**
 986          * @param string $file
 987          */
 988         function __construct( $file ) {
 989                 $this->handle = fopen( $file, "wt" );
 990                 $this->filename = $file;
 991         }
 992
 993         /**
 994          * @param string $string
 995          */
 996         function writeCloseStream( $string ) {
 997                 parent::writeCloseStream( $string );
 998                 if ( $this->handle ) {
 999                         fclose( $this->handle );
1000                         $this->handle = false;
1001                 }
1002         }
1003
1004         /**
1005          * @param string $string
1006          */
1007         function write( $string ) {
1008                 fputs( $this->handle, $string );
1009         }
1010
1011         /**
1012          * @param string $newname
1013          */
1014         function closeRenameAndReopen( $newname ) {
1015                 $this->closeAndRename( $newname, true );
1016         }
1017
1018         /**
1019          * @param string $newname
1020          * @throws MWException
1021          */
1022         function renameOrException( $newname ) {
1023                         if ( !rename( $this->filename, $newname ) ) {
1024                                 throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" );
1025                         }
1026         }
1027
1028         /**
1029          * @param array $newname
1030          * @return string
1031          * @throws MWException
1032          */
1033         function checkRenameArgCount( $newname ) {
1034                 if ( is_array( $newname ) ) {
1035                         if ( count( $newname ) > 1 ) {
1036                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
1037                         } else {
1038                                 $newname = $newname[0];
1039                         }
1040                 }
1041                 return $newname;
1042         }
1043
1044         /**
1045          * @param string $newname
1046          * @param bool $open
1047          */
1048         function closeAndRename( $newname, $open = false ) {
1049                 $newname = $this->checkRenameArgCount( $newname );
1050                 if ( $newname ) {
1051                         if ( $this->handle ) {
1052                                 fclose( $this->handle );
1053                                 $this->handle = false;
1054                         }
1055                         $this->renameOrException( $newname );
1056                         if ( $open ) {
1057                                 $this->handle = fopen( $this->filename, "wt" );
1058                         }
1059                 }
1060         }
1061
1062         /**
1063          * @return string|null
1064          */
1065         function getFilenames() {
1066                 return $this->filename;
1067         }
1068 }
1069
1070 /**
1071  * Stream outputter to send data to a file via some filter program.
1072  * Even if compression is available in a library, using a separate
1073  * program can allow us to make use of a multi-processor system.
1074  * @ingroup Dump
1075  */
1076 class DumpPipeOutput extends DumpFileOutput {
1077         protected $command, $filename;
1078         protected $procOpenResource = false;
1079
1080         /**
1081          * @param string $command
1082          * @param string $file
1083          */
1084         function __construct( $command, $file = null ) {
1085                 if ( !is_null( $file ) ) {
1086                         $command .= " > " . wfEscapeShellArg( $file );
1087                 }
1088
1089                 $this->startCommand( $command );
1090                 $this->command = $command;
1091                 $this->filename = $file;
1092         }
1093
1094         /**
1095          * @param string $string
1096          */
1097         function writeCloseStream( $string ) {
1098                 parent::writeCloseStream( $string );
1099                 if ( $this->procOpenResource ) {
1100                         proc_close( $this->procOpenResource );
1101                         $this->procOpenResource = false;
1102                 }
1103         }
1104
1105         /**
1106          * @param string $command
1107          */
1108         function startCommand( $command ) {
1109                 $spec = array(
1110                         0 => array( "pipe", "r" ),
1111                 );
1112                 $pipes = array();
1113                 $this->procOpenResource = proc_open( $command, $spec, $pipes );
1114                 $this->handle = $pipes[0];
1115         }
1116
1117         /**
1118          * @param string $newname
1119          */
1120         function closeRenameAndReopen( $newname ) {
1121                 $this->closeAndRename( $newname, true );
1122         }
1123
1124         /**
1125          * @param string $newname
1126          * @param bool $open
1127          */
1128         function closeAndRename( $newname, $open = false ) {
1129                 $newname = $this->checkRenameArgCount( $newname );
1130                 if ( $newname ) {
1131                         if ( $this->handle ) {
1132                                 fclose( $this->handle );
1133                                 $this->handle = false;
1134                         }
1135                         if ( $this->procOpenResource ) {
1136                                 proc_close( $this->procOpenResource );
1137                                 $this->procOpenResource = false;
1138                         }
1139                         $this->renameOrException( $newname );
1140                         if ( $open ) {
1141                                 $command = $this->command;
1142                                 $command .= " > " . wfEscapeShellArg( $this->filename );
1143                                 $this->startCommand( $command );
1144                         }
1145                 }
1146         }
1147 }
1148
1149 /**
1150  * Sends dump output via the gzip compressor.
1151  * @ingroup Dump
1152  */
1153 class DumpGZipOutput extends DumpPipeOutput {
1154         /**
1155          * @param string $file
1156          */
1157         function __construct( $file ) {
1158                 parent::__construct( "gzip", $file );
1159         }
1160 }
1161
1162 /**
1163  * Sends dump output via the bgzip2 compressor.
1164  * @ingroup Dump
1165  */
1166 class DumpBZip2Output extends DumpPipeOutput {
1167         /**
1168          * @param string $file
1169          */
1170         function __construct( $file ) {
1171                 parent::__construct( "bzip2", $file );
1172         }
1173 }
1174
1175 /**
1176  * Sends dump output via the p7zip compressor.
1177  * @ingroup Dump
1178  */
1179 class Dump7ZipOutput extends DumpPipeOutput {
1180         /**
1181          * @param string $file
1182          */
1183         function __construct( $file ) {
1184                 $command = $this->setup7zCommand( $file );
1185                 parent::__construct( $command );
1186                 $this->filename = $file;
1187         }
1188
1189         /**
1190          * @param string $file
1191          * @return string
1192          */
1193         function setup7zCommand( $file ) {
1194                 $command = "7za a -bd -si -mx=4 " . wfEscapeShellArg( $file );
1195                 // Suppress annoying useless crap from p7zip
1196                 // Unfortunately this could suppress real error messages too
1197                 $command .= ' >' . wfGetNull() . ' 2>&1';
1198                 return $command;
1199         }
1200
1201         /**
1202          * @param string $newname
1203          * @param bool $open
1204          */
1205         function closeAndRename( $newname, $open = false ) {
1206                 $newname = $this->checkRenameArgCount( $newname );
1207                 if ( $newname ) {
1208                         fclose( $this->handle );
1209                         proc_close( $this->procOpenResource );
1210                         $this->renameOrException( $newname );
1211                         if ( $open ) {
1212                                 $command = $this->setup7zCommand( $this->filename );
1213                                 $this->startCommand( $command );
1214                         }
1215                 }
1216         }
1217 }
1218
1219 /**
1220  * Dump output filter class.
1221  * This just does output filtering and streaming; XML formatting is done
1222  * higher up, so be careful in what you do.
1223  * @ingroup Dump
1224  */
1225 class DumpFilter {
1226         /**
1227          * @var DumpOutput
1228          * FIXME will need to be made protected whenever legacy code
1229          * is updated.
1230          */
1231         public $sink;
1232
1233         /**
1234          * @var bool
1235          */
1236         protected $sendingThisPage;
1237
1238         /**
1239          * @param DumpOutput $sink
1240          */
1241         function __construct( &$sink ) {
1242                 $this->sink =& $sink;
1243         }
1244
1245         /**
1246          * @param string $string
1247          */
1248         function writeOpenStream( $string ) {
1249                 $this->sink->writeOpenStream( $string );
1250         }
1251
1252         /**
1253          * @param string $string
1254          */
1255         function writeCloseStream( $string ) {
1256                 $this->sink->writeCloseStream( $string );
1257         }
1258
1259         /**
1260          * @param object $page
1261          * @param string $string
1262          */
1263         function writeOpenPage( $page, $string ) {
1264                 $this->sendingThisPage = $this->pass( $page, $string );
1265                 if ( $this->sendingThisPage ) {
1266                         $this->sink->writeOpenPage( $page, $string );
1267                 }
1268         }
1269
1270         /**
1271          * @param string $string
1272          */
1273         function writeClosePage( $string ) {
1274                 if ( $this->sendingThisPage ) {
1275                         $this->sink->writeClosePage( $string );
1276                         $this->sendingThisPage = false;
1277                 }
1278         }
1279
1280         /**
1281          * @param object $rev
1282          * @param string $string
1283          */
1284         function writeRevision( $rev, $string ) {
1285                 if ( $this->sendingThisPage ) {
1286                         $this->sink->writeRevision( $rev, $string );
1287                 }
1288         }
1289
1290         /**
1291          * @param object $rev
1292          * @param string $string
1293          */
1294         function writeLogItem( $rev, $string ) {
1295                 $this->sink->writeRevision( $rev, $string );
1296         }
1297
1298         /**
1299          * @param string $newname
1300          */
1301         function closeRenameAndReopen( $newname ) {
1302                 $this->sink->closeRenameAndReopen( $newname );
1303         }
1304
1305         /**
1306          * @param string $newname
1307          * @param bool $open
1308          */
1309         function closeAndRename( $newname, $open = false ) {
1310                 $this->sink->closeAndRename( $newname, $open );
1311         }
1312
1313         /**
1314          * @return array
1315          */
1316         function getFilenames() {
1317                 return $this->sink->getFilenames();
1318         }
1319
1320         /**
1321          * Override for page-based filter types.
1322          * @param object $page
1323          * @return bool
1324          */
1325         function pass( $page ) {
1326                 return true;
1327         }
1328 }
1329
1330 /**
1331  * Simple dump output filter to exclude all talk pages.
1332  * @ingroup Dump
1333  */
1334 class DumpNotalkFilter extends DumpFilter {
1335         /**
1336          * @param object $page
1337          * @return bool
1338          */
1339         function pass( $page ) {
1340                 return !MWNamespace::isTalk( $page->page_namespace );
1341         }
1342 }
1343
1344 /**
1345  * Dump output filter to include or exclude pages in a given set of namespaces.
1346  * @ingroup Dump
1347  */
1348 class DumpNamespaceFilter extends DumpFilter {
1349         /** @var bool */
1350         public $invert = false;
1351
1352         /** @var array */
1353         public $namespaces = array();
1354
1355         /**
1356          * @param DumpOutput $sink
1357          * @param array $param
1358          * @throws MWException
1359          */
1360         function __construct( &$sink, $param ) {
1361                 parent::__construct( $sink );
1362
1363                 $constants = array(
1364                         "NS_MAIN"           => NS_MAIN,
1365                         "NS_TALK"           => NS_TALK,
1366                         "NS_USER"           => NS_USER,
1367                         "NS_USER_TALK"      => NS_USER_TALK,
1368                         "NS_PROJECT"        => NS_PROJECT,
1369                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
1370                         "NS_FILE"           => NS_FILE,
1371                         "NS_FILE_TALK"      => NS_FILE_TALK,
1372                         "NS_IMAGE"          => NS_IMAGE, // NS_IMAGE is an alias for NS_FILE
1373                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
1374                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
1375                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
1376                         "NS_TEMPLATE"       => NS_TEMPLATE,
1377                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
1378                         "NS_HELP"           => NS_HELP,
1379                         "NS_HELP_TALK"      => NS_HELP_TALK,
1380                         "NS_CATEGORY"       => NS_CATEGORY,
1381                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
1382
1383                 if ( $param { 0 } == '!' ) {
1384                         $this->invert = true;
1385                         $param = substr( $param, 1 );
1386                 }
1387
1388                 foreach ( explode( ',', $param ) as $key ) {
1389                         $key = trim( $key );
1390                         if ( isset( $constants[$key] ) ) {
1391                                 $ns = $constants[$key];
1392                                 $this->namespaces[$ns] = true;
1393                         } elseif ( is_numeric( $key ) ) {
1394                                 $ns = intval( $key );
1395                                 $this->namespaces[$ns] = true;
1396                         } else {
1397                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
1398                         }
1399                 }
1400         }
1401
1402         /**
1403          * @param object $page
1404          * @return bool
1405          */
1406         function pass( $page ) {
1407                 $match = isset( $this->namespaces[$page->page_namespace] );
1408                 return $this->invert xor $match;
1409         }
1410 }
1411
1412 /**
1413  * Dump output filter to include only the last revision in each page sequence.
1414  * @ingroup Dump
1415  */
1416 class DumpLatestFilter extends DumpFilter {
1417         public $page;
1418
1419         public $pageString;
1420
1421         public $rev;
1422
1423         public $revString;
1424
1425         /**
1426          * @param object $page
1427          * @param string $string
1428          */
1429         function writeOpenPage( $page, $string ) {
1430                 $this->page = $page;
1431                 $this->pageString = $string;
1432         }
1433
1434         /**
1435          * @param string $string
1436          */
1437         function writeClosePage( $string ) {
1438                 if ( $this->rev ) {
1439                         $this->sink->writeOpenPage( $this->page, $this->pageString );
1440                         $this->sink->writeRevision( $this->rev, $this->revString );
1441                         $this->sink->writeClosePage( $string );
1442                 }
1443                 $this->rev = null;
1444                 $this->revString = null;
1445                 $this->page = null;
1446                 $this->pageString = null;
1447         }
1448
1449         /**
1450          * @param object $rev
1451          * @param string $string
1452          */
1453         function writeRevision( $rev, $string ) {
1454                 if ( $rev->rev_id == $this->page->page_latest ) {
1455                         $this->rev = $rev;
1456                         $this->revString = $string;
1457                 }
1458         }
1459 }
1460
1461 /**
1462  * Base class for output stream; prints to stdout or buffer or wherever.
1463  * @ingroup Dump
1464  */
1465 class DumpMultiWriter {
1466
1467         /**
1468          * @param array $sinks
1469          */
1470         function __construct( $sinks ) {
1471                 $this->sinks = $sinks;
1472                 $this->count = count( $sinks );
1473         }
1474
1475         /**
1476          * @param string $string
1477          */
1478         function writeOpenStream( $string ) {
1479                 for ( $i = 0; $i < $this->count; $i++ ) {
1480                         $this->sinks[$i]->writeOpenStream( $string );
1481                 }
1482         }
1483
1484         /**
1485          * @param string $string
1486          */
1487         function writeCloseStream( $string ) {
1488                 for ( $i = 0; $i < $this->count; $i++ ) {
1489                         $this->sinks[$i]->writeCloseStream( $string );
1490                 }
1491         }
1492
1493         /**
1494          * @param object $page
1495          * @param string $string
1496          */
1497         function writeOpenPage( $page, $string ) {
1498                 for ( $i = 0; $i < $this->count; $i++ ) {
1499                         $this->sinks[$i]->writeOpenPage( $page, $string );
1500                 }
1501         }
1502
1503         /**
1504          * @param string $string
1505          */
1506         function writeClosePage( $string ) {
1507                 for ( $i = 0; $i < $this->count; $i++ ) {
1508                         $this->sinks[$i]->writeClosePage( $string );
1509                 }
1510         }
1511
1512         /**
1513          * @param object $rev
1514          * @param string $string
1515          */
1516         function writeRevision( $rev, $string ) {
1517                 for ( $i = 0; $i < $this->count; $i++ ) {
1518                         $this->sinks[$i]->writeRevision( $rev, $string );
1519                 }
1520         }
1521
1522         /**
1523          * @param array $newnames
1524          */
1525         function closeRenameAndReopen( $newnames ) {
1526                 $this->closeAndRename( $newnames, true );
1527         }
1528
1529         /**
1530          * @param array $newnames
1531          * @param bool $open
1532          */
1533         function closeAndRename( $newnames, $open = false ) {
1534                 for ( $i = 0; $i < $this->count; $i++ ) {
1535                         $this->sinks[$i]->closeAndRename( $newnames[$i], $open );
1536                 }
1537         }
1538
1539         /**
1540          * @return array
1541          */
1542         function getFilenames() {
1543                 $filenames = array();
1544                 for ( $i = 0; $i < $this->count; $i++ ) {
1545                         $filenames[] = $this->sinks[$i]->getFilenames();
1546                 }
1547                 return $filenames;
1548         }
1549 }