includes/Export.php

   1 <?php
   2 /**
   3  * Base classes for dumps and export
   4  *
   5  * Copyright © 2003, 2005, 2006 Brion Vibber <brion@pobox.com>
   6  * https://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  */
  25
  26 /**
  27  * @defgroup Dump Dump
  28  */
  29
  30 /**
  31  * @ingroup SpecialPage Dump
  32  */
  33 class WikiExporter {
  34         /** @var bool Return distinct author list (when not returning full history) */
  35         public $list_authors = false;
  36
  37         /** @var bool */
  38         public $dumpUploads = false;
  39
  40         /** @var bool */
  41         public $dumpUploadFileContents = false;
  42
  43         /** @var string */
  44         protected $author_list = "";
  45
  46         const FULL = 1;
  47         const CURRENT = 2;
  48         const STABLE = 4; // extension defined
  49         const LOGS = 8;
  50         const RANGE = 16;
  51
  52         const BUFFER = 0;
  53         const STREAM = 1;
  54
  55         const TEXT = 0;
  56         const STUB = 1;
  57
  58         /** @var int */
  59         protected $buffer;
  60
  61         /** @var int */
  62         protected $text;
  63
  64         /** @var DumpOutput */
  65         protected $sink;
  66
  67         /**
  68          * Returns the export schema version.
  69          * @return string
  70          */
  71         public static function schemaVersion() {
  72                 return "0.8";
  73         }
  74
  75         /**
  76          * If using WikiExporter::STREAM to stream a large amount of data,
  77          * provide a database connection which is not managed by
  78          * LoadBalancer to read from: some history blob types will
  79          * make additional queries to pull source data while the
  80          * main query is still running.
  81          *
  82          * @param DatabaseBase $db
  83          * @param int|array $history One of WikiExporter::FULL, WikiExporter::CURRENT,
  84          *   WikiExporter::RANGE or WikiExporter::STABLE, or an associative array:
  85          *   - offset: non-inclusive offset at which to start the query
  86          *   - limit: maximum number of rows to return
  87          *   - dir: "asc" or "desc" timestamp order
  88          * @param int $buffer One of WikiExporter::BUFFER or WikiExporter::STREAM
  89          * @param int $text One of WikiExporter::TEXT or WikiExporter::STUB
  90          */
  91         function __construct( $db, $history = WikiExporter::CURRENT,
  92                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  93                 $this->db = $db;
  94                 $this->history = $history;
  95                 $this->buffer = $buffer;
  96                 $this->writer = new XmlDumpWriter();
  97                 $this->sink = new DumpOutput();
  98                 $this->text = $text;
  99         }
 100
 101         /**
 102          * Set the DumpOutput or DumpFilter object which will receive
 103          * various row objects and XML output for filtering. Filters
 104          * can be chained or used as callbacks.
 105          *
 106          * @param DumpOutput $sink
 107          */
 108         public function setOutputSink( &$sink ) {
 109                 $this->sink =& $sink;
 110         }
 111
 112         public function openStream() {
 113                 $output = $this->writer->openStream();
 114                 $this->sink->writeOpenStream( $output );
 115         }
 116
 117         public function closeStream() {
 118                 $output = $this->writer->closeStream();
 119                 $this->sink->writeCloseStream( $output );
 120         }
 121
 122         /**
 123          * Dumps a series of page and revision records for all pages
 124          * in the database, either including complete history or only
 125          * the most recent version.
 126          */
 127         public function allPages() {
 128                 $this->dumpFrom( '' );
 129         }
 130
 131         /**
 132          * Dumps a series of page and revision records for those pages
 133          * in the database falling within the page_id range given.
 134          * @param int $start Inclusive lower limit (this id is included)
 135          * @param int $end Exclusive upper limit (this id is not included)
 136          *   If 0, no upper limit.
 137          */
 138         public function pagesByRange( $start, $end ) {
 139                 $condition = 'page_id >= ' . intval( $start );
 140                 if ( $end ) {
 141                         $condition .= ' AND page_id < ' . intval( $end );
 142                 }
 143                 $this->dumpFrom( $condition );
 144         }
 145
 146         /**
 147          * Dumps a series of page and revision records for those pages
 148          * in the database with revisions falling within the rev_id range given.
 149          * @param int $start Inclusive lower limit (this id is included)
 150          * @param int $end Exclusive upper limit (this id is not included)
 151          *   If 0, no upper limit.
 152          */
 153         public function revsByRange( $start, $end ) {
 154                 $condition = 'rev_id >= ' . intval( $start );
 155                 if ( $end ) {
 156                         $condition .= ' AND rev_id < ' . intval( $end );
 157                 }
 158                 $this->dumpFrom( $condition );
 159         }
 160
 161         /**
 162          * @param Title $title
 163          */
 164         public function pageByTitle( $title ) {
 165                 $this->dumpFrom(
 166                         'page_namespace=' . $title->getNamespace() .
 167                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 168         }
 169
 170         /**
 171          * @param string $name
 172          * @throws MWException
 173          */
 174         public function pageByName( $name ) {
 175                 $title = Title::newFromText( $name );
 176                 if ( is_null( $title ) ) {
 177                         throw new MWException( "Can't export invalid title" );
 178                 } else {
 179                         $this->pageByTitle( $title );
 180                 }
 181         }
 182
 183         /**
 184          * @param array $names
 185          */
 186         public function pagesByName( $names ) {
 187                 foreach ( $names as $name ) {
 188                         $this->pageByName( $name );
 189                 }
 190         }
 191
 192         public function allLogs() {
 193                 $this->dumpFrom( '' );
 194         }
 195
 196         /**
 197          * @param int $start
 198          * @param int $end
 199          */
 200         public function logsByRange( $start, $end ) {
 201                 $condition = 'log_id >= ' . intval( $start );
 202                 if ( $end ) {
 203                         $condition .= ' AND log_id < ' . intval( $end );
 204                 }
 205                 $this->dumpFrom( $condition );
 206         }
 207
 208         /**
 209          * Generates the distinct list of authors of an article
 210          * Not called by default (depends on $this->list_authors)
 211          * Can be set by Special:Export when not exporting whole history
 212          *
 213          * @param array $cond
 214          */
 215         protected function do_list_authors( $cond ) {
 216                 wfProfileIn( __METHOD__ );
 217                 $this->author_list = "<contributors>";
 218                 // rev_deleted
 219
 220                 $res = $this->db->select(
 221                         array( 'page', 'revision' ),
 222                         array( 'DISTINCT rev_user_text', 'rev_user' ),
 223                         array(
 224                                 $this->db->bitAnd( 'rev_deleted', Revision::DELETED_USER ) . ' = 0',
 225                                 $cond,
 226                                 'page_id = rev_id',
 227                         ),
 228                         __METHOD__
 229                 );
 230
 231                 foreach ( $res as $row ) {
 232                         $this->author_list .= "<contributor>" .
 233                                 "<username>" .
 234                                 htmlentities( $row->rev_user_text ) .
 235                                 "</username>" .
 236                                 "<id>" .
 237                                 $row->rev_user .
 238                                 "</id>" .
 239                                 "</contributor>";
 240                 }
 241                 $this->author_list .= "</contributors>";
 242                 wfProfileOut( __METHOD__ );
 243         }
 244
 245         /**
 246          * @param string $cond
 247          * @throws MWException
 248          * @throws Exception
 249          */
 250         protected function dumpFrom( $cond = '' ) {
 251                 wfProfileIn( __METHOD__ );
 252                 # For logging dumps...
 253                 if ( $this->history & self::LOGS ) {
 254                         $where = array( 'user_id = log_user' );
 255                         # Hide private logs
 256                         $hideLogs = LogEventsList::getExcludeClause( $this->db );
 257                         if ( $hideLogs ) {
 258                                 $where[] = $hideLogs;
 259                         }
 260                         # Add on any caller specified conditions
 261                         if ( $cond ) {
 262                                 $where[] = $cond;
 263                         }
 264                         # Get logging table name for logging.* clause
 265                         $logging = $this->db->tableName( 'logging' );
 266
 267                         if ( $this->buffer == WikiExporter::STREAM ) {
 268                                 $prev = $this->db->bufferResults( false );
 269                         }
 270                         $wrapper = null; // Assuring $wrapper is not undefined, if exception occurs early
 271                         try {
 272                                 $result = $this->db->select( array( 'logging', 'user' ),
 273                                         array( "{$logging}.*", 'user_name' ), // grab the user name
 274                                         $where,
 275                                         __METHOD__,
 276                                         array( 'ORDER BY' => 'log_id', 'USE INDEX' => array( 'logging' => 'PRIMARY' ) )
 277                                 );
 278                                 $wrapper = $this->db->resultObject( $result );
 279                                 $this->outputLogStream( $wrapper );
 280                                 if ( $this->buffer == WikiExporter::STREAM ) {
 281                                         $this->db->bufferResults( $prev );
 282                                 }
 283                         } catch ( Exception $e ) {
 284                                 // Throwing the exception does not reliably free the resultset, and
 285                                 // would also leave the connection in unbuffered mode.
 286
 287                                 // Freeing result
 288                                 try {
 289                                         if ( $wrapper ) {
 290                                                 $wrapper->free();
 291                                         }
 292                                 } catch ( Exception $e2 ) {
 293                                         // Already in panic mode -> ignoring $e2 as $e has
 294                                         // higher priority
 295                                 }
 296
 297                                 // Putting database back in previous buffer mode
 298                                 try {
 299                                         if ( $this->buffer == WikiExporter::STREAM ) {
 300                                                 $this->db->bufferResults( $prev );
 301                                         }
 302                                 } catch ( Exception $e2 ) {
 303                                         // Already in panic mode -> ignoring $e2 as $e has
 304                                         // higher priority
 305                                 }
 306
 307                                 // Inform caller about problem
 308                                 wfProfileOut( __METHOD__ );
 309                                 throw $e;
 310                         }
 311                 # For page dumps...
 312                 } else {
 313                         $tables = array( 'page', 'revision' );
 314                         $opts = array( 'ORDER BY' => 'page_id ASC' );
 315                         $opts['USE INDEX'] = array();
 316                         $join = array();
 317                         if ( is_array( $this->history ) ) {
 318                                 # Time offset/limit for all pages/history...
 319                                 $revJoin = 'page_id=rev_page';
 320                                 # Set time order
 321                                 if ( $this->history['dir'] == 'asc' ) {
 322                                         $op = '>';
 323                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 324                                 } else {
 325                                         $op = '<';
 326                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 327                                 }
 328                                 # Set offset
 329                                 if ( !empty( $this->history['offset'] ) ) {
 330                                         $revJoin .= " AND rev_timestamp $op " .
 331                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 332                                 }
 333                                 $join['revision'] = array( 'INNER JOIN', $revJoin );
 334                                 # Set query limit
 335                                 if ( !empty( $this->history['limit'] ) ) {
 336                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 337                                 }
 338                         } elseif ( $this->history & WikiExporter::FULL ) {
 339                                 # Full history dumps...
 340                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 341                         } elseif ( $this->history & WikiExporter::CURRENT ) {
 342                                 # Latest revision dumps...
 343                                 if ( $this->list_authors && $cond != '' ) { // List authors, if so desired
 344                                         $this->do_list_authors( $cond );
 345                                 }
 346                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 347                         } elseif ( $this->history & WikiExporter::STABLE ) {
 348                                 # "Stable" revision dumps...
 349                                 # Default JOIN, to be overridden...
 350                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 351                                 # One, and only one hook should set this, and return false
 352                                 if ( wfRunHooks( 'WikiExporter::dumpStableQuery', array( &$tables, &$opts, &$join ) ) ) {
 353                                         wfProfileOut( __METHOD__ );
 354                                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 355                                 }
 356                         } elseif ( $this->history & WikiExporter::RANGE ) {
 357                                 # Dump of revisions within a specified range
 358                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 359                                 $opts['ORDER BY'] = array( 'rev_page ASC', 'rev_id ASC' );
 360                         } else {
 361                                 # Unknown history specification parameter?
 362                                 wfProfileOut( __METHOD__ );
 363                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 364                         }
 365                         # Query optimization hacks
 366                         if ( $cond == '' ) {
 367                                 $opts[] = 'STRAIGHT_JOIN';
 368                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 369                         }
 370                         # Build text join options
 371                         if ( $this->text != WikiExporter::STUB ) { // 1-pass
 372                                 $tables[] = 'text';
 373                                 $join['text'] = array( 'INNER JOIN', 'rev_text_id=old_id' );
 374                         }
 375
 376                         if ( $this->buffer == WikiExporter::STREAM ) {
 377                                 $prev = $this->db->bufferResults( false );
 378                         }
 379
 380                         $wrapper = null; // Assuring $wrapper is not undefined, if exception occurs early
 381                         try {
 382                                 wfRunHooks( 'ModifyExportQuery',
 383                                                 array( $this->db, &$tables, &$cond, &$opts, &$join ) );
 384
 385                                 # Do the query!
 386                                 $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 387                                 $wrapper = $this->db->resultObject( $result );
 388                                 # Output dump results
 389                                 $this->outputPageStream( $wrapper );
 390
 391                                 if ( $this->buffer == WikiExporter::STREAM ) {
 392                                         $this->db->bufferResults( $prev );
 393                                 }
 394                         } catch ( Exception $e ) {
 395                                 // Throwing the exception does not reliably free the resultset, and
 396                                 // would also leave the connection in unbuffered mode.
 397
 398                                 // Freeing result
 399                                 try {
 400                                         if ( $wrapper ) {
 401                                                 $wrapper->free();
 402                                         }
 403                                 } catch ( Exception $e2 ) {
 404                                         // Already in panic mode -> ignoring $e2 as $e has
 405                                         // higher priority
 406                                 }
 407
 408                                 // Putting database back in previous buffer mode
 409                                 try {
 410                                         if ( $this->buffer == WikiExporter::STREAM ) {
 411                                                 $this->db->bufferResults( $prev );
 412                                         }
 413                                 } catch ( Exception $e2 ) {
 414                                         // Already in panic mode -> ignoring $e2 as $e has
 415                                         // higher priority
 416                                 }
 417
 418                                 // Inform caller about problem
 419                                 throw $e;
 420                         }
 421                 }
 422                 wfProfileOut( __METHOD__ );
 423         }
 424
 425         /**
 426          * Runs through a query result set dumping page and revision records.
 427          * The result set should be sorted/grouped by page to avoid duplicate
 428          * page records in the output.
 429          *
 430          * Should be safe for
 431          * streaming (non-buffered) queries, as long as it was made on a
 432          * separate database connection not managed by LoadBalancer; some
 433          * blob storage types will make queries to pull source data.
 434          *
 435          * @param ResultWrapper $resultset
 436          */
 437         protected function outputPageStream( $resultset ) {
 438                 $last = null;
 439                 foreach ( $resultset as $row ) {
 440                         if ( $last === null ||
 441                                 $last->page_namespace != $row->page_namespace ||
 442                                 $last->page_title != $row->page_title ) {
 443                                 if ( $last !== null ) {
 444                                         $output = '';
 445                                         if ( $this->dumpUploads ) {
 446                                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 447                                         }
 448                                         $output .= $this->writer->closePage();
 449                                         $this->sink->writeClosePage( $output );
 450                                 }
 451                                 $output = $this->writer->openPage( $row );
 452                                 $this->sink->writeOpenPage( $row, $output );
 453                                 $last = $row;
 454                         }
 455                         $output = $this->writer->writeRevision( $row );
 456                         $this->sink->writeRevision( $row, $output );
 457                 }
 458                 if ( $last !== null ) {
 459                         $output = '';
 460                         if ( $this->dumpUploads ) {
 461                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 462                         }
 463                         $output .= $this->author_list;
 464                         $output .= $this->writer->closePage();
 465                         $this->sink->writeClosePage( $output );
 466                 }
 467         }
 468
 469         /**
 470          * @param array $resultset
 471          */
 472         protected function outputLogStream( $resultset ) {
 473                 foreach ( $resultset as $row ) {
 474                         $output = $this->writer->writeLogItem( $row );
 475                         $this->sink->writeLogItem( $row, $output );
 476                 }
 477         }
 478 }
 479
 480 /**
 481  * @ingroup Dump
 482  */
 483 class XmlDumpWriter {
 484         /**
 485          * Returns the export schema version.
 486          * @deprecated since 1.20; use WikiExporter::schemaVersion() instead
 487          * @return string
 488          */
 489         function schemaVersion() {
 490                 wfDeprecated( __METHOD__, '1.20' );
 491                 return WikiExporter::schemaVersion();
 492         }
 493
 494         /**
 495          * Opens the XML output stream's root "<mediawiki>" element.
 496          * This does not include an xml directive, so is safe to include
 497          * as a subelement in a larger XML stream. Namespace and XML Schema
 498          * references are included.
 499          *
 500          * Output will be encoded in UTF-8.
 501          *
 502          * @return string
 503          */
 504         function openStream() {
 505                 global $wgLanguageCode;
 506                 $ver = WikiExporter::schemaVersion();
 507                 return Xml::element( 'mediawiki', array(
 508                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 509                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 510                         /*
 511                          * When a new version of the schema is created, it needs staging on mediawiki.org.
 512                          * This requires a change in the operations/mediawiki-config git repo.
 513                          *
 514                          * Create a changeset like https://gerrit.wikimedia.org/r/#/c/149643/ in which
 515                          * you copy in the new xsd file.
 516                          *
 517                          * After it is reviewed, merged and deployed (sync-docroot), the index.html needs purging.
 518                          * echo "http://www.mediawiki.org/xml/index.html" | mwscript purgeList.php --wiki=aawiki
 519                          */
 520                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 521                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 522                         'version'            => $ver,
 523                         'xml:lang'           => $wgLanguageCode ),
 524                         null ) .
 525                         "\n" .
 526                         $this->siteInfo();
 527         }
 528
 529         /**
 530          * @return string
 531          */
 532         function siteInfo() {
 533                 $info = array(
 534                         $this->sitename(),
 535                         $this->homelink(),
 536                         $this->generator(),
 537                         $this->caseSetting(),
 538                         $this->namespaces() );
 539                 return "  <siteinfo>\n    " .
 540                         implode( "\n    ", $info ) .
 541                         "\n  </siteinfo>\n";
 542         }
 543
 544         /**
 545          * @return string
 546          */
 547         function sitename() {
 548                 global $wgSitename;
 549                 return Xml::element( 'sitename', array(), $wgSitename );
 550         }
 551
 552         /**
 553          * @return string
 554          */
 555         function generator() {
 556                 global $wgVersion;
 557                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 558         }
 559
 560         /**
 561          * @return string
 562          */
 563         function homelink() {
 564                 return Xml::element( 'base', array(), Title::newMainPage()->getCanonicalURL() );
 565         }
 566
 567         /**
 568          * @return string
 569          */
 570         function caseSetting() {
 571                 global $wgCapitalLinks;
 572                 // "case-insensitive" option is reserved for future
 573                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 574                 return Xml::element( 'case', array(), $sensitivity );
 575         }
 576
 577         /**
 578          * @return string
 579          */
 580         function namespaces() {
 581                 global $wgContLang;
 582                 $spaces = "<namespaces>\n";
 583                 foreach ( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 584                         $spaces .= '      ' .
 585                                 Xml::element( 'namespace',
 586                                         array(
 587                                                 'key' => $ns,
 588                                                 'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 589                                         ), $title ) . "\n";
 590                 }
 591                 $spaces .= "    </namespaces>";
 592                 return $spaces;
 593         }
 594
 595         /**
 596          * Closes the output stream with the closing root element.
 597          * Call when finished dumping things.
 598          *
 599          * @return string
 600          */
 601         function closeStream() {
 602                 return "</mediawiki>\n";
 603         }
 604
 605         /**
 606          * Opens a "<page>" section on the output stream, with data
 607          * from the given database row.
 608          *
 609          * @param object $row
 610          * @return string
 611          */
 612         public function openPage( $row ) {
 613                 $out = "  <page>\n";
 614                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 615                 $out .= '    ' . Xml::elementClean( 'title', array(), self::canonicalTitle( $title ) ) . "\n";
 616                 $out .= '    ' . Xml::element( 'ns', array(), strval( $row->page_namespace ) ) . "\n";
 617                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 618                 if ( $row->page_is_redirect ) {
 619                         $page = WikiPage::factory( $title );
 620                         $redirect = $page->getRedirectTarget();
 621                         if ( $redirect instanceof Title && $redirect->isValidRedirectTarget() ) {
 622                                 $out .= '    ';
 623                                 $out .= Xml::element( 'redirect', array( 'title' => self::canonicalTitle( $redirect ) ) );
 624                                 $out .= "\n";
 625                         }
 626                 }
 627
 628                 if ( $row->page_restrictions != '' ) {
 629                         $out .= '    ' . Xml::element( 'restrictions', array(),
 630                                 strval( $row->page_restrictions ) ) . "\n";
 631                 }
 632
 633                 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 634
 635                 return $out;
 636         }
 637
 638         /**
 639          * Closes a "<page>" section on the output stream.
 640          *
 641          * @access private
 642          * @return string
 643          */
 644         function closePage() {
 645                 return "  </page>\n";
 646         }
 647
 648         /**
 649          * Dumps a "<revision>" section on the output stream, with
 650          * data filled in from the given database row.
 651          *
 652          * @param object $row
 653          * @return string
 654          * @access private
 655          */
 656         function writeRevision( $row ) {
 657                 wfProfileIn( __METHOD__ );
 658
 659                 $out = "    <revision>\n";
 660                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 661                 if ( isset( $row->rev_parent_id ) && $row->rev_parent_id ) {
 662                         $out .= "      " . Xml::element( 'parentid', null, strval( $row->rev_parent_id ) ) . "\n";
 663                 }
 664
 665                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 666
 667                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_USER ) ) {
 668                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 669                 } else {
 670                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 671                 }
 672
 673                 if ( isset( $row->rev_minor_edit ) && $row->rev_minor_edit ) {
 674                         $out .= "      <minor/>\n";
 675                 }
 676                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_COMMENT ) ) {
 677                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 678                 } elseif ( $row->rev_comment != '' ) {
 679                         $out .= "      " . Xml::elementClean( 'comment', array(), strval( $row->rev_comment ) ) . "\n";
 680                 }
 681
 682                 if ( isset( $row->rev_content_model ) && !is_null( $row->rev_content_model ) ) {
 683                         $content_model = strval( $row->rev_content_model );
 684                 } else {
 685                         // probably using $wgContentHandlerUseDB = false;
 686                         $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 687                         $content_model = ContentHandler::getDefaultModelFor( $title );
 688                 }
 689
 690                 $content_handler = ContentHandler::getForModelID( $content_model );
 691
 692                 if ( isset( $row->rev_content_format ) && !is_null( $row->rev_content_format ) ) {
 693                         $content_format = strval( $row->rev_content_format );
 694                 } else {
 695                         // probably using $wgContentHandlerUseDB = false;
 696                         $content_format = $content_handler->getDefaultFormat();
 697                 }
 698
 699                 $text = '';
 700                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_TEXT ) ) {
 701                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 702                 } elseif ( isset( $row->old_text ) ) {
 703                         // Raw text from the database may have invalid chars
 704                         $text = strval( Revision::getRevisionText( $row ) );
 705                         $text = $content_handler->exportTransform( $text, $content_format );
 706                         $out .= "      " . Xml::elementClean( 'text',
 707                                 array( 'xml:space' => 'preserve', 'bytes' => intval( $row->rev_len ) ),
 708                                 strval( $text ) ) . "\n";
 709                 } else {
 710                         // Stub output
 711                         $out .= "      " . Xml::element( 'text',
 712                                 array( 'id' => $row->rev_text_id, 'bytes' => intval( $row->rev_len ) ),
 713                                 "" ) . "\n";
 714                 }
 715
 716                 if ( isset( $row->rev_sha1 )
 717                         && $row->rev_sha1
 718                         && !( $row->rev_deleted & Revision::DELETED_TEXT )
 719                 ) {
 720                         $out .= "      " . Xml::element( 'sha1', null, strval( $row->rev_sha1 ) ) . "\n";
 721                 } else {
 722                         $out .= "      <sha1/>\n";
 723                 }
 724
 725                 $out .= "      " . Xml::element( 'model', null, strval( $content_model ) ) . "\n";
 726                 $out .= "      " . Xml::element( 'format', null, strval( $content_format ) ) . "\n";
 727
 728                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 729
 730                 $out .= "    </revision>\n";
 731
 732                 wfProfileOut( __METHOD__ );
 733                 return $out;
 734         }
 735
 736         /**
 737          * Dumps a "<logitem>" section on the output stream, with
 738          * data filled in from the given database row.
 739          *
 740          * @param object $row
 741          * @return string
 742          * @access private
 743          */
 744         function writeLogItem( $row ) {
 745                 wfProfileIn( __METHOD__ );
 746
 747                 $out = "  <logitem>\n";
 748                 $out .= "    " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 749
 750                 $out .= $this->writeTimestamp( $row->log_timestamp, "    " );
 751
 752                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 753                         $out .= "    " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 754                 } else {
 755                         $out .= $this->writeContributor( $row->log_user, $row->user_name, "    " );
 756                 }
 757
 758                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 759                         $out .= "    " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 760                 } elseif ( $row->log_comment != '' ) {
 761                         $out .= "    " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 762                 }
 763
 764                 $out .= "    " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 765                 $out .= "    " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 766
 767                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 768                         $out .= "    " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 769                 } else {
 770                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 771                         $out .= "    " . Xml::elementClean( 'logtitle', null, self::canonicalTitle( $title ) ) . "\n";
 772                         $out .= "    " . Xml::elementClean( 'params',
 773                                 array( 'xml:space' => 'preserve' ),
 774                                 strval( $row->log_params ) ) . "\n";
 775                 }
 776
 777                 $out .= "  </logitem>\n";
 778
 779                 wfProfileOut( __METHOD__ );
 780                 return $out;
 781         }
 782
 783         /**
 784          * @param string $timestamp
 785          * @param string $indent Default to six spaces
 786          * @return string
 787          */
 788         function writeTimestamp( $timestamp, $indent = "      " ) {
 789                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 790                 return $indent . Xml::element( 'timestamp', null, $ts ) . "\n";
 791         }
 792
 793         /**
 794          * @param int $id
 795          * @param string $text
 796          * @param string $indent Default to six spaces
 797          * @return string
 798          */
 799         function writeContributor( $id, $text, $indent = "      " ) {
 800                 $out = $indent . "<contributor>\n";
 801                 if ( $id || !IP::isValid( $text ) ) {
 802                         $out .= $indent . "  " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 803                         $out .= $indent . "  " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 804                 } else {
 805                         $out .= $indent . "  " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 806                 }
 807                 $out .= $indent . "</contributor>\n";
 808                 return $out;
 809         }
 810
 811         /**
 812          * Warning! This data is potentially inconsistent. :(
 813          * @param object $row
 814          * @param bool $dumpContents
 815          * @return string
 816          */
 817         function writeUploads( $row, $dumpContents = false ) {
 818                 if ( $row->page_namespace == NS_FILE ) {
 819                         $img = wfLocalFile( $row->page_title );
 820                         if ( $img && $img->exists() ) {
 821                                 $out = '';
 822                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 823                                         $out .= $this->writeUpload( $ver, $dumpContents );
 824                                 }
 825                                 $out .= $this->writeUpload( $img, $dumpContents );
 826                                 return $out;
 827                         }
 828                 }
 829                 return '';
 830         }
 831
 832         /**
 833          * @param File $file
 834          * @param bool $dumpContents
 835          * @return string
 836          */
 837         function writeUpload( $file, $dumpContents = false ) {
 838                 if ( $file->isOld() ) {
 839                         $archiveName = "      " .
 840                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 841                 } else {
 842                         $archiveName = '';
 843                 }
 844                 if ( $dumpContents ) {
 845                         $be = $file->getRepo()->getBackend();
 846                         # Dump file as base64
 847                         # Uses only XML-safe characters, so does not need escaping
 848                         # @todo Too bad this loads the contents into memory (script might swap)
 849                         $contents = '      <contents encoding="base64">' .
 850                                 chunk_split( base64_encode(
 851                                         $be->getFileContents( array( 'src' => $file->getPath() ) ) ) ) .
 852                                 "      </contents>\n";
 853                 } else {
 854                         $contents = '';
 855                 }
 856                 if ( $file->isDeleted( File::DELETED_COMMENT ) ) {
 857                         $comment = Xml::element( 'comment', array( 'deleted' => 'deleted' ) );
 858                 } else {
 859                         $comment = Xml::elementClean( 'comment', null, $file->getDescription() );
 860                 }
 861                 return "    <upload>\n" .
 862                         $this->writeTimestamp( $file->getTimestamp() ) .
 863                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 864                         "      " . $comment . "\n" .
 865                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 866                         $archiveName .
 867                         "      " . Xml::element( 'src', null, $file->getCanonicalURL() ) . "\n" .
 868                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 869                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 870                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 871                         $contents .
 872                         "    </upload>\n";
 873         }
 874
 875         /**
 876          * Return prefixed text form of title, but using the content language's
 877          * canonical namespace. This skips any special-casing such as gendered
 878          * user namespaces -- which while useful, are not yet listed in the
 879          * XML "<siteinfo>" data so are unsafe in export.
 880          *
 881          * @param Title $title
 882          * @return string
 883          * @since 1.18
 884          */
 885         public static function canonicalTitle( Title $title ) {
 886                 if ( $title->isExternal() ) {
 887                         return $title->getPrefixedText();
 888                 }
 889
 890                 global $wgContLang;
 891                 $prefix = str_replace( '_', ' ', $wgContLang->getNsText( $title->getNamespace() ) );
 892
 893                 if ( $prefix !== '' ) {
 894                         $prefix .= ':';
 895                 }
 896
 897                 return $prefix . $title->getText();
 898         }
 899 }
 900
 901 /**
 902  * Base class for output stream; prints to stdout or buffer or wherever.
 903  * @ingroup Dump
 904  */
 905 class DumpOutput {
 906
 907         /**
 908          * @param string $string
 909          */
 910         function writeOpenStream( $string ) {
 911                 $this->write( $string );
 912         }
 913
 914         /**
 915          * @param string $string
 916          */
 917         function writeCloseStream( $string ) {
 918                 $this->write( $string );
 919         }
 920
 921         /**
 922          * @param object $page
 923          * @param string $string
 924          */
 925         function writeOpenPage( $page, $string ) {
 926                 $this->write( $string );
 927         }
 928
 929         /**
 930          * @param string $string
 931          */
 932         function writeClosePage( $string ) {
 933                 $this->write( $string );
 934         }
 935
 936         /**
 937          * @param object $rev
 938          * @param string $string
 939          */
 940         function writeRevision( $rev, $string ) {
 941                 $this->write( $string );
 942         }
 943
 944         /**
 945          * @param object $rev
 946          * @param string $string
 947          */
 948         function writeLogItem( $rev, $string ) {
 949                 $this->write( $string );
 950         }
 951
 952         /**
 953          * Override to write to a different stream type.
 954          * @param string $string
 955          * @return bool
 956          */
 957         function write( $string ) {
 958                 print $string;
 959         }
 960
 961         /**
 962          * Close the old file, move it to a specified name,
 963          * and reopen new file with the old name. Use this
 964          * for writing out a file in multiple pieces
 965          * at specified checkpoints (e.g. every n hours).
 966          * @param string|array $newname File name. May be a string or an array with one element
 967          */
 968         function closeRenameAndReopen( $newname ) {
 969         }
 970
 971         /**
 972          * Close the old file, and move it to a specified name.
 973          * Use this for the last piece of a file written out
 974          * at specified checkpoints (e.g. every n hours).
 975          * @param string|array $newname File name. May be a string or an array with one element
 976          * @param bool $open If true, a new file with the old filename will be opened
 977          *   again for writing (default: false)
 978          */
 979         function closeAndRename( $newname, $open = false ) {
 980         }
 981
 982         /**
 983          * Returns the name of the file or files which are
 984          * being written to, if there are any.
 985          * @return null
 986          */
 987         function getFilenames() {
 988                 return null;
 989         }
 990 }
 991
 992 /**
 993  * Stream outputter to send data to a file.
 994  * @ingroup Dump
 995  */
 996 class DumpFileOutput extends DumpOutput {
 997         protected $handle = false, $filename;
 998
 999         /**
1000          * @param string $file
1001          */
1002         function __construct( $file ) {
1003                 $this->handle = fopen( $file, "wt" );
1004                 $this->filename = $file;
1005         }
1006
1007         /**
1008          * @param string $string
1009          */
1010         function writeCloseStream( $string ) {
1011                 parent::writeCloseStream( $string );
1012                 if ( $this->handle ) {
1013                         fclose( $this->handle );
1014                         $this->handle = false;
1015                 }
1016         }
1017
1018         /**
1019          * @param string $string
1020          */
1021         function write( $string ) {
1022                 fputs( $this->handle, $string );
1023         }
1024
1025         /**
1026          * @param string $newname
1027          */
1028         function closeRenameAndReopen( $newname ) {
1029                 $this->closeAndRename( $newname, true );
1030         }
1031
1032         /**
1033          * @param string $newname
1034          * @throws MWException
1035          */
1036         function renameOrException( $newname ) {
1037                         if ( !rename( $this->filename, $newname ) ) {
1038                                 throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" );
1039                         }
1040         }
1041
1042         /**
1043          * @param array $newname
1044          * @return string
1045          * @throws MWException
1046          */
1047         function checkRenameArgCount( $newname ) {
1048                 if ( is_array( $newname ) ) {
1049                         if ( count( $newname ) > 1 ) {
1050                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
1051                         } else {
1052                                 $newname = $newname[0];
1053                         }
1054                 }
1055                 return $newname;
1056         }
1057
1058         /**
1059          * @param string $newname
1060          * @param bool $open
1061          */
1062         function closeAndRename( $newname, $open = false ) {
1063                 $newname = $this->checkRenameArgCount( $newname );
1064                 if ( $newname ) {
1065                         if ( $this->handle ) {
1066                                 fclose( $this->handle );
1067                                 $this->handle = false;
1068                         }
1069                         $this->renameOrException( $newname );
1070                         if ( $open ) {
1071                                 $this->handle = fopen( $this->filename, "wt" );
1072                         }
1073                 }
1074         }
1075
1076         /**
1077          * @return string|null
1078          */
1079         function getFilenames() {
1080                 return $this->filename;
1081         }
1082 }
1083
1084 /**
1085  * Stream outputter to send data to a file via some filter program.
1086  * Even if compression is available in a library, using a separate
1087  * program can allow us to make use of a multi-processor system.
1088  * @ingroup Dump
1089  */
1090 class DumpPipeOutput extends DumpFileOutput {
1091         protected $command, $filename;
1092         protected $procOpenResource = false;
1093
1094         /**
1095          * @param string $command
1096          * @param string $file
1097          */
1098         function __construct( $command, $file = null ) {
1099                 if ( !is_null( $file ) ) {
1100                         $command .= " > " . wfEscapeShellArg( $file );
1101                 }
1102
1103                 $this->startCommand( $command );
1104                 $this->command = $command;
1105                 $this->filename = $file;
1106         }
1107
1108         /**
1109          * @param string $string
1110          */
1111         function writeCloseStream( $string ) {
1112                 parent::writeCloseStream( $string );
1113                 if ( $this->procOpenResource ) {
1114                         proc_close( $this->procOpenResource );
1115                         $this->procOpenResource = false;
1116                 }
1117         }
1118
1119         /**
1120          * @param string $command
1121          */
1122         function startCommand( $command ) {
1123                 $spec = array(
1124                         0 => array( "pipe", "r" ),
1125                 );
1126                 $pipes = array();
1127                 $this->procOpenResource = proc_open( $command, $spec, $pipes );
1128                 $this->handle = $pipes[0];
1129         }
1130
1131         /**
1132          * @param string $newname
1133          */
1134         function closeRenameAndReopen( $newname ) {
1135                 $this->closeAndRename( $newname, true );
1136         }
1137
1138         /**
1139          * @param string $newname
1140          * @param bool $open
1141          */
1142         function closeAndRename( $newname, $open = false ) {
1143                 $newname = $this->checkRenameArgCount( $newname );
1144                 if ( $newname ) {
1145                         if ( $this->handle ) {
1146                                 fclose( $this->handle );
1147                                 $this->handle = false;
1148                         }
1149                         if ( $this->procOpenResource ) {
1150                                 proc_close( $this->procOpenResource );
1151                                 $this->procOpenResource = false;
1152                         }
1153                         $this->renameOrException( $newname );
1154                         if ( $open ) {
1155                                 $command = $this->command;
1156                                 $command .= " > " . wfEscapeShellArg( $this->filename );
1157                                 $this->startCommand( $command );
1158                         }
1159                 }
1160         }
1161 }
1162
1163 /**
1164  * Sends dump output via the gzip compressor.
1165  * @ingroup Dump
1166  */
1167 class DumpGZipOutput extends DumpPipeOutput {
1168         /**
1169          * @param string $file
1170          */
1171         function __construct( $file ) {
1172                 parent::__construct( "gzip", $file );
1173         }
1174 }
1175
1176 /**
1177  * Sends dump output via the bgzip2 compressor.
1178  * @ingroup Dump
1179  */
1180 class DumpBZip2Output extends DumpPipeOutput {
1181         /**
1182          * @param string $file
1183          */
1184         function __construct( $file ) {
1185                 parent::__construct( "bzip2", $file );
1186         }
1187 }
1188
1189 /**
1190  * Sends dump output via the p7zip compressor.
1191  * @ingroup Dump
1192  */
1193 class Dump7ZipOutput extends DumpPipeOutput {
1194         /**
1195          * @param string $file
1196          */
1197         function __construct( $file ) {
1198                 $command = $this->setup7zCommand( $file );
1199                 parent::__construct( $command );
1200                 $this->filename = $file;
1201         }
1202
1203         /**
1204          * @param string $file
1205          * @return string
1206          */
1207         function setup7zCommand( $file ) {
1208                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
1209                 // Suppress annoying useless crap from p7zip
1210                 // Unfortunately this could suppress real error messages too
1211                 $command .= ' >' . wfGetNull() . ' 2>&1';
1212                 return $command;
1213         }
1214
1215         /**
1216          * @param string $newname
1217          * @param bool $open
1218          */
1219         function closeAndRename( $newname, $open = false ) {
1220                 $newname = $this->checkRenameArgCount( $newname );
1221                 if ( $newname ) {
1222                         fclose( $this->handle );
1223                         proc_close( $this->procOpenResource );
1224                         $this->renameOrException( $newname );
1225                         if ( $open ) {
1226                                 $command = $this->setup7zCommand( $this->filename );
1227                                 $this->startCommand( $command );
1228                         }
1229                 }
1230         }
1231 }
1232
1233 /**
1234  * Dump output filter class.
1235  * This just does output filtering and streaming; XML formatting is done
1236  * higher up, so be careful in what you do.
1237  * @ingroup Dump
1238  */
1239 class DumpFilter {
1240         /**
1241          * @var DumpOutput
1242          * FIXME will need to be made protected whenever legacy code
1243          * is updated.
1244          */
1245         public $sink;
1246
1247         /**
1248          * @var bool
1249          */
1250         protected $sendingThisPage;
1251
1252         /**
1253          * @param DumpOutput $sink
1254          */
1255         function __construct( &$sink ) {
1256                 $this->sink =& $sink;
1257         }
1258
1259         /**
1260          * @param string $string
1261          */
1262         function writeOpenStream( $string ) {
1263                 $this->sink->writeOpenStream( $string );
1264         }
1265
1266         /**
1267          * @param string $string
1268          */
1269         function writeCloseStream( $string ) {
1270                 $this->sink->writeCloseStream( $string );
1271         }
1272
1273         /**
1274          * @param object $page
1275          * @param string $string
1276          */
1277         function writeOpenPage( $page, $string ) {
1278                 $this->sendingThisPage = $this->pass( $page, $string );
1279                 if ( $this->sendingThisPage ) {
1280                         $this->sink->writeOpenPage( $page, $string );
1281                 }
1282         }
1283
1284         /**
1285          * @param string $string
1286          */
1287         function writeClosePage( $string ) {
1288                 if ( $this->sendingThisPage ) {
1289                         $this->sink->writeClosePage( $string );
1290                         $this->sendingThisPage = false;
1291                 }
1292         }
1293
1294         /**
1295          * @param object $rev
1296          * @param string $string
1297          */
1298         function writeRevision( $rev, $string ) {
1299                 if ( $this->sendingThisPage ) {
1300                         $this->sink->writeRevision( $rev, $string );
1301                 }
1302         }
1303
1304         /**
1305          * @param object $rev
1306          * @param string $string
1307          */
1308         function writeLogItem( $rev, $string ) {
1309                 $this->sink->writeRevision( $rev, $string );
1310         }
1311
1312         /**
1313          * @param string $newname
1314          */
1315         function closeRenameAndReopen( $newname ) {
1316                 $this->sink->closeRenameAndReopen( $newname );
1317         }
1318
1319         /**
1320          * @param string $newname
1321          * @param bool $open
1322          */
1323         function closeAndRename( $newname, $open = false ) {
1324                 $this->sink->closeAndRename( $newname, $open );
1325         }
1326
1327         /**
1328          * @return array
1329          */
1330         function getFilenames() {
1331                 return $this->sink->getFilenames();
1332         }
1333
1334         /**
1335          * Override for page-based filter types.
1336          * @param object $page
1337          * @return bool
1338          */
1339         function pass( $page ) {
1340                 return true;
1341         }
1342 }
1343
1344 /**
1345  * Simple dump output filter to exclude all talk pages.
1346  * @ingroup Dump
1347  */
1348 class DumpNotalkFilter extends DumpFilter {
1349         /**
1350          * @param object $page
1351          * @return bool
1352          */
1353         function pass( $page ) {
1354                 return !MWNamespace::isTalk( $page->page_namespace );
1355         }
1356 }
1357
1358 /**
1359  * Dump output filter to include or exclude pages in a given set of namespaces.
1360  * @ingroup Dump
1361  */
1362 class DumpNamespaceFilter extends DumpFilter {
1363         /** @var bool */
1364         protected $invert = false;
1365
1366         /** @var array */
1367         protected $namespaces = array();
1368
1369         /**
1370          * @param DumpOutput $sink
1371          * @param array $param
1372          * @throws MWException
1373          */
1374         function __construct( &$sink, $param ) {
1375                 parent::__construct( $sink );
1376
1377                 $constants = array(
1378                         "NS_MAIN"           => NS_MAIN,
1379                         "NS_TALK"           => NS_TALK,
1380                         "NS_USER"           => NS_USER,
1381                         "NS_USER_TALK"      => NS_USER_TALK,
1382                         "NS_PROJECT"        => NS_PROJECT,
1383                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
1384                         "NS_FILE"           => NS_FILE,
1385                         "NS_FILE_TALK"      => NS_FILE_TALK,
1386                         "NS_IMAGE"          => NS_IMAGE, // NS_IMAGE is an alias for NS_FILE
1387                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
1388                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
1389                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
1390                         "NS_TEMPLATE"       => NS_TEMPLATE,
1391                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
1392                         "NS_HELP"           => NS_HELP,
1393                         "NS_HELP_TALK"      => NS_HELP_TALK,
1394                         "NS_CATEGORY"       => NS_CATEGORY,
1395                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
1396
1397                 if ( $param { 0 } == '!' ) {
1398                         $this->invert = true;
1399                         $param = substr( $param, 1 );
1400                 }
1401
1402                 foreach ( explode( ',', $param ) as $key ) {
1403                         $key = trim( $key );
1404                         if ( isset( $constants[$key] ) ) {
1405                                 $ns = $constants[$key];
1406                                 $this->namespaces[$ns] = true;
1407                         } elseif ( is_numeric( $key ) ) {
1408                                 $ns = intval( $key );
1409                                 $this->namespaces[$ns] = true;
1410                         } else {
1411                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
1412                         }
1413                 }
1414         }
1415
1416         /**
1417          * @param object $page
1418          * @return bool
1419          */
1420         function pass( $page ) {
1421                 $match = isset( $this->namespaces[$page->page_namespace] );
1422                 return $this->invert xor $match;
1423         }
1424 }
1425
1426 /**
1427  * Dump output filter to include only the last revision in each page sequence.
1428  * @ingroup Dump
1429  */
1430 class DumpLatestFilter extends DumpFilter {
1431         protected $page;
1432
1433         protected $pageString;
1434
1435         protected $rev;
1436
1437         protected $revString;
1438
1439         /**
1440          * @param object $page
1441          * @param string $string
1442          */
1443         function writeOpenPage( $page, $string ) {
1444                 $this->page = $page;
1445                 $this->pageString = $string;
1446         }
1447
1448         /**
1449          * @param string $string
1450          */
1451         function writeClosePage( $string ) {
1452                 if ( $this->rev ) {
1453                         $this->sink->writeOpenPage( $this->page, $this->pageString );
1454                         $this->sink->writeRevision( $this->rev, $this->revString );
1455                         $this->sink->writeClosePage( $string );
1456                 }
1457                 $this->rev = null;
1458                 $this->revString = null;
1459                 $this->page = null;
1460                 $this->pageString = null;
1461         }
1462
1463         /**
1464          * @param object $rev
1465          * @param string $string
1466          */
1467         function writeRevision( $rev, $string ) {
1468                 if ( $rev->rev_id == $this->page->page_latest ) {
1469                         $this->rev = $rev;
1470                         $this->revString = $string;
1471                 }
1472         }
1473 }
1474
1475 /**
1476  * Base class for output stream; prints to stdout or buffer or wherever.
1477  * @ingroup Dump
1478  */
1479 class DumpMultiWriter {
1480
1481         /**
1482          * @param array $sinks
1483          */
1484         function __construct( $sinks ) {
1485                 $this->sinks = $sinks;
1486                 $this->count = count( $sinks );
1487         }
1488
1489         /**
1490          * @param string $string
1491          */
1492         function writeOpenStream( $string ) {
1493                 for ( $i = 0; $i < $this->count; $i++ ) {
1494                         $this->sinks[$i]->writeOpenStream( $string );
1495                 }
1496         }
1497
1498         /**
1499          * @param string $string
1500          */
1501         function writeCloseStream( $string ) {
1502                 for ( $i = 0; $i < $this->count; $i++ ) {
1503                         $this->sinks[$i]->writeCloseStream( $string );
1504                 }
1505         }
1506
1507         /**
1508          * @param object $page
1509          * @param string $string
1510          */
1511         function writeOpenPage( $page, $string ) {
1512                 for ( $i = 0; $i < $this->count; $i++ ) {
1513                         $this->sinks[$i]->writeOpenPage( $page, $string );
1514                 }
1515         }
1516
1517         /**
1518          * @param string $string
1519          */
1520         function writeClosePage( $string ) {
1521                 for ( $i = 0; $i < $this->count; $i++ ) {
1522                         $this->sinks[$i]->writeClosePage( $string );
1523                 }
1524         }
1525
1526         /**
1527          * @param object $rev
1528          * @param string $string
1529          */
1530         function writeRevision( $rev, $string ) {
1531                 for ( $i = 0; $i < $this->count; $i++ ) {
1532                         $this->sinks[$i]->writeRevision( $rev, $string );
1533                 }
1534         }
1535
1536         /**
1537          * @param array $newnames
1538          */
1539         function closeRenameAndReopen( $newnames ) {
1540                 $this->closeAndRename( $newnames, true );
1541         }
1542
1543         /**
1544          * @param array $newnames
1545          * @param bool $open
1546          */
1547         function closeAndRename( $newnames, $open = false ) {
1548                 for ( $i = 0; $i < $this->count; $i++ ) {
1549                         $this->sinks[$i]->closeAndRename( $newnames[$i], $open );
1550                 }
1551         }
1552
1553         /**
1554          * @return array
1555          */
1556         function getFilenames() {
1557                 $filenames = array();
1558                 for ( $i = 0; $i < $this->count; $i++ ) {
1559                         $filenames[] = $this->sinks[$i]->getFilenames();
1560                 }
1561                 return $filenames;
1562         }
1563 }