includes/Export.php

   1 <?php
   2 /**
   3  * Base classes for dumps and export
   4  *
   5  * Copyright © 2003, 2005, 2006 Brion Vibber <brion@pobox.com>
   6  * https://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  */
  25
  26 /**
  27  * @defgroup Dump Dump
  28  */
  29
  30 /**
  31  * @ingroup SpecialPage Dump
  32  */
  33 class WikiExporter {
  34         /** @var bool Return distinct author list (when not returning full history) */
  35         public $list_authors = false;
  36
  37         /** @var bool */
  38         public $dumpUploads = false;
  39
  40         /** @var bool */
  41         public $dumpUploadFileContents = false;
  42
  43         /** @var string */
  44         protected $author_list = "";
  45
  46         const FULL = 1;
  47         const CURRENT = 2;
  48         const STABLE = 4; // extension defined
  49         const LOGS = 8;
  50         const RANGE = 16;
  51
  52         const BUFFER = 0;
  53         const STREAM = 1;
  54
  55         const TEXT = 0;
  56         const STUB = 1;
  57
  58         /** @var int */
  59         protected $buffer;
  60
  61         /** @var int */
  62         protected $text;
  63
  64         /** @var DumpOutput */
  65         protected $sink;
  66
  67         /**
  68          * Returns the export schema version.
  69          * @return string
  70          */
  71         public static function schemaVersion() {
  72                 return "0.8";
  73         }
  74
  75         /**
  76          * If using WikiExporter::STREAM to stream a large amount of data,
  77          * provide a database connection which is not managed by
  78          * LoadBalancer to read from: some history blob types will
  79          * make additional queries to pull source data while the
  80          * main query is still running.
  81          *
  82          * @param DatabaseBase $db
  83          * @param int|array $history One of WikiExporter::FULL, WikiExporter::CURRENT,
  84          *   WikiExporter::RANGE or WikiExporter::STABLE, or an associative array:
  85          *   - offset: non-inclusive offset at which to start the query
  86          *   - limit: maximum number of rows to return
  87          *   - dir: "asc" or "desc" timestamp order
  88          * @param int $buffer One of WikiExporter::BUFFER or WikiExporter::STREAM
  89          * @param int $text One of WikiExporter::TEXT or WikiExporter::STUB
  90          */
  91         function __construct( $db, $history = WikiExporter::CURRENT,
  92                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  93                 $this->db = $db;
  94                 $this->history = $history;
  95                 $this->buffer = $buffer;
  96                 $this->writer = new XmlDumpWriter();
  97                 $this->sink = new DumpOutput();
  98                 $this->text = $text;
  99         }
 100
 101         /**
 102          * Set the DumpOutput or DumpFilter object which will receive
 103          * various row objects and XML output for filtering. Filters
 104          * can be chained or used as callbacks.
 105          *
 106          * @param DumpOutput $sink
 107          */
 108         public function setOutputSink( &$sink ) {
 109                 $this->sink =& $sink;
 110         }
 111
 112         public function openStream() {
 113                 $output = $this->writer->openStream();
 114                 $this->sink->writeOpenStream( $output );
 115         }
 116
 117         public function closeStream() {
 118                 $output = $this->writer->closeStream();
 119                 $this->sink->writeCloseStream( $output );
 120         }
 121
 122         /**
 123          * Dumps a series of page and revision records for all pages
 124          * in the database, either including complete history or only
 125          * the most recent version.
 126          */
 127         public function allPages() {
 128                 $this->dumpFrom( '' );
 129         }
 130
 131         /**
 132          * Dumps a series of page and revision records for those pages
 133          * in the database falling within the page_id range given.
 134          * @param int $start Inclusive lower limit (this id is included)
 135          * @param int $end Exclusive upper limit (this id is not included)
 136          *   If 0, no upper limit.
 137          */
 138         public function pagesByRange( $start, $end ) {
 139                 $condition = 'page_id >= ' . intval( $start );
 140                 if ( $end ) {
 141                         $condition .= ' AND page_id < ' . intval( $end );
 142                 }
 143                 $this->dumpFrom( $condition );
 144         }
 145
 146         /**
 147          * Dumps a series of page and revision records for those pages
 148          * in the database with revisions falling within the rev_id range given.
 149          * @param int $start Inclusive lower limit (this id is included)
 150          * @param int $end Exclusive upper limit (this id is not included)
 151          *   If 0, no upper limit.
 152          */
 153         public function revsByRange( $start, $end ) {
 154                 $condition = 'rev_id >= ' . intval( $start );
 155                 if ( $end ) {
 156                         $condition .= ' AND rev_id < ' . intval( $end );
 157                 }
 158                 $this->dumpFrom( $condition );
 159         }
 160
 161         /**
 162          * @param Title $title
 163          */
 164         public function pageByTitle( $title ) {
 165                 $this->dumpFrom(
 166                         'page_namespace=' . $title->getNamespace() .
 167                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 168         }
 169
 170         /**
 171          * @param string $name
 172          * @throws MWException
 173          */
 174         public function pageByName( $name ) {
 175                 $title = Title::newFromText( $name );
 176                 if ( is_null( $title ) ) {
 177                         throw new MWException( "Can't export invalid title" );
 178                 } else {
 179                         $this->pageByTitle( $title );
 180                 }
 181         }
 182
 183         /**
 184          * @param array $names
 185          */
 186         public function pagesByName( $names ) {
 187                 foreach ( $names as $name ) {
 188                         $this->pageByName( $name );
 189                 }
 190         }
 191
 192         public function allLogs() {
 193                 $this->dumpFrom( '' );
 194         }
 195
 196         /**
 197          * @param int $start
 198          * @param int $end
 199          */
 200         public function logsByRange( $start, $end ) {
 201                 $condition = 'log_id >= ' . intval( $start );
 202                 if ( $end ) {
 203                         $condition .= ' AND log_id < ' . intval( $end );
 204                 }
 205                 $this->dumpFrom( $condition );
 206         }
 207
 208         /**
 209          * Generates the distinct list of authors of an article
 210          * Not called by default (depends on $this->list_authors)
 211          * Can be set by Special:Export when not exporting whole history
 212          *
 213          * @param array $cond
 214          */
 215         protected function do_list_authors( $cond ) {
 216                 wfProfileIn( __METHOD__ );
 217                 $this->author_list = "<contributors>";
 218                 // rev_deleted
 219
 220                 $res = $this->db->select(
 221                         array( 'page', 'revision' ),
 222                         array( 'DISTINCT rev_user_text', 'rev_user' ),
 223                         array(
 224                                 $this->db->bitAnd( 'rev_deleted', Revision::DELETED_USER ) . ' = 0',
 225                                 $cond,
 226                                 'page_id = rev_id',
 227                         ),
 228                         __METHOD__
 229                 );
 230
 231                 foreach ( $res as $row ) {
 232                         $this->author_list .= "<contributor>" .
 233                                 "<username>" .
 234                                 htmlentities( $row->rev_user_text ) .
 235                                 "</username>" .
 236                                 "<id>" .
 237                                 $row->rev_user .
 238                                 "</id>" .
 239                                 "</contributor>";
 240                 }
 241                 $this->author_list .= "</contributors>";
 242                 wfProfileOut( __METHOD__ );
 243         }
 244
 245         /**
 246          * @param string $cond
 247          * @throws MWException
 248          * @throws Exception
 249          */
 250         protected function dumpFrom( $cond = '' ) {
 251                 wfProfileIn( __METHOD__ );
 252                 # For logging dumps...
 253                 if ( $this->history & self::LOGS ) {
 254                         $where = array( 'user_id = log_user' );
 255                         # Hide private logs
 256                         $hideLogs = LogEventsList::getExcludeClause( $this->db );
 257                         if ( $hideLogs ) {
 258                                 $where[] = $hideLogs;
 259                         }
 260                         # Add on any caller specified conditions
 261                         if ( $cond ) {
 262                                 $where[] = $cond;
 263                         }
 264                         # Get logging table name for logging.* clause
 265                         $logging = $this->db->tableName( 'logging' );
 266
 267                         if ( $this->buffer == WikiExporter::STREAM ) {
 268                                 $prev = $this->db->bufferResults( false );
 269                         }
 270                         $wrapper = null; // Assuring $wrapper is not undefined, if exception occurs early
 271                         try {
 272                                 $result = $this->db->select( array( 'logging', 'user' ),
 273                                         array( "{$logging}.*", 'user_name' ), // grab the user name
 274                                         $where,
 275                                         __METHOD__,
 276                                         array( 'ORDER BY' => 'log_id', 'USE INDEX' => array( 'logging' => 'PRIMARY' ) )
 277                                 );
 278                                 $wrapper = $this->db->resultObject( $result );
 279                                 $this->outputLogStream( $wrapper );
 280                                 if ( $this->buffer == WikiExporter::STREAM ) {
 281                                         $this->db->bufferResults( $prev );
 282                                 }
 283                         } catch ( Exception $e ) {
 284                                 // Throwing the exception does not reliably free the resultset, and
 285                                 // would also leave the connection in unbuffered mode.
 286
 287                                 // Freeing result
 288                                 try {
 289                                         if ( $wrapper ) {
 290                                                 $wrapper->free();
 291                                         }
 292                                 } catch ( Exception $e2 ) {
 293                                         // Already in panic mode -> ignoring $e2 as $e has
 294                                         // higher priority
 295                                 }
 296
 297                                 // Putting database back in previous buffer mode
 298                                 try {
 299                                         if ( $this->buffer == WikiExporter::STREAM ) {
 300                                                 $this->db->bufferResults( $prev );
 301                                         }
 302                                 } catch ( Exception $e2 ) {
 303                                         // Already in panic mode -> ignoring $e2 as $e has
 304                                         // higher priority
 305                                 }
 306
 307                                 // Inform caller about problem
 308                                 wfProfileOut( __METHOD__ );
 309                                 throw $e;
 310                         }
 311                 # For page dumps...
 312                 } else {
 313                         $tables = array( 'page', 'revision' );
 314                         $opts = array( 'ORDER BY' => 'page_id ASC' );
 315                         $opts['USE INDEX'] = array();
 316                         $join = array();
 317                         if ( is_array( $this->history ) ) {
 318                                 # Time offset/limit for all pages/history...
 319                                 $revJoin = 'page_id=rev_page';
 320                                 # Set time order
 321                                 if ( $this->history['dir'] == 'asc' ) {
 322                                         $op = '>';
 323                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 324                                 } else {
 325                                         $op = '<';
 326                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 327                                 }
 328                                 # Set offset
 329                                 if ( !empty( $this->history['offset'] ) ) {
 330                                         $revJoin .= " AND rev_timestamp $op " .
 331                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 332                                 }
 333                                 $join['revision'] = array( 'INNER JOIN', $revJoin );
 334                                 # Set query limit
 335                                 if ( !empty( $this->history['limit'] ) ) {
 336                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 337                                 }
 338                         } elseif ( $this->history & WikiExporter::FULL ) {
 339                                 # Full history dumps...
 340                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 341                         } elseif ( $this->history & WikiExporter::CURRENT ) {
 342                                 # Latest revision dumps...
 343                                 if ( $this->list_authors && $cond != '' ) { // List authors, if so desired
 344                                         $this->do_list_authors( $cond );
 345                                 }
 346                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 347                         } elseif ( $this->history & WikiExporter::STABLE ) {
 348                                 # "Stable" revision dumps...
 349                                 # Default JOIN, to be overridden...
 350                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 351                                 # One, and only one hook should set this, and return false
 352                                 if ( wfRunHooks( 'WikiExporter::dumpStableQuery', array( &$tables, &$opts, &$join ) ) ) {
 353                                         wfProfileOut( __METHOD__ );
 354                                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 355                                 }
 356                         } elseif ( $this->history & WikiExporter::RANGE ) {
 357                                 # Dump of revisions within a specified range
 358                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 359                                 $opts['ORDER BY'] = array( 'rev_page ASC', 'rev_id ASC' );
 360                         } else {
 361                                 # Unknown history specification parameter?
 362                                 wfProfileOut( __METHOD__ );
 363                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 364                         }
 365                         # Query optimization hacks
 366                         if ( $cond == '' ) {
 367                                 $opts[] = 'STRAIGHT_JOIN';
 368                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 369                         }
 370                         # Build text join options
 371                         if ( $this->text != WikiExporter::STUB ) { // 1-pass
 372                                 $tables[] = 'text';
 373                                 $join['text'] = array( 'INNER JOIN', 'rev_text_id=old_id' );
 374                         }
 375
 376                         if ( $this->buffer == WikiExporter::STREAM ) {
 377                                 $prev = $this->db->bufferResults( false );
 378                         }
 379
 380                         $wrapper = null; // Assuring $wrapper is not undefined, if exception occurs early
 381                         try {
 382                                 wfRunHooks( 'ModifyExportQuery',
 383                                                 array( $this->db, &$tables, &$cond, &$opts, &$join ) );
 384
 385                                 # Do the query!
 386                                 $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 387                                 $wrapper = $this->db->resultObject( $result );
 388                                 # Output dump results
 389                                 $this->outputPageStream( $wrapper );
 390
 391                                 if ( $this->buffer == WikiExporter::STREAM ) {
 392                                         $this->db->bufferResults( $prev );
 393                                 }
 394                         } catch ( Exception $e ) {
 395                                 // Throwing the exception does not reliably free the resultset, and
 396                                 // would also leave the connection in unbuffered mode.
 397
 398                                 // Freeing result
 399                                 try {
 400                                         if ( $wrapper ) {
 401                                                 $wrapper->free();
 402                                         }
 403                                 } catch ( Exception $e2 ) {
 404                                         // Already in panic mode -> ignoring $e2 as $e has
 405                                         // higher priority
 406                                 }
 407
 408                                 // Putting database back in previous buffer mode
 409                                 try {
 410                                         if ( $this->buffer == WikiExporter::STREAM ) {
 411                                                 $this->db->bufferResults( $prev );
 412                                         }
 413                                 } catch ( Exception $e2 ) {
 414                                         // Already in panic mode -> ignoring $e2 as $e has
 415                                         // higher priority
 416                                 }
 417
 418                                 // Inform caller about problem
 419                                 throw $e;
 420                         }
 421                 }
 422                 wfProfileOut( __METHOD__ );
 423         }
 424
 425         /**
 426          * Runs through a query result set dumping page and revision records.
 427          * The result set should be sorted/grouped by page to avoid duplicate
 428          * page records in the output.
 429          *
 430          * Should be safe for
 431          * streaming (non-buffered) queries, as long as it was made on a
 432          * separate database connection not managed by LoadBalancer; some
 433          * blob storage types will make queries to pull source data.
 434          *
 435          * @param ResultWrapper $resultset
 436          */
 437         protected function outputPageStream( $resultset ) {
 438                 $last = null;
 439                 foreach ( $resultset as $row ) {
 440                         if ( $last === null ||
 441                                 $last->page_namespace != $row->page_namespace ||
 442                                 $last->page_title != $row->page_title ) {
 443                                 if ( $last !== null ) {
 444                                         $output = '';
 445                                         if ( $this->dumpUploads ) {
 446                                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 447                                         }
 448                                         $output .= $this->writer->closePage();
 449                                         $this->sink->writeClosePage( $output );
 450                                 }
 451                                 $output = $this->writer->openPage( $row );
 452                                 $this->sink->writeOpenPage( $row, $output );
 453                                 $last = $row;
 454                         }
 455                         $output = $this->writer->writeRevision( $row );
 456                         $this->sink->writeRevision( $row, $output );
 457                 }
 458                 if ( $last !== null ) {
 459                         $output = '';
 460                         if ( $this->dumpUploads ) {
 461                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 462                         }
 463                         $output .= $this->author_list;
 464                         $output .= $this->writer->closePage();
 465                         $this->sink->writeClosePage( $output );
 466                 }
 467         }
 468
 469         /**
 470          * @param array $resultset
 471          */
 472         protected function outputLogStream( $resultset ) {
 473                 foreach ( $resultset as $row ) {
 474                         $output = $this->writer->writeLogItem( $row );
 475                         $this->sink->writeLogItem( $row, $output );
 476                 }
 477         }
 478 }
 479
 480 /**
 481  * @ingroup Dump
 482  */
 483 class XmlDumpWriter {
 484         /**
 485          * Returns the export schema version.
 486          * @deprecated since 1.20; use WikiExporter::schemaVersion() instead
 487          * @return string
 488          */
 489         function schemaVersion() {
 490                 wfDeprecated( __METHOD__, '1.20' );
 491                 return WikiExporter::schemaVersion();
 492         }
 493
 494         /**
 495          * Opens the XML output stream's root "<mediawiki>" element.
 496          * This does not include an xml directive, so is safe to include
 497          * as a subelement in a larger XML stream. Namespace and XML Schema
 498          * references are included.
 499          *
 500          * Output will be encoded in UTF-8.
 501          *
 502          * @return string
 503          */
 504         function openStream() {
 505                 global $wgLanguageCode;
 506                 $ver = WikiExporter::schemaVersion();
 507                 return Xml::element( 'mediawiki', array(
 508                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 509                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 510                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 511                                 #TODO: how do we get a new version up there?
 512                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 513                         'version'            => $ver,
 514                         'xml:lang'           => $wgLanguageCode ),
 515                         null ) .
 516                         "\n" .
 517                         $this->siteInfo();
 518         }
 519
 520         /**
 521          * @return string
 522          */
 523         function siteInfo() {
 524                 $info = array(
 525                         $this->sitename(),
 526                         $this->homelink(),
 527                         $this->generator(),
 528                         $this->caseSetting(),
 529                         $this->namespaces() );
 530                 return "  <siteinfo>\n    " .
 531                         implode( "\n    ", $info ) .
 532                         "\n  </siteinfo>\n";
 533         }
 534
 535         /**
 536          * @return string
 537          */
 538         function sitename() {
 539                 global $wgSitename;
 540                 return Xml::element( 'sitename', array(), $wgSitename );
 541         }
 542
 543         /**
 544          * @return string
 545          */
 546         function generator() {
 547                 global $wgVersion;
 548                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 549         }
 550
 551         /**
 552          * @return string
 553          */
 554         function homelink() {
 555                 return Xml::element( 'base', array(), Title::newMainPage()->getCanonicalURL() );
 556         }
 557
 558         /**
 559          * @return string
 560          */
 561         function caseSetting() {
 562                 global $wgCapitalLinks;
 563                 // "case-insensitive" option is reserved for future
 564                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 565                 return Xml::element( 'case', array(), $sensitivity );
 566         }
 567
 568         /**
 569          * @return string
 570          */
 571         function namespaces() {
 572                 global $wgContLang;
 573                 $spaces = "<namespaces>\n";
 574                 foreach ( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 575                         $spaces .= '      ' .
 576                                 Xml::element( 'namespace',
 577                                         array(
 578                                                 'key' => $ns,
 579                                                 'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 580                                         ), $title ) . "\n";
 581                 }
 582                 $spaces .= "    </namespaces>";
 583                 return $spaces;
 584         }
 585
 586         /**
 587          * Closes the output stream with the closing root element.
 588          * Call when finished dumping things.
 589          *
 590          * @return string
 591          */
 592         function closeStream() {
 593                 return "</mediawiki>\n";
 594         }
 595
 596         /**
 597          * Opens a "<page>" section on the output stream, with data
 598          * from the given database row.
 599          *
 600          * @param object $row
 601          * @return string
 602          */
 603         public function openPage( $row ) {
 604                 $out = "  <page>\n";
 605                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 606                 $out .= '    ' . Xml::elementClean( 'title', array(), self::canonicalTitle( $title ) ) . "\n";
 607                 $out .= '    ' . Xml::element( 'ns', array(), strval( $row->page_namespace ) ) . "\n";
 608                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 609                 if ( $row->page_is_redirect ) {
 610                         $page = WikiPage::factory( $title );
 611                         $redirect = $page->getRedirectTarget();
 612                         if ( $redirect instanceof Title && $redirect->isValidRedirectTarget() ) {
 613                                 $out .= '    ';
 614                                 $out .= Xml::element( 'redirect', array( 'title' => self::canonicalTitle( $redirect ) ) );
 615                                 $out .= "\n";
 616                         }
 617                 }
 618
 619                 if ( $row->page_restrictions != '' ) {
 620                         $out .= '    ' . Xml::element( 'restrictions', array(),
 621                                 strval( $row->page_restrictions ) ) . "\n";
 622                 }
 623
 624                 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 625
 626                 return $out;
 627         }
 628
 629         /**
 630          * Closes a "<page>" section on the output stream.
 631          *
 632          * @access private
 633          * @return string
 634          */
 635         function closePage() {
 636                 return "  </page>\n";
 637         }
 638
 639         /**
 640          * Dumps a "<revision>" section on the output stream, with
 641          * data filled in from the given database row.
 642          *
 643          * @param object $row
 644          * @return string
 645          * @access private
 646          */
 647         function writeRevision( $row ) {
 648                 wfProfileIn( __METHOD__ );
 649
 650                 $out = "    <revision>\n";
 651                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 652                 if ( isset( $row->rev_parent_id ) && $row->rev_parent_id ) {
 653                         $out .= "      " . Xml::element( 'parentid', null, strval( $row->rev_parent_id ) ) . "\n";
 654                 }
 655
 656                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 657
 658                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_USER ) ) {
 659                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 660                 } else {
 661                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 662                 }
 663
 664                 if ( isset( $row->rev_minor_edit ) && $row->rev_minor_edit ) {
 665                         $out .= "      <minor/>\n";
 666                 }
 667                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_COMMENT ) ) {
 668                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 669                 } elseif ( $row->rev_comment != '' ) {
 670                         $out .= "      " . Xml::elementClean( 'comment', array(), strval( $row->rev_comment ) ) . "\n";
 671                 }
 672
 673                 $text = '';
 674                 if ( isset( $row->rev_deleted ) && ( $row->rev_deleted & Revision::DELETED_TEXT ) ) {
 675                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 676                 } elseif ( isset( $row->old_text ) ) {
 677                         // Raw text from the database may have invalid chars
 678                         $text = strval( Revision::getRevisionText( $row ) );
 679                         $out .= "      " . Xml::elementClean( 'text',
 680                                 array( 'xml:space' => 'preserve', 'bytes' => intval( $row->rev_len ) ),
 681                                 strval( $text ) ) . "\n";
 682                 } else {
 683                         // Stub output
 684                         $out .= "      " . Xml::element( 'text',
 685                                 array( 'id' => $row->rev_text_id, 'bytes' => intval( $row->rev_len ) ),
 686                                 "" ) . "\n";
 687                 }
 688
 689                 if ( isset( $row->rev_sha1 )
 690                         && $row->rev_sha1
 691                         && !( $row->rev_deleted & Revision::DELETED_TEXT )
 692                 ) {
 693                         $out .= "      " . Xml::element( 'sha1', null, strval( $row->rev_sha1 ) ) . "\n";
 694                 } else {
 695                         $out .= "      <sha1/>\n";
 696                 }
 697
 698                 if ( isset( $row->rev_content_model ) && !is_null( $row->rev_content_model ) ) {
 699                         $content_model = strval( $row->rev_content_model );
 700                 } else {
 701                         // probably using $wgContentHandlerUseDB = false;
 702                         // @todo test!
 703                         $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 704                         $content_model = ContentHandler::getDefaultModelFor( $title );
 705                 }
 706
 707                 $out .= "      " . Xml::element( 'model', null, strval( $content_model ) ) . "\n";
 708
 709                 if ( isset( $row->rev_content_format ) && !is_null( $row->rev_content_format ) ) {
 710                         $content_format = strval( $row->rev_content_format );
 711                 } else {
 712                         // probably using $wgContentHandlerUseDB = false;
 713                         // @todo test!
 714                         $content_handler = ContentHandler::getForModelID( $content_model );
 715                         $content_format = $content_handler->getDefaultFormat();
 716                 }
 717
 718                 $out .= "      " . Xml::element( 'format', null, strval( $content_format ) ) . "\n";
 719
 720                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 721
 722                 $out .= "    </revision>\n";
 723
 724                 wfProfileOut( __METHOD__ );
 725                 return $out;
 726         }
 727
 728         /**
 729          * Dumps a "<logitem>" section on the output stream, with
 730          * data filled in from the given database row.
 731          *
 732          * @param object $row
 733          * @return string
 734          * @access private
 735          */
 736         function writeLogItem( $row ) {
 737                 wfProfileIn( __METHOD__ );
 738
 739                 $out = "  <logitem>\n";
 740                 $out .= "    " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 741
 742                 $out .= $this->writeTimestamp( $row->log_timestamp, "    " );
 743
 744                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 745                         $out .= "    " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 746                 } else {
 747                         $out .= $this->writeContributor( $row->log_user, $row->user_name, "    " );
 748                 }
 749
 750                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 751                         $out .= "    " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 752                 } elseif ( $row->log_comment != '' ) {
 753                         $out .= "    " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 754                 }
 755
 756                 $out .= "    " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 757                 $out .= "    " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 758
 759                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 760                         $out .= "    " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 761                 } else {
 762                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 763                         $out .= "    " . Xml::elementClean( 'logtitle', null, self::canonicalTitle( $title ) ) . "\n";
 764                         $out .= "    " . Xml::elementClean( 'params',
 765                                 array( 'xml:space' => 'preserve' ),
 766                                 strval( $row->log_params ) ) . "\n";
 767                 }
 768
 769                 $out .= "  </logitem>\n";
 770
 771                 wfProfileOut( __METHOD__ );
 772                 return $out;
 773         }
 774
 775         /**
 776          * @param string $timestamp
 777          * @param string $indent Default to six spaces
 778          * @return string
 779          */
 780         function writeTimestamp( $timestamp, $indent = "      " ) {
 781                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 782                 return $indent . Xml::element( 'timestamp', null, $ts ) . "\n";
 783         }
 784
 785         /**
 786          * @param int $id
 787          * @param string $text
 788          * @param string $indent Default to six spaces
 789          * @return string
 790          */
 791         function writeContributor( $id, $text, $indent = "      " ) {
 792                 $out = $indent . "<contributor>\n";
 793                 if ( $id || !IP::isValid( $text ) ) {
 794                         $out .= $indent . "  " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 795                         $out .= $indent . "  " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 796                 } else {
 797                         $out .= $indent . "  " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 798                 }
 799                 $out .= $indent . "</contributor>\n";
 800                 return $out;
 801         }
 802
 803         /**
 804          * Warning! This data is potentially inconsistent. :(
 805          * @param object $row
 806          * @param bool $dumpContents
 807          * @return string
 808          */
 809         function writeUploads( $row, $dumpContents = false ) {
 810                 if ( $row->page_namespace == NS_FILE ) {
 811                         $img = wfLocalFile( $row->page_title );
 812                         if ( $img && $img->exists() ) {
 813                                 $out = '';
 814                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 815                                         $out .= $this->writeUpload( $ver, $dumpContents );
 816                                 }
 817                                 $out .= $this->writeUpload( $img, $dumpContents );
 818                                 return $out;
 819                         }
 820                 }
 821                 return '';
 822         }
 823
 824         /**
 825          * @param File $file
 826          * @param bool $dumpContents
 827          * @return string
 828          */
 829         function writeUpload( $file, $dumpContents = false ) {
 830                 if ( $file->isOld() ) {
 831                         $archiveName = "      " .
 832                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 833                 } else {
 834                         $archiveName = '';
 835                 }
 836                 if ( $dumpContents ) {
 837                         $be = $file->getRepo()->getBackend();
 838                         # Dump file as base64
 839                         # Uses only XML-safe characters, so does not need escaping
 840                         # @todo Too bad this loads the contents into memory (script might swap)
 841                         $contents = '      <contents encoding="base64">' .
 842                                 chunk_split( base64_encode(
 843                                         $be->getFileContents( array( 'src' => $file->getPath() ) ) ) ) .
 844                                 "      </contents>\n";
 845                 } else {
 846                         $contents = '';
 847                 }
 848                 if ( $file->isDeleted( File::DELETED_COMMENT ) ) {
 849                         $comment = Xml::element( 'comment', array( 'deleted' => 'deleted' ) );
 850                 } else {
 851                         $comment = Xml::elementClean( 'comment', null, $file->getDescription() );
 852                 }
 853                 return "    <upload>\n" .
 854                         $this->writeTimestamp( $file->getTimestamp() ) .
 855                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 856                         "      " . $comment . "\n" .
 857                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 858                         $archiveName .
 859                         "      " . Xml::element( 'src', null, $file->getCanonicalURL() ) . "\n" .
 860                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 861                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 862                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 863                         $contents .
 864                         "    </upload>\n";
 865         }
 866
 867         /**
 868          * Return prefixed text form of title, but using the content language's
 869          * canonical namespace. This skips any special-casing such as gendered
 870          * user namespaces -- which while useful, are not yet listed in the
 871          * XML "<siteinfo>" data so are unsafe in export.
 872          *
 873          * @param Title $title
 874          * @return string
 875          * @since 1.18
 876          */
 877         public static function canonicalTitle( Title $title ) {
 878                 if ( $title->isExternal() ) {
 879                         return $title->getPrefixedText();
 880                 }
 881
 882                 global $wgContLang;
 883                 $prefix = str_replace( '_', ' ', $wgContLang->getNsText( $title->getNamespace() ) );
 884
 885                 if ( $prefix !== '' ) {
 886                         $prefix .= ':';
 887                 }
 888
 889                 return $prefix . $title->getText();
 890         }
 891 }
 892
 893 /**
 894  * Base class for output stream; prints to stdout or buffer or wherever.
 895  * @ingroup Dump
 896  */
 897 class DumpOutput {
 898
 899         /**
 900          * @param string $string
 901          */
 902         function writeOpenStream( $string ) {
 903                 $this->write( $string );
 904         }
 905
 906         /**
 907          * @param string $string
 908          */
 909         function writeCloseStream( $string ) {
 910                 $this->write( $string );
 911         }
 912
 913         /**
 914          * @param object $page
 915          * @param string $string
 916          */
 917         function writeOpenPage( $page, $string ) {
 918                 $this->write( $string );
 919         }
 920
 921         /**
 922          * @param string $string
 923          */
 924         function writeClosePage( $string ) {
 925                 $this->write( $string );
 926         }
 927
 928         /**
 929          * @param object $rev
 930          * @param string $string
 931          */
 932         function writeRevision( $rev, $string ) {
 933                 $this->write( $string );
 934         }
 935
 936         /**
 937          * @param object $rev
 938          * @param string $string
 939          */
 940         function writeLogItem( $rev, $string ) {
 941                 $this->write( $string );
 942         }
 943
 944         /**
 945          * Override to write to a different stream type.
 946          * @param string $string
 947          * @return bool
 948          */
 949         function write( $string ) {
 950                 print $string;
 951         }
 952
 953         /**
 954          * Close the old file, move it to a specified name,
 955          * and reopen new file with the old name. Use this
 956          * for writing out a file in multiple pieces
 957          * at specified checkpoints (e.g. every n hours).
 958          * @param string|array $newname File name. May be a string or an array with one element
 959          */
 960         function closeRenameAndReopen( $newname ) {
 961         }
 962
 963         /**
 964          * Close the old file, and move it to a specified name.
 965          * Use this for the last piece of a file written out
 966          * at specified checkpoints (e.g. every n hours).
 967          * @param string|array $newname File name. May be a string or an array with one element
 968          * @param bool $open If true, a new file with the old filename will be opened
 969          *   again for writing (default: false)
 970          */
 971         function closeAndRename( $newname, $open = false ) {
 972         }
 973
 974         /**
 975          * Returns the name of the file or files which are
 976          * being written to, if there are any.
 977          * @return null
 978          */
 979         function getFilenames() {
 980                 return null;
 981         }
 982 }
 983
 984 /**
 985  * Stream outputter to send data to a file.
 986  * @ingroup Dump
 987  */
 988 class DumpFileOutput extends DumpOutput {
 989         protected $handle = false, $filename;
 990
 991         /**
 992          * @param string $file
 993          */
 994         function __construct( $file ) {
 995                 $this->handle = fopen( $file, "wt" );
 996                 $this->filename = $file;
 997         }
 998
 999         /**
1000          * @param string $string
1001          */
1002         function writeCloseStream( $string ) {
1003                 parent::writeCloseStream( $string );
1004                 if ( $this->handle ) {
1005                         fclose( $this->handle );
1006                         $this->handle = false;
1007                 }
1008         }
1009
1010         /**
1011          * @param string $string
1012          */
1013         function write( $string ) {
1014                 fputs( $this->handle, $string );
1015         }
1016
1017         /**
1018          * @param string $newname
1019          */
1020         function closeRenameAndReopen( $newname ) {
1021                 $this->closeAndRename( $newname, true );
1022         }
1023
1024         /**
1025          * @param string $newname
1026          * @throws MWException
1027          */
1028         function renameOrException( $newname ) {
1029                         if ( !rename( $this->filename, $newname ) ) {
1030                                 throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" );
1031                         }
1032         }
1033
1034         /**
1035          * @param array $newname
1036          * @return string
1037          * @throws MWException
1038          */
1039         function checkRenameArgCount( $newname ) {
1040                 if ( is_array( $newname ) ) {
1041                         if ( count( $newname ) > 1 ) {
1042                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
1043                         } else {
1044                                 $newname = $newname[0];
1045                         }
1046                 }
1047                 return $newname;
1048         }
1049
1050         /**
1051          * @param string $newname
1052          * @param bool $open
1053          */
1054         function closeAndRename( $newname, $open = false ) {
1055                 $newname = $this->checkRenameArgCount( $newname );
1056                 if ( $newname ) {
1057                         if ( $this->handle ) {
1058                                 fclose( $this->handle );
1059                                 $this->handle = false;
1060                         }
1061                         $this->renameOrException( $newname );
1062                         if ( $open ) {
1063                                 $this->handle = fopen( $this->filename, "wt" );
1064                         }
1065                 }
1066         }
1067
1068         /**
1069          * @return string|null
1070          */
1071         function getFilenames() {
1072                 return $this->filename;
1073         }
1074 }
1075
1076 /**
1077  * Stream outputter to send data to a file via some filter program.
1078  * Even if compression is available in a library, using a separate
1079  * program can allow us to make use of a multi-processor system.
1080  * @ingroup Dump
1081  */
1082 class DumpPipeOutput extends DumpFileOutput {
1083         protected $command, $filename;
1084         protected $procOpenResource = false;
1085
1086         /**
1087          * @param string $command
1088          * @param string $file
1089          */
1090         function __construct( $command, $file = null ) {
1091                 if ( !is_null( $file ) ) {
1092                         $command .= " > " . wfEscapeShellArg( $file );
1093                 }
1094
1095                 $this->startCommand( $command );
1096                 $this->command = $command;
1097                 $this->filename = $file;
1098         }
1099
1100         /**
1101          * @param string $string
1102          */
1103         function writeCloseStream( $string ) {
1104                 parent::writeCloseStream( $string );
1105                 if ( $this->procOpenResource ) {
1106                         proc_close( $this->procOpenResource );
1107                         $this->procOpenResource = false;
1108                 }
1109         }
1110
1111         /**
1112          * @param string $command
1113          */
1114         function startCommand( $command ) {
1115                 $spec = array(
1116                         0 => array( "pipe", "r" ),
1117                 );
1118                 $pipes = array();
1119                 $this->procOpenResource = proc_open( $command, $spec, $pipes );
1120                 $this->handle = $pipes[0];
1121         }
1122
1123         /**
1124          * @param string $newname
1125          */
1126         function closeRenameAndReopen( $newname ) {
1127                 $this->closeAndRename( $newname, true );
1128         }
1129
1130         /**
1131          * @param string $newname
1132          * @param bool $open
1133          */
1134         function closeAndRename( $newname, $open = false ) {
1135                 $newname = $this->checkRenameArgCount( $newname );
1136                 if ( $newname ) {
1137                         if ( $this->handle ) {
1138                                 fclose( $this->handle );
1139                                 $this->handle = false;
1140                         }
1141                         if ( $this->procOpenResource ) {
1142                                 proc_close( $this->procOpenResource );
1143                                 $this->procOpenResource = false;
1144                         }
1145                         $this->renameOrException( $newname );
1146                         if ( $open ) {
1147                                 $command = $this->command;
1148                                 $command .= " > " . wfEscapeShellArg( $this->filename );
1149                                 $this->startCommand( $command );
1150                         }
1151                 }
1152         }
1153 }
1154
1155 /**
1156  * Sends dump output via the gzip compressor.
1157  * @ingroup Dump
1158  */
1159 class DumpGZipOutput extends DumpPipeOutput {
1160         /**
1161          * @param string $file
1162          */
1163         function __construct( $file ) {
1164                 parent::__construct( "gzip", $file );
1165         }
1166 }
1167
1168 /**
1169  * Sends dump output via the bgzip2 compressor.
1170  * @ingroup Dump
1171  */
1172 class DumpBZip2Output extends DumpPipeOutput {
1173         /**
1174          * @param string $file
1175          */
1176         function __construct( $file ) {
1177                 parent::__construct( "bzip2", $file );
1178         }
1179 }
1180
1181 /**
1182  * Sends dump output via the p7zip compressor.
1183  * @ingroup Dump
1184  */
1185 class Dump7ZipOutput extends DumpPipeOutput {
1186         /**
1187          * @param string $file
1188          */
1189         function __construct( $file ) {
1190                 $command = $this->setup7zCommand( $file );
1191                 parent::__construct( $command );
1192                 $this->filename = $file;
1193         }
1194
1195         /**
1196          * @param string $file
1197          * @return string
1198          */
1199         function setup7zCommand( $file ) {
1200                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
1201                 // Suppress annoying useless crap from p7zip
1202                 // Unfortunately this could suppress real error messages too
1203                 $command .= ' >' . wfGetNull() . ' 2>&1';
1204                 return $command;
1205         }
1206
1207         /**
1208          * @param string $newname
1209          * @param bool $open
1210          */
1211         function closeAndRename( $newname, $open = false ) {
1212                 $newname = $this->checkRenameArgCount( $newname );
1213                 if ( $newname ) {
1214                         fclose( $this->handle );
1215                         proc_close( $this->procOpenResource );
1216                         $this->renameOrException( $newname );
1217                         if ( $open ) {
1218                                 $command = $this->setup7zCommand( $this->filename );
1219                                 $this->startCommand( $command );
1220                         }
1221                 }
1222         }
1223 }
1224
1225 /**
1226  * Dump output filter class.
1227  * This just does output filtering and streaming; XML formatting is done
1228  * higher up, so be careful in what you do.
1229  * @ingroup Dump
1230  */
1231 class DumpFilter {
1232         /**
1233          * @var DumpOutput
1234          * FIXME will need to be made protected whenever legacy code
1235          * is updated.
1236          */
1237         public $sink;
1238
1239         /**
1240          * @var bool
1241          */
1242         protected $sendingThisPage;
1243
1244         /**
1245          * @param DumpOutput $sink
1246          */
1247         function __construct( &$sink ) {
1248                 $this->sink =& $sink;
1249         }
1250
1251         /**
1252          * @param string $string
1253          */
1254         function writeOpenStream( $string ) {
1255                 $this->sink->writeOpenStream( $string );
1256         }
1257
1258         /**
1259          * @param string $string
1260          */
1261         function writeCloseStream( $string ) {
1262                 $this->sink->writeCloseStream( $string );
1263         }
1264
1265         /**
1266          * @param object $page
1267          * @param string $string
1268          */
1269         function writeOpenPage( $page, $string ) {
1270                 $this->sendingThisPage = $this->pass( $page, $string );
1271                 if ( $this->sendingThisPage ) {
1272                         $this->sink->writeOpenPage( $page, $string );
1273                 }
1274         }
1275
1276         /**
1277          * @param string $string
1278          */
1279         function writeClosePage( $string ) {
1280                 if ( $this->sendingThisPage ) {
1281                         $this->sink->writeClosePage( $string );
1282                         $this->sendingThisPage = false;
1283                 }
1284         }
1285
1286         /**
1287          * @param object $rev
1288          * @param string $string
1289          */
1290         function writeRevision( $rev, $string ) {
1291                 if ( $this->sendingThisPage ) {
1292                         $this->sink->writeRevision( $rev, $string );
1293                 }
1294         }
1295
1296         /**
1297          * @param object $rev
1298          * @param string $string
1299          */
1300         function writeLogItem( $rev, $string ) {
1301                 $this->sink->writeRevision( $rev, $string );
1302         }
1303
1304         /**
1305          * @param string $newname
1306          */
1307         function closeRenameAndReopen( $newname ) {
1308                 $this->sink->closeRenameAndReopen( $newname );
1309         }
1310
1311         /**
1312          * @param string $newname
1313          * @param bool $open
1314          */
1315         function closeAndRename( $newname, $open = false ) {
1316                 $this->sink->closeAndRename( $newname, $open );
1317         }
1318
1319         /**
1320          * @return array
1321          */
1322         function getFilenames() {
1323                 return $this->sink->getFilenames();
1324         }
1325
1326         /**
1327          * Override for page-based filter types.
1328          * @param object $page
1329          * @return bool
1330          */
1331         function pass( $page ) {
1332                 return true;
1333         }
1334 }
1335
1336 /**
1337  * Simple dump output filter to exclude all talk pages.
1338  * @ingroup Dump
1339  */
1340 class DumpNotalkFilter extends DumpFilter {
1341         /**
1342          * @param object $page
1343          * @return bool
1344          */
1345         function pass( $page ) {
1346                 return !MWNamespace::isTalk( $page->page_namespace );
1347         }
1348 }
1349
1350 /**
1351  * Dump output filter to include or exclude pages in a given set of namespaces.
1352  * @ingroup Dump
1353  */
1354 class DumpNamespaceFilter extends DumpFilter {
1355         /** @var bool */
1356         protected $invert = false;
1357
1358         /** @var array */
1359         protected $namespaces = array();
1360
1361         /**
1362          * @param DumpOutput $sink
1363          * @param array $param
1364          * @throws MWException
1365          */
1366         function __construct( &$sink, $param ) {
1367                 parent::__construct( $sink );
1368
1369                 $constants = array(
1370                         "NS_MAIN"           => NS_MAIN,
1371                         "NS_TALK"           => NS_TALK,
1372                         "NS_USER"           => NS_USER,
1373                         "NS_USER_TALK"      => NS_USER_TALK,
1374                         "NS_PROJECT"        => NS_PROJECT,
1375                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
1376                         "NS_FILE"           => NS_FILE,
1377                         "NS_FILE_TALK"      => NS_FILE_TALK,
1378                         "NS_IMAGE"          => NS_IMAGE, // NS_IMAGE is an alias for NS_FILE
1379                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
1380                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
1381                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
1382                         "NS_TEMPLATE"       => NS_TEMPLATE,
1383                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
1384                         "NS_HELP"           => NS_HELP,
1385                         "NS_HELP_TALK"      => NS_HELP_TALK,
1386                         "NS_CATEGORY"       => NS_CATEGORY,
1387                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
1388
1389                 if ( $param { 0 } == '!' ) {
1390                         $this->invert = true;
1391                         $param = substr( $param, 1 );
1392                 }
1393
1394                 foreach ( explode( ',', $param ) as $key ) {
1395                         $key = trim( $key );
1396                         if ( isset( $constants[$key] ) ) {
1397                                 $ns = $constants[$key];
1398                                 $this->namespaces[$ns] = true;
1399                         } elseif ( is_numeric( $key ) ) {
1400                                 $ns = intval( $key );
1401                                 $this->namespaces[$ns] = true;
1402                         } else {
1403                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
1404                         }
1405                 }
1406         }
1407
1408         /**
1409          * @param object $page
1410          * @return bool
1411          */
1412         function pass( $page ) {
1413                 $match = isset( $this->namespaces[$page->page_namespace] );
1414                 return $this->invert xor $match;
1415         }
1416 }
1417
1418 /**
1419  * Dump output filter to include only the last revision in each page sequence.
1420  * @ingroup Dump
1421  */
1422 class DumpLatestFilter extends DumpFilter {
1423         protected $page;
1424
1425         protected $pageString;
1426
1427         protected $rev;
1428
1429         protected $revString;
1430
1431         /**
1432          * @param object $page
1433          * @param string $string
1434          */
1435         function writeOpenPage( $page, $string ) {
1436                 $this->page = $page;
1437                 $this->pageString = $string;
1438         }
1439
1440         /**
1441          * @param string $string
1442          */
1443         function writeClosePage( $string ) {
1444                 if ( $this->rev ) {
1445                         $this->sink->writeOpenPage( $this->page, $this->pageString );
1446                         $this->sink->writeRevision( $this->rev, $this->revString );
1447                         $this->sink->writeClosePage( $string );
1448                 }
1449                 $this->rev = null;
1450                 $this->revString = null;
1451                 $this->page = null;
1452                 $this->pageString = null;
1453         }
1454
1455         /**
1456          * @param object $rev
1457          * @param string $string
1458          */
1459         function writeRevision( $rev, $string ) {
1460                 if ( $rev->rev_id == $this->page->page_latest ) {
1461                         $this->rev = $rev;
1462                         $this->revString = $string;
1463                 }
1464         }
1465 }
1466
1467 /**
1468  * Base class for output stream; prints to stdout or buffer or wherever.
1469  * @ingroup Dump
1470  */
1471 class DumpMultiWriter {
1472
1473         /**
1474          * @param array $sinks
1475          */
1476         function __construct( $sinks ) {
1477                 $this->sinks = $sinks;
1478                 $this->count = count( $sinks );
1479         }
1480
1481         /**
1482          * @param string $string
1483          */
1484         function writeOpenStream( $string ) {
1485                 for ( $i = 0; $i < $this->count; $i++ ) {
1486                         $this->sinks[$i]->writeOpenStream( $string );
1487                 }
1488         }
1489
1490         /**
1491          * @param string $string
1492          */
1493         function writeCloseStream( $string ) {
1494                 for ( $i = 0; $i < $this->count; $i++ ) {
1495                         $this->sinks[$i]->writeCloseStream( $string );
1496                 }
1497         }
1498
1499         /**
1500          * @param object $page
1501          * @param string $string
1502          */
1503         function writeOpenPage( $page, $string ) {
1504                 for ( $i = 0; $i < $this->count; $i++ ) {
1505                         $this->sinks[$i]->writeOpenPage( $page, $string );
1506                 }
1507         }
1508
1509         /**
1510          * @param string $string
1511          */
1512         function writeClosePage( $string ) {
1513                 for ( $i = 0; $i < $this->count; $i++ ) {
1514                         $this->sinks[$i]->writeClosePage( $string );
1515                 }
1516         }
1517
1518         /**
1519          * @param object $rev
1520          * @param string $string
1521          */
1522         function writeRevision( $rev, $string ) {
1523                 for ( $i = 0; $i < $this->count; $i++ ) {
1524                         $this->sinks[$i]->writeRevision( $rev, $string );
1525                 }
1526         }
1527
1528         /**
1529          * @param array $newnames
1530          */
1531         function closeRenameAndReopen( $newnames ) {
1532                 $this->closeAndRename( $newnames, true );
1533         }
1534
1535         /**
1536          * @param array $newnames
1537          * @param bool $open
1538          */
1539         function closeAndRename( $newnames, $open = false ) {
1540                 for ( $i = 0; $i < $this->count; $i++ ) {
1541                         $this->sinks[$i]->closeAndRename( $newnames[$i], $open );
1542                 }
1543         }
1544
1545         /**
1546          * @return array
1547          */
1548         function getFilenames() {
1549                 $filenames = array();
1550                 for ( $i = 0; $i < $this->count; $i++ ) {
1551                         $filenames[] = $this->sinks[$i]->getFilenames();
1552                 }
1553                 return $filenames;
1554         }
1555 }
1556
1557 /**
1558  * @param string $string
1559  * @return string
1560  * @todo FIXME: Only used in OAI extension. Move over there.
1561  */
1562 function xmlsafe( $string ) {
1563         wfProfileIn( __FUNCTION__ );
1564
1565         /**
1566          * The page may contain old data which has not been properly normalized.
1567          * Invalid UTF-8 sequences or forbidden control characters will make our
1568          * XML output invalid, so be sure to strip them out.
1569          */
1570         $string = UtfNormal::cleanUp( $string );
1571
1572         $string = htmlspecialchars( $string );
1573         wfProfileOut( __FUNCTION__ );
1574         return $string;
1575 }