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