includes/Export.php

   1 <?php
   2 # Copyright (C) 2003, 2005, 2006 Brion Vibber <brion@pobox.com>
   3 # http://www.mediawiki.org/
   4 #
   5 # This program is free software; you can redistribute it and/or modify
   6 # it under the terms of the GNU General Public License as published by
   7 # the Free Software Foundation; either version 2 of the License, or
   8 # (at your option) any later version.
   9 #
  10 # This program is distributed in the hope that it will be useful,
  11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13 # GNU General Public License for more details.
  14 #
  15 # You should have received a copy of the GNU General Public License along
  16 # with this program; if not, write to the Free Software Foundation, Inc.,
  17 # 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18 # http://www.gnu.org/copyleft/gpl.html
  19
  20 /**
  21  *
  22  * @package MediaWiki
  23  * @subpackage SpecialPage
  24  */
  25
  26 class WikiExporter {
  27         var $list_authors = false ; # Return distinct author list (when not returning full history)
  28         var $author_list = "" ;
  29
  30         const FULL = 0;
  31         const CURRENT = 1;
  32
  33         const BUFFER = 0;
  34         const STREAM = 1;
  35
  36         const TEXT = 0;
  37         const STUB = 1;
  38
  39         /**
  40          * If using WikiExporter::STREAM to stream a large amount of data,
  41          * provide a database connection which is not managed by
  42          * LoadBalancer to read from: some history blob types will
  43          * make additional queries to pull source data while the
  44          * main query is still running.
  45          *
  46          * @param Database $db
  47          * @param mixed $history one of WikiExporter::FULL or WikiExporter::CURRENT, or an
  48          *                       associative array:
  49          *                         offset: non-inclusive offset at which to start the query
  50          *                         limit: maximum number of rows to return
  51          *                         dir: "asc" or "desc" timestamp order
  52          * @param int $buffer one of WikiExporter::BUFFER or WikiExporter::STREAM
  53          */
  54         function WikiExporter( &$db, $history = WikiExporter::CURRENT,
  55                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  56                 $this->db =& $db;
  57                 $this->history = $history;
  58                 $this->buffer  = $buffer;
  59                 $this->writer  = new XmlDumpWriter();
  60                 $this->sink    = new DumpOutput();
  61                 $this->text    = $text;
  62         }
  63
  64         /**
  65          * Set the DumpOutput or DumpFilter object which will receive
  66          * various row objects and XML output for filtering. Filters
  67          * can be chained or used as callbacks.
  68          *
  69          * @param mixed $callback
  70          */
  71         function setOutputSink( &$sink ) {
  72                 $this->sink =& $sink;
  73         }
  74
  75         function openStream() {
  76                 $output = $this->writer->openStream();
  77                 $this->sink->writeOpenStream( $output );
  78         }
  79
  80         function closeStream() {
  81                 $output = $this->writer->closeStream();
  82                 $this->sink->writeCloseStream( $output );
  83         }
  84
  85         /**
  86          * Dumps a series of page and revision records for all pages
  87          * in the database, either including complete history or only
  88          * the most recent version.
  89          */
  90         function allPages() {
  91                 return $this->dumpFrom( '' );
  92         }
  93
  94         /**
  95          * Dumps a series of page and revision records for those pages
  96          * in the database falling within the page_id range given.
  97          * @param int $start Inclusive lower limit (this id is included)
  98          * @param int $end   Exclusive upper limit (this id is not included)
  99          *                   If 0, no upper limit.
 100          */
 101         function pagesByRange( $start, $end ) {
 102                 $condition = 'page_id >= ' . intval( $start );
 103                 if( $end ) {
 104                         $condition .= ' AND page_id < ' . intval( $end );
 105                 }
 106                 return $this->dumpFrom( $condition );
 107         }
 108
 109         /**
 110          * @param Title $title
 111          */
 112         function pageByTitle( $title ) {
 113                 return $this->dumpFrom(
 114                         'page_namespace=' . $title->getNamespace() .
 115                         ' AND page_title=' . $this->db->addQuotes( $title->getDbKey() ) );
 116         }
 117
 118         function pageByName( $name ) {
 119                 $title = Title::newFromText( $name );
 120                 if( is_null( $title ) ) {
 121                         return new WikiError( "Can't export invalid title" );
 122                 } else {
 123                         return $this->pageByTitle( $title );
 124                 }
 125         }
 126
 127         function pagesByName( $names ) {
 128                 foreach( $names as $name ) {
 129                         $this->pageByName( $name );
 130                 }
 131         }
 132
 133
 134         // -------------------- private implementation below --------------------
 135
 136         # Generates the distinct list of authors of an article
 137         # Not called by default (depends on $this->list_authors)
 138         # Can be set by Special:Export when not exporting whole history
 139         function do_list_authors ( $page , $revision , $cond ) {
 140                 $fname = "do_list_authors" ;
 141                 wfProfileIn( $fname );
 142                 $this->author_list = "<contributors>";
 143                 $sql = "SELECT DISTINCT rev_user_text,rev_user FROM {$page},{$revision} WHERE page_id=rev_page AND " . $cond ;
 144                 $result = $this->db->query( $sql, $fname );
 145                 $resultset = $this->db->resultObject( $result );
 146                 while( $row = $resultset->fetchObject() ) {
 147                         $this->author_list .= "<contributor>" .
 148                                 "<username>" .
 149                                 htmlentities( $row->rev_user_text )  .
 150                                 "</username>" .
 151                                 "<id>" .
 152                                 $row->rev_user .
 153                                 "</id>" .
 154                                 "</contributor>";
 155                 }
 156                 wfProfileOut( $fname );
 157                 $this->author_list .= "</contributors>";
 158         }
 159
 160         function dumpFrom( $cond = '' ) {
 161                 $fname = 'WikiExporter::dumpFrom';
 162                 wfProfileIn( $fname );
 163
 164                 $page     = $this->db->tableName( 'page' );
 165                 $revision = $this->db->tableName( 'revision' );
 166                 $text     = $this->db->tableName( 'text' );
 167
 168                 $order = 'ORDER BY page_id';
 169                 $limit = '';
 170
 171                 if( $this->history == WikiExporter::FULL ) {
 172                         $join = 'page_id=rev_page';
 173                 } elseif( $this->history == WikiExporter::CURRENT ) {
 174                         if ( $this->list_authors && $cond != '' )  { // List authors, if so desired
 175                                 $this->do_list_authors ( $page , $revision , $cond );
 176                         }
 177                         $join = 'page_id=rev_page AND page_latest=rev_id';
 178                 } elseif ( is_array( $this->history ) ) {
 179                         $join = 'page_id=rev_page';
 180                         if ( $this->history['dir'] == 'asc' ) {
 181                                 $op = '>';
 182                                 $order .= ', rev_timestamp';
 183                         } else {
 184                                 $op = '<';
 185                                 $order .= ', rev_timestamp DESC';
 186                         }
 187                         if ( !empty( $this->history['offset'] ) ) {
 188                                 $join .= " AND rev_timestamp $op " . $this->db->addQuotes(
 189                                         $this->db->timestamp( $this->history['offset'] ) );
 190                         }
 191                         if ( !empty( $this->history['limit'] ) ) {
 192                                 $limitNum = intval( $this->history['limit'] );
 193                                 if ( $limitNum > 0 ) {
 194                                         $limit = "LIMIT $limitNum";
 195                                 }
 196                         }
 197                 } else {
 198                         wfProfileOut( $fname );
 199                         return new WikiError( "$fname given invalid history dump type." );
 200                 }
 201                 $where = ( $cond == '' ) ? '' : "$cond AND";
 202
 203                 if( $this->buffer == WikiExporter::STREAM ) {
 204                         $prev = $this->db->bufferResults( false );
 205                 }
 206                 if( $cond == '' ) {
 207                         // Optimization hack for full-database dump
 208                         $revindex = $pageindex = $this->db->useIndexClause("PRIMARY");
 209                         $straight = ' /*! STRAIGHT_JOIN */ ';
 210                 } else {
 211                         $pageindex = '';
 212                         $revindex = '';
 213                         $straight = '';
 214                 }
 215                 if( $this->text == WikiExporter::STUB ) {
 216                         $sql = "SELECT $straight * FROM
 217                                         $page $pageindex,
 218                                         $revision $revindex
 219                                         WHERE $where $join
 220                                         $order $limit";
 221                 } else {
 222                         $sql = "SELECT $straight * FROM
 223                                         $page $pageindex,
 224                                         $revision $revindex,
 225                                         $text
 226                                         WHERE $where $join AND rev_text_id=old_id
 227                                         $order $limit";
 228                 }
 229                 $result = $this->db->query( $sql, $fname );
 230                 $wrapper = $this->db->resultObject( $result );
 231                 $this->outputStream( $wrapper );
 232
 233                 if ( $this->list_authors ) {
 234                         $this->outputStream( $wrapper );
 235                 }
 236
 237                 if( $this->buffer == WikiExporter::STREAM ) {
 238                         $this->db->bufferResults( $prev );
 239                 }
 240
 241                 wfProfileOut( $fname );
 242         }
 243
 244         /**
 245          * Runs through a query result set dumping page and revision records.
 246          * The result set should be sorted/grouped by page to avoid duplicate
 247          * page records in the output.
 248          *
 249          * The result set will be freed once complete. Should be safe for
 250          * streaming (non-buffered) queries, as long as it was made on a
 251          * separate database connection not managed by LoadBalancer; some
 252          * blob storage types will make queries to pull source data.
 253          *
 254          * @param ResultWrapper $resultset
 255          * @access private
 256          */
 257         function outputStream( $resultset ) {
 258                 $last = null;
 259                 while( $row = $resultset->fetchObject() ) {
 260                         if( is_null( $last ) ||
 261                                 $last->page_namespace != $row->page_namespace ||
 262                                 $last->page_title     != $row->page_title ) {
 263                                 if( isset( $last ) ) {
 264                                         $output = $this->writer->closePage();
 265                                         $this->sink->writeClosePage( $output );
 266                                 }
 267                                 $output = $this->writer->openPage( $row );
 268                                 $this->sink->writeOpenPage( $row, $output );
 269                                 $last = $row;
 270                         }
 271                         $output = $this->writer->writeRevision( $row );
 272                         $this->sink->writeRevision( $row, $output );
 273                 }
 274                 if( isset( $last ) ) {
 275                         $output = $this->author_list . $this->writer->closePage();
 276                         $this->sink->writeClosePage( $output );
 277                 }
 278                 $resultset->free();
 279         }
 280 }
 281
 282 class XmlDumpWriter {
 283
 284         /**
 285          * Returns the export schema version.
 286          * @return string
 287          */
 288         function schemaVersion() {
 289                 return "0.3"; // FIXME: upgrade to 0.4 when updated XSD is ready, for the revision deletion bits
 290         }
 291
 292         /**
 293          * Opens the XML output stream's root <mediawiki> element.
 294          * This does not include an xml directive, so is safe to include
 295          * as a subelement in a larger XML stream. Namespace and XML Schema
 296          * references are included.
 297          *
 298          * Output will be encoded in UTF-8.
 299          *
 300          * @return string
 301          */
 302         function openStream() {
 303                 global $wgContLanguageCode;
 304                 $ver = $this->schemaVersion();
 305                 return wfElement( 'mediawiki', array(
 306                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 307                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 308                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 309                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 310                         'version'            => $ver,
 311                         'xml:lang'           => $wgContLanguageCode ),
 312                         null ) .
 313                         "\n" .
 314                         $this->siteInfo();
 315         }
 316
 317         function siteInfo() {
 318                 $info = array(
 319                         $this->sitename(),
 320                         $this->homelink(),
 321                         $this->generator(),
 322                         $this->caseSetting(),
 323                         $this->namespaces() );
 324                 return "  <siteinfo>\n    " .
 325                         implode( "\n    ", $info ) .
 326                         "\n  </siteinfo>\n";
 327         }
 328
 329         function sitename() {
 330                 global $wgSitename;
 331                 return wfElement( 'sitename', array(), $wgSitename );
 332         }
 333
 334         function generator() {
 335                 global $wgVersion;
 336                 return wfElement( 'generator', array(), "MediaWiki $wgVersion" );
 337         }
 338
 339         function homelink() {
 340                 return wfElement( 'base', array(), Title::newMainPage()->getFullUrl() );
 341         }
 342
 343         function caseSetting() {
 344                 global $wgCapitalLinks;
 345                 // "case-insensitive" option is reserved for future
 346                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 347                 return wfElement( 'case', array(), $sensitivity );
 348         }
 349
 350         function namespaces() {
 351                 global $wgContLang;
 352                 $spaces = "  <namespaces>\n";
 353                 foreach( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 354                         $spaces .= '      ' . wfElement( 'namespace', array( 'key' => $ns ), $title ) . "\n";
 355                 }
 356                 $spaces .= "    </namespaces>";
 357                 return $spaces;
 358         }
 359
 360         /**
 361          * Closes the output stream with the closing root element.
 362          * Call when finished dumping things.
 363          */
 364         function closeStream() {
 365                 return "</mediawiki>\n";
 366         }
 367
 368
 369         /**
 370          * Opens a <page> section on the output stream, with data
 371          * from the given database row.
 372          *
 373          * @param object $row
 374          * @return string
 375          * @access private
 376          */
 377         function openPage( $row ) {
 378                 $out = "  <page>\n";
 379                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 380                 $out .= '    ' . wfElementClean( 'title', array(), $title->getPrefixedText() ) . "\n";
 381                 $out .= '    ' . wfElement( 'id', array(), strval( $row->page_id ) ) . "\n";
 382                 if( '' != $row->page_restrictions ) {
 383                         $out .= '    ' . wfElement( 'restrictions', array(),
 384                                 strval( $row->page_restrictions ) ) . "\n";
 385                 }
 386                 return $out;
 387         }
 388
 389         /**
 390          * Closes a <page> section on the output stream.
 391          *
 392          * @access private
 393          */
 394         function closePage() {
 395                 return "  </page>\n";
 396         }
 397
 398         /**
 399          * Dumps a <revision> section on the output stream, with
 400          * data filled in from the given database row.
 401          *
 402          * @param object $row
 403          * @return string
 404          * @access private
 405          */
 406         function writeRevision( $row ) {
 407                 $fname = 'WikiExporter::dumpRev';
 408                 wfProfileIn( $fname );
 409
 410                 $out  = "    <revision>\n";
 411                 $out .= "      " . wfElement( 'id', null, strval( $row->rev_id ) ) . "\n";
 412
 413                 $ts = wfTimestamp( TS_ISO_8601, $row->rev_timestamp );
 414                 $out .= "      " . wfElement( 'timestamp', null, $ts ) . "\n";
 415
 416                 if( $row->rev_deleted & Revision::DELETED_USER ) {
 417                         $out .= "      " . wfElement( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 418                 } else {
 419                         $out .= "      <contributor>\n";
 420                         if( $row->rev_user ) {
 421                                 $out .= "        " . wfElementClean( 'username', null, strval( $row->rev_user_text ) ) . "\n";
 422                                 $out .= "        " . wfElement( 'id', null, strval( $row->rev_user ) ) . "\n";
 423                         } else {
 424                                 $out .= "        " . wfElementClean( 'ip', null, strval( $row->rev_user_text ) ) . "\n";
 425                         }
 426                         $out .= "      </contributor>\n";
 427                 }
 428
 429                 if( $row->rev_minor_edit ) {
 430                         $out .=  "      <minor/>\n";
 431                 }
 432                 if( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 433                         $out .= "      " . wfElement( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 434                 } elseif( $row->rev_comment != '' ) {
 435                         $out .= "      " . wfElementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 436                 }
 437
 438                 if( $row->rev_deleted & Revision::DELETED_TEXT ) {
 439                         $out .= "      " . wfElement( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 440                 } elseif( isset( $row->old_text ) ) {
 441                         // Raw text from the database may have invalid chars
 442                         $text = strval( Revision::getRevisionText( $row ) );
 443                         $out .= "      " . wfElementClean( 'text',
 444                                 array( 'xml:space' => 'preserve' ),
 445                                 strval( $text ) ) . "\n";
 446                 } else {
 447                         // Stub output
 448                         $out .= "      " . wfElement( 'text',
 449                                 array( 'id' => $row->rev_text_id ),
 450                                 "" ) . "\n";
 451                 }
 452
 453                 $out .= "    </revision>\n";
 454
 455                 wfProfileOut( $fname );
 456                 return $out;
 457         }
 458
 459 }
 460
 461
 462 /**
 463  * Base class for output stream; prints to stdout or buffer or whereever.
 464  */
 465 class DumpOutput {
 466         function writeOpenStream( $string ) {
 467                 $this->write( $string );
 468         }
 469
 470         function writeCloseStream( $string ) {
 471                 $this->write( $string );
 472         }
 473
 474         function writeOpenPage( $page, $string ) {
 475                 $this->write( $string );
 476         }
 477
 478         function writeClosePage( $string ) {
 479                 $this->write( $string );
 480         }
 481
 482         function writeRevision( $rev, $string ) {
 483                 $this->write( $string );
 484         }
 485
 486         /**
 487          * Override to write to a different stream type.
 488          * @return bool
 489          */
 490         function write( $string ) {
 491                 print $string;
 492         }
 493 }
 494
 495 /**
 496  * Stream outputter to send data to a file.
 497  */
 498 class DumpFileOutput extends DumpOutput {
 499         var $handle;
 500
 501         function DumpFileOutput( $file ) {
 502                 $this->handle = fopen( $file, "wt" );
 503         }
 504
 505         function write( $string ) {
 506                 fputs( $this->handle, $string );
 507         }
 508 }
 509
 510 /**
 511  * Stream outputter to send data to a file via some filter program.
 512  * Even if compression is available in a library, using a separate
 513  * program can allow us to make use of a multi-processor system.
 514  */
 515 class DumpPipeOutput extends DumpFileOutput {
 516         function DumpPipeOutput( $command, $file = null ) {
 517                 if( !is_null( $file ) ) {
 518                         $command .=  " > " . wfEscapeShellArg( $file );
 519                 }
 520                 $this->handle = popen( $command, "w" );
 521         }
 522 }
 523
 524 /**
 525  * Sends dump output via the gzip compressor.
 526  */
 527 class DumpGZipOutput extends DumpPipeOutput {
 528         function DumpGZipOutput( $file ) {
 529                 parent::DumpPipeOutput( "gzip", $file );
 530         }
 531 }
 532
 533 /**
 534  * Sends dump output via the bgzip2 compressor.
 535  */
 536 class DumpBZip2Output extends DumpPipeOutput {
 537         function DumpBZip2Output( $file ) {
 538                 parent::DumpPipeOutput( "bzip2", $file );
 539         }
 540 }
 541
 542 /**
 543  * Sends dump output via the p7zip compressor.
 544  */
 545 class Dump7ZipOutput extends DumpPipeOutput {
 546         function Dump7ZipOutput( $file ) {
 547                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 548                 // Suppress annoying useless crap from p7zip
 549                 // Unfortunately this could suppress real error messages too
 550                 $command .= " >/dev/null 2>&1";
 551                 parent::DumpPipeOutput( $command );
 552         }
 553 }
 554
 555
 556
 557 /**
 558  * Dump output filter class.
 559  * This just does output filtering and streaming; XML formatting is done
 560  * higher up, so be careful in what you do.
 561  */
 562 class DumpFilter {
 563         function DumpFilter( &$sink ) {
 564                 $this->sink =& $sink;
 565         }
 566
 567         function writeOpenStream( $string ) {
 568                 $this->sink->writeOpenStream( $string );
 569         }
 570
 571         function writeCloseStream( $string ) {
 572                 $this->sink->writeCloseStream( $string );
 573         }
 574
 575         function writeOpenPage( $page, $string ) {
 576                 $this->sendingThisPage = $this->pass( $page, $string );
 577                 if( $this->sendingThisPage ) {
 578                         $this->sink->writeOpenPage( $page, $string );
 579                 }
 580         }
 581
 582         function writeClosePage( $string ) {
 583                 if( $this->sendingThisPage ) {
 584                         $this->sink->writeClosePage( $string );
 585                         $this->sendingThisPage = false;
 586                 }
 587         }
 588
 589         function writeRevision( $rev, $string ) {
 590                 if( $this->sendingThisPage ) {
 591                         $this->sink->writeRevision( $rev, $string );
 592                 }
 593         }
 594
 595         /**
 596          * Override for page-based filter types.
 597          * @return bool
 598          */
 599         function pass( $page ) {
 600                 return true;
 601         }
 602 }
 603
 604 /**
 605  * Simple dump output filter to exclude all talk pages.
 606  */
 607 class DumpNotalkFilter extends DumpFilter {
 608         function pass( $page ) {
 609                 return !Namespace::isTalk( $page->page_namespace );
 610         }
 611 }
 612
 613 /**
 614  * Dump output filter to include or exclude pages in a given set of namespaces.
 615  */
 616 class DumpNamespaceFilter extends DumpFilter {
 617         var $invert = false;
 618         var $namespaces = array();
 619
 620         function DumpNamespaceFilter( &$sink, $param ) {
 621                 parent::DumpFilter( $sink );
 622
 623                 $constants = array(
 624                         "NS_MAIN"           => NS_MAIN,
 625                         "NS_TALK"           => NS_TALK,
 626                         "NS_USER"           => NS_USER,
 627                         "NS_USER_TALK"      => NS_USER_TALK,
 628                         "NS_PROJECT"        => NS_PROJECT,
 629                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
 630                         "NS_IMAGE"          => NS_IMAGE,
 631                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
 632                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
 633                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
 634                         "NS_TEMPLATE"       => NS_TEMPLATE,
 635                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
 636                         "NS_HELP"           => NS_HELP,
 637                         "NS_HELP_TALK"      => NS_HELP_TALK,
 638                         "NS_CATEGORY"       => NS_CATEGORY,
 639                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
 640
 641                 if( $param{0} == '!' ) {
 642                         $this->invert = true;
 643                         $param = substr( $param, 1 );
 644                 }
 645
 646                 foreach( explode( ',', $param ) as $key ) {
 647                         $key = trim( $key );
 648                         if( isset( $constants[$key] ) ) {
 649                                 $ns = $constants[$key];
 650                                 $this->namespaces[$ns] = true;
 651                         } elseif( is_numeric( $key ) ) {
 652                                 $ns = intval( $key );
 653                                 $this->namespaces[$ns] = true;
 654                         } else {
 655                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
 656                         }
 657                 }
 658         }
 659
 660         function pass( $page ) {
 661                 $match = isset( $this->namespaces[$page->page_namespace] );
 662                 return $this->invert xor $match;
 663         }
 664 }
 665
 666
 667 /**
 668  * Dump output filter to include only the last revision in each page sequence.
 669  */
 670 class DumpLatestFilter extends DumpFilter {
 671         var $page, $pageString, $rev, $revString;
 672
 673         function writeOpenPage( $page, $string ) {
 674                 $this->page = $page;
 675                 $this->pageString = $string;
 676         }
 677
 678         function writeClosePage( $string ) {
 679                 if( $this->rev ) {
 680                         $this->sink->writeOpenPage( $this->page, $this->pageString );
 681                         $this->sink->writeRevision( $this->rev, $this->revString );
 682                         $this->sink->writeClosePage( $string );
 683                 }
 684                 $this->rev = null;
 685                 $this->revString = null;
 686                 $this->page = null;
 687                 $this->pageString = null;
 688         }
 689
 690         function writeRevision( $rev, $string ) {
 691                 if( $rev->rev_id == $this->page->page_latest ) {
 692                         $this->rev = $rev;
 693                         $this->revString = $string;
 694                 }
 695         }
 696 }
 697
 698 /**
 699  * Base class for output stream; prints to stdout or buffer or whereever.
 700  */
 701 class DumpMultiWriter {
 702         function DumpMultiWriter( $sinks ) {
 703                 $this->sinks = $sinks;
 704                 $this->count = count( $sinks );
 705         }
 706
 707         function writeOpenStream( $string ) {
 708                 for( $i = 0; $i < $this->count; $i++ ) {
 709                         $this->sinks[$i]->writeOpenStream( $string );
 710                 }
 711         }
 712
 713         function writeCloseStream( $string ) {
 714                 for( $i = 0; $i < $this->count; $i++ ) {
 715                         $this->sinks[$i]->writeCloseStream( $string );
 716                 }
 717         }
 718
 719         function writeOpenPage( $page, $string ) {
 720                 for( $i = 0; $i < $this->count; $i++ ) {
 721                         $this->sinks[$i]->writeOpenPage( $page, $string );
 722                 }
 723         }
 724
 725         function writeClosePage( $string ) {
 726                 for( $i = 0; $i < $this->count; $i++ ) {
 727                         $this->sinks[$i]->writeClosePage( $string );
 728                 }
 729         }
 730
 731         function writeRevision( $rev, $string ) {
 732                 for( $i = 0; $i < $this->count; $i++ ) {
 733                         $this->sinks[$i]->writeRevision( $rev, $string );
 734                 }
 735         }
 736 }
 737
 738 function xmlsafe( $string ) {
 739         $fname = 'xmlsafe';
 740         wfProfileIn( $fname );
 741
 742         /**
 743          * The page may contain old data which has not been properly normalized.
 744          * Invalid UTF-8 sequences or forbidden control characters will make our
 745          * XML output invalid, so be sure to strip them out.
 746          */
 747         $string = UtfNormal::cleanUp( $string );
 748
 749         $string = htmlspecialchars( $string );
 750         wfProfileOut( $fname );
 751         return $string;
 752 }
 753
 754 ?>