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  *
  23  * @addtogroup SpecialPage
  24  */
  25 class WikiExporter {
  26         var $list_authors = false ; # Return distinct author list (when not returning full history)
  27         var $author_list = "" ;
  28
  29         const FULL = 0;
  30         const CURRENT = 1;
  31
  32         const BUFFER = 0;
  33         const STREAM = 1;
  34
  35         const TEXT = 0;
  36         const STUB = 1;
  37
  38         /**
  39          * If using WikiExporter::STREAM to stream a large amount of data,
  40          * provide a database connection which is not managed by
  41          * LoadBalancer to read from: some history blob types will
  42          * make additional queries to pull source data while the
  43          * main query is still running.
  44          *
  45          * @param Database $db
  46          * @param mixed $history one of WikiExporter::FULL or WikiExporter::CURRENT, or an
  47          *                       associative array:
  48          *                         offset: non-inclusive offset at which to start the query
  49          *                         limit: maximum number of rows to return
  50          *                         dir: "asc" or "desc" timestamp order
  51          * @param int $buffer one of WikiExporter::BUFFER or WikiExporter::STREAM
  52          */
  53         function __construct( &$db, $history = WikiExporter::CURRENT,
  54                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  55                 $this->db =& $db;
  56                 $this->history = $history;
  57                 $this->buffer  = $buffer;
  58                 $this->writer  = new XmlDumpWriter();
  59                 $this->sink    = new DumpOutput();
  60                 $this->text    = $text;
  61         }
  62
  63         /**
  64          * Set the DumpOutput or DumpFilter object which will receive
  65          * various row objects and XML output for filtering. Filters
  66          * can be chained or used as callbacks.
  67          *
  68          * @param mixed $callback
  69          */
  70         function setOutputSink( &$sink ) {
  71                 $this->sink =& $sink;
  72         }
  73
  74         function openStream() {
  75                 $output = $this->writer->openStream();
  76                 $this->sink->writeOpenStream( $output );
  77         }
  78
  79         function closeStream() {
  80                 $output = $this->writer->closeStream();
  81                 $this->sink->writeCloseStream( $output );
  82         }
  83
  84         /**
  85          * Dumps a series of page and revision records for all pages
  86          * in the database, either including complete history or only
  87          * the most recent version.
  88          */
  89         function allPages() {
  90                 return $this->dumpFrom( '' );
  91         }
  92
  93         /**
  94          * Dumps a series of page and revision records for those pages
  95          * in the database falling within the page_id range given.
  96          * @param int $start Inclusive lower limit (this id is included)
  97          * @param int $end   Exclusive upper limit (this id is not included)
  98          *                   If 0, no upper limit.
  99          */
 100         function pagesByRange( $start, $end ) {
 101                 $condition = 'page_id >= ' . intval( $start );
 102                 if( $end ) {
 103                         $condition .= ' AND page_id < ' . intval( $end );
 104                 }
 105                 return $this->dumpFrom( $condition );
 106         }
 107
 108         /**
 109          * @param Title $title
 110          */
 111         function pageByTitle( $title ) {
 112                 return $this->dumpFrom(
 113                         'page_namespace=' . $title->getNamespace() .
 114                         ' AND page_title=' . $this->db->addQuotes( $title->getDbKey() ) );
 115         }
 116
 117         function pageByName( $name ) {
 118                 $title = Title::newFromText( $name );
 119                 if( is_null( $title ) ) {
 120                         return new WikiError( "Can't export invalid title" );
 121                 } else {
 122                         return $this->pageByTitle( $title );
 123                 }
 124         }
 125
 126         function pagesByName( $names ) {
 127                 foreach( $names as $name ) {
 128                         $this->pageByName( $name );
 129                 }
 130         }
 131
 132
 133         // -------------------- private implementation below --------------------
 134
 135         # Generates the distinct list of authors of an article
 136         # Not called by default (depends on $this->list_authors)
 137         # Can be set by Special:Export when not exporting whole history
 138         function do_list_authors ( $page , $revision , $cond ) {
 139                 $fname = "do_list_authors" ;
 140                 wfProfileIn( $fname );
 141                 $this->author_list = "<contributors>";
 142                 //rev_deleted
 143                 $nothidden = '(rev_deleted & '.Revision::DELETED_USER.') = 0';
 144
 145                 $sql = "SELECT DISTINCT rev_user_text,rev_user FROM {$page},{$revision} WHERE page_id=rev_page AND $nothidden AND " . $cond ;
 146                 $result = $this->db->query( $sql, $fname );
 147                 $resultset = $this->db->resultObject( $result );
 148                 while( $row = $resultset->fetchObject() ) {
 149                         $this->author_list .= "<contributor>" .
 150                                 "<username>" .
 151                                 htmlentities( $row->rev_user_text )  .
 152                                 "</username>" .
 153                                 "<id>" .
 154                                 $row->rev_user .
 155                                 "</id>" .
 156                                 "</contributor>";
 157                 }
 158                 wfProfileOut( $fname );
 159                 $this->author_list .= "</contributors>";
 160         }
 161
 162         function dumpFrom( $cond = '' ) {
 163                 $fname = 'WikiExporter::dumpFrom';
 164                 wfProfileIn( $fname );
 165
 166                 $page     = $this->db->tableName( 'page' );
 167                 $revision = $this->db->tableName( 'revision' );
 168                 $text     = $this->db->tableName( 'text' );
 169
 170                 $order = 'ORDER BY page_id';
 171                 $limit = '';
 172
 173                 if( $this->history == WikiExporter::FULL ) {
 174                         $join = 'page_id=rev_page';
 175                 } elseif( $this->history == WikiExporter::CURRENT ) {
 176                         if ( $this->list_authors && $cond != '' )  { // List authors, if so desired
 177                                 $this->do_list_authors ( $page , $revision , $cond );
 178                         }
 179                         $join = 'page_id=rev_page AND page_latest=rev_id';
 180                 } elseif ( is_array( $this->history ) ) {
 181                         $join = 'page_id=rev_page';
 182                         if ( $this->history['dir'] == 'asc' ) {
 183                                 $op = '>';
 184                                 $order .= ', rev_timestamp';
 185                         } else {
 186                                 $op = '<';
 187                                 $order .= ', rev_timestamp DESC';
 188                         }
 189                         if ( !empty( $this->history['offset'] ) ) {
 190                                 $join .= " AND rev_timestamp $op " . $this->db->addQuotes(
 191                                         $this->db->timestamp( $this->history['offset'] ) );
 192                         }
 193                         if ( !empty( $this->history['limit'] ) ) {
 194                                 $limitNum = intval( $this->history['limit'] );
 195                                 if ( $limitNum > 0 ) {
 196                                         $limit = "LIMIT $limitNum";
 197                                 }
 198                         }
 199                 } else {
 200                         wfProfileOut( $fname );
 201                         return new WikiError( "$fname given invalid history dump type." );
 202                 }
 203                 $where = ( $cond == '' ) ? '' : "$cond AND";
 204
 205                 if( $this->buffer == WikiExporter::STREAM ) {
 206                         $prev = $this->db->bufferResults( false );
 207                 }
 208                 if( $cond == '' ) {
 209                         // Optimization hack for full-database dump
 210                         $revindex = $pageindex = $this->db->useIndexClause("PRIMARY");
 211                         $straight = ' /*! STRAIGHT_JOIN */ ';
 212                 } else {
 213                         $pageindex = '';
 214                         $revindex = '';
 215                         $straight = '';
 216                 }
 217                 if( $this->text == WikiExporter::STUB ) {
 218                         $sql = "SELECT $straight * FROM
 219                                         $page $pageindex,
 220                                         $revision $revindex
 221                                         WHERE $where $join
 222                                         $order $limit";
 223                 } else {
 224                         $sql = "SELECT $straight * FROM
 225                                         $page $pageindex,
 226                                         $revision $revindex,
 227                                         $text
 228                                         WHERE $where $join AND rev_text_id=old_id
 229                                         $order $limit";
 230                 }
 231                 $result = $this->db->query( $sql, $fname );
 232                 $wrapper = $this->db->resultObject( $result );
 233                 $this->outputStream( $wrapper );
 234
 235                 if ( $this->list_authors ) {
 236                         $this->outputStream( $wrapper );
 237                 }
 238
 239                 if( $this->buffer == WikiExporter::STREAM ) {
 240                         $this->db->bufferResults( $prev );
 241                 }
 242
 243                 wfProfileOut( $fname );
 244         }
 245
 246         /**
 247          * Runs through a query result set dumping page and revision records.
 248          * The result set should be sorted/grouped by page to avoid duplicate
 249          * page records in the output.
 250          *
 251          * The result set will be freed once complete. Should be safe for
 252          * streaming (non-buffered) queries, as long as it was made on a
 253          * separate database connection not managed by LoadBalancer; some
 254          * blob storage types will make queries to pull source data.
 255          *
 256          * @param ResultWrapper $resultset
 257          * @access private
 258          */
 259         function outputStream( $resultset ) {
 260                 $last = null;
 261                 while( $row = $resultset->fetchObject() ) {
 262                         if( is_null( $last ) ||
 263                                 $last->page_namespace != $row->page_namespace ||
 264                                 $last->page_title     != $row->page_title ) {
 265                                 if( isset( $last ) ) {
 266                                         $output = $this->writer->closePage();
 267                                         $this->sink->writeClosePage( $output );
 268                                 }
 269                                 $output = $this->writer->openPage( $row );
 270                                 $this->sink->writeOpenPage( $row, $output );
 271                                 $last = $row;
 272                         }
 273                         $output = $this->writer->writeRevision( $row );
 274                         $this->sink->writeRevision( $row, $output );
 275                 }
 276                 if( isset( $last ) ) {
 277                         $output = $this->author_list . $this->writer->closePage();
 278                         $this->sink->writeClosePage( $output );
 279                 }
 280                 $resultset->free();
 281         }
 282 }
 283
 284 class XmlDumpWriter {
 285
 286         /**
 287          * Returns the export schema version.
 288          * @return string
 289          */
 290         function schemaVersion() {
 291                 return "0.3"; // FIXME: upgrade to 0.4 when updated XSD is ready, for the revision deletion bits
 292         }
 293
 294         /**
 295          * Opens the XML output stream's root <mediawiki> element.
 296          * This does not include an xml directive, so is safe to include
 297          * as a subelement in a larger XML stream. Namespace and XML Schema
 298          * references are included.
 299          *
 300          * Output will be encoded in UTF-8.
 301          *
 302          * @return string
 303          */
 304         function openStream() {
 305                 global $wgContLanguageCode;
 306                 $ver = $this->schemaVersion();
 307                 return wfElement( 'mediawiki', array(
 308                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 309                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 310                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 311                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 312                         'version'            => $ver,
 313                         'xml:lang'           => $wgContLanguageCode ),
 314                         null ) .
 315                         "\n" .
 316                         $this->siteInfo();
 317         }
 318
 319         function siteInfo() {
 320                 $info = array(
 321                         $this->sitename(),
 322                         $this->homelink(),
 323                         $this->generator(),
 324                         $this->caseSetting(),
 325                         $this->namespaces() );
 326                 return "  <siteinfo>\n    " .
 327                         implode( "\n    ", $info ) .
 328                         "\n  </siteinfo>\n";
 329         }
 330
 331         function sitename() {
 332                 global $wgSitename;
 333                 return wfElement( 'sitename', array(), $wgSitename );
 334         }
 335
 336         function generator() {
 337                 global $wgVersion;
 338                 return wfElement( 'generator', array(), "MediaWiki $wgVersion" );
 339         }
 340
 341         function homelink() {
 342                 return wfElement( 'base', array(), Title::newMainPage()->getFullUrl() );
 343         }
 344
 345         function caseSetting() {
 346                 global $wgCapitalLinks;
 347                 // "case-insensitive" option is reserved for future
 348                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 349                 return wfElement( 'case', array(), $sensitivity );
 350         }
 351
 352         function namespaces() {
 353                 global $wgContLang;
 354                 $spaces = "  <namespaces>\n";
 355                 foreach( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 356                         $spaces .= '      ' . wfElement( 'namespace', array( 'key' => $ns ), $title ) . "\n";
 357                 }
 358                 $spaces .= "    </namespaces>";
 359                 return $spaces;
 360         }
 361
 362         /**
 363          * Closes the output stream with the closing root element.
 364          * Call when finished dumping things.
 365          */
 366         function closeStream() {
 367                 return "</mediawiki>\n";
 368         }
 369
 370
 371         /**
 372          * Opens a <page> section on the output stream, with data
 373          * from the given database row.
 374          *
 375          * @param object $row
 376          * @return string
 377          * @access private
 378          */
 379         function openPage( $row ) {
 380                 $out = "  <page>\n";
 381                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 382                 $out .= '    ' . wfElementClean( 'title', array(), $title->getPrefixedText() ) . "\n";
 383                 $out .= '    ' . wfElement( 'id', array(), strval( $row->page_id ) ) . "\n";
 384                 if( '' != $row->page_restrictions ) {
 385                         $out .= '    ' . wfElement( 'restrictions', array(),
 386                                 strval( $row->page_restrictions ) ) . "\n";
 387                 }
 388                 return $out;
 389         }
 390
 391         /**
 392          * Closes a <page> section on the output stream.
 393          *
 394          * @access private
 395          */
 396         function closePage() {
 397                 return "  </page>\n";
 398         }
 399
 400         /**
 401          * Dumps a <revision> section on the output stream, with
 402          * data filled in from the given database row.
 403          *
 404          * @param object $row
 405          * @return string
 406          * @access private
 407          */
 408         function writeRevision( $row ) {
 409                 $fname = 'WikiExporter::dumpRev';
 410                 wfProfileIn( $fname );
 411
 412                 $out  = "    <revision>\n";
 413                 $out .= "      " . wfElement( 'id', null, strval( $row->rev_id ) ) . "\n";
 414
 415                 $ts = wfTimestamp( TS_ISO_8601, $row->rev_timestamp );
 416                 $out .= "      " . wfElement( 'timestamp', null, $ts ) . "\n";
 417
 418                 if( $row->rev_deleted & Revision::DELETED_USER ) {
 419                         $out .= "      " . wfElement( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 420                 } else {
 421                         $out .= "      <contributor>\n";
 422                         if( $row->rev_user ) {
 423                                 $out .= "        " . wfElementClean( 'username', null, strval( $row->rev_user_text ) ) . "\n";
 424                                 $out .= "        " . wfElement( 'id', null, strval( $row->rev_user ) ) . "\n";
 425                         } else {
 426                                 $out .= "        " . wfElementClean( 'ip', null, strval( $row->rev_user_text ) ) . "\n";
 427                         }
 428                         $out .= "      </contributor>\n";
 429                 }
 430
 431                 if( $row->rev_minor_edit ) {
 432                         $out .=  "      <minor/>\n";
 433                 }
 434                 if( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 435                         $out .= "      " . wfElement( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 436                 } elseif( $row->rev_comment != '' ) {
 437                         $out .= "      " . wfElementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 438                 }
 439
 440                 if( $row->rev_deleted & Revision::DELETED_TEXT ) {
 441                         $out .= "      " . wfElement( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 442                 } elseif( isset( $row->old_text ) ) {
 443                         // Raw text from the database may have invalid chars
 444                         $text = strval( Revision::getRevisionText( $row ) );
 445                         $out .= "      " . wfElementClean( 'text',
 446                                 array( 'xml:space' => 'preserve' ),
 447                                 strval( $text ) ) . "\n";
 448                 } else {
 449                         // Stub output
 450                         $out .= "      " . wfElement( 'text',
 451                                 array( 'id' => $row->rev_text_id ),
 452                                 "" ) . "\n";
 453                 }
 454
 455                 $out .= "    </revision>\n";
 456
 457                 wfProfileOut( $fname );
 458                 return $out;
 459         }
 460
 461 }
 462
 463
 464 /**
 465  * Base class for output stream; prints to stdout or buffer or whereever.
 466  */
 467 class DumpOutput {
 468         function writeOpenStream( $string ) {
 469                 $this->write( $string );
 470         }
 471
 472         function writeCloseStream( $string ) {
 473                 $this->write( $string );
 474         }
 475
 476         function writeOpenPage( $page, $string ) {
 477                 $this->write( $string );
 478         }
 479
 480         function writeClosePage( $string ) {
 481                 $this->write( $string );
 482         }
 483
 484         function writeRevision( $rev, $string ) {
 485                 $this->write( $string );
 486         }
 487
 488         /**
 489          * Override to write to a different stream type.
 490          * @return bool
 491          */
 492         function write( $string ) {
 493                 print $string;
 494         }
 495 }
 496
 497 /**
 498  * Stream outputter to send data to a file.
 499  */
 500 class DumpFileOutput extends DumpOutput {
 501         var $handle;
 502
 503         function DumpFileOutput( $file ) {
 504                 $this->handle = fopen( $file, "wt" );
 505         }
 506
 507         function write( $string ) {
 508                 fputs( $this->handle, $string );
 509         }
 510 }
 511
 512 /**
 513  * Stream outputter to send data to a file via some filter program.
 514  * Even if compression is available in a library, using a separate
 515  * program can allow us to make use of a multi-processor system.
 516  */
 517 class DumpPipeOutput extends DumpFileOutput {
 518         function DumpPipeOutput( $command, $file = null ) {
 519                 if( !is_null( $file ) ) {
 520                         $command .=  " > " . wfEscapeShellArg( $file );
 521                 }
 522                 $this->handle = popen( $command, "w" );
 523         }
 524 }
 525
 526 /**
 527  * Sends dump output via the gzip compressor.
 528  */
 529 class DumpGZipOutput extends DumpPipeOutput {
 530         function DumpGZipOutput( $file ) {
 531                 parent::DumpPipeOutput( "gzip", $file );
 532         }
 533 }
 534
 535 /**
 536  * Sends dump output via the bgzip2 compressor.
 537  */
 538 class DumpBZip2Output extends DumpPipeOutput {
 539         function DumpBZip2Output( $file ) {
 540                 parent::DumpPipeOutput( "bzip2", $file );
 541         }
 542 }
 543
 544 /**
 545  * Sends dump output via the p7zip compressor.
 546  */
 547 class Dump7ZipOutput extends DumpPipeOutput {
 548         function Dump7ZipOutput( $file ) {
 549                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 550                 // Suppress annoying useless crap from p7zip
 551                 // Unfortunately this could suppress real error messages too
 552                 $command .= " >/dev/null 2>&1";
 553                 parent::DumpPipeOutput( $command );
 554         }
 555 }
 556
 557
 558
 559 /**
 560  * Dump output filter class.
 561  * This just does output filtering and streaming; XML formatting is done
 562  * higher up, so be careful in what you do.
 563  */
 564 class DumpFilter {
 565         function DumpFilter( &$sink ) {
 566                 $this->sink =& $sink;
 567         }
 568
 569         function writeOpenStream( $string ) {
 570                 $this->sink->writeOpenStream( $string );
 571         }
 572
 573         function writeCloseStream( $string ) {
 574                 $this->sink->writeCloseStream( $string );
 575         }
 576
 577         function writeOpenPage( $page, $string ) {
 578                 $this->sendingThisPage = $this->pass( $page, $string );
 579                 if( $this->sendingThisPage ) {
 580                         $this->sink->writeOpenPage( $page, $string );
 581                 }
 582         }
 583
 584         function writeClosePage( $string ) {
 585                 if( $this->sendingThisPage ) {
 586                         $this->sink->writeClosePage( $string );
 587                         $this->sendingThisPage = false;
 588                 }
 589         }
 590
 591         function writeRevision( $rev, $string ) {
 592                 if( $this->sendingThisPage ) {
 593                         $this->sink->writeRevision( $rev, $string );
 594                 }
 595         }
 596
 597         /**
 598          * Override for page-based filter types.
 599          * @return bool
 600          */
 601         function pass( $page ) {
 602                 return true;
 603         }
 604 }
 605
 606 /**
 607  * Simple dump output filter to exclude all talk pages.
 608  */
 609 class DumpNotalkFilter extends DumpFilter {
 610         function pass( $page ) {
 611                 return !Namespace::isTalk( $page->page_namespace );
 612         }
 613 }
 614
 615 /**
 616  * Dump output filter to include or exclude pages in a given set of namespaces.
 617  */
 618 class DumpNamespaceFilter extends DumpFilter {
 619         var $invert = false;
 620         var $namespaces = array();
 621
 622         function DumpNamespaceFilter( &$sink, $param ) {
 623                 parent::DumpFilter( $sink );
 624
 625                 $constants = array(
 626                         "NS_MAIN"           => NS_MAIN,
 627                         "NS_TALK"           => NS_TALK,
 628                         "NS_USER"           => NS_USER,
 629                         "NS_USER_TALK"      => NS_USER_TALK,
 630                         "NS_PROJECT"        => NS_PROJECT,
 631                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
 632                         "NS_IMAGE"          => NS_IMAGE,
 633                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
 634                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
 635                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
 636                         "NS_TEMPLATE"       => NS_TEMPLATE,
 637                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
 638                         "NS_HELP"           => NS_HELP,
 639                         "NS_HELP_TALK"      => NS_HELP_TALK,
 640                         "NS_CATEGORY"       => NS_CATEGORY,
 641                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
 642
 643                 if( $param{0} == '!' ) {
 644                         $this->invert = true;
 645                         $param = substr( $param, 1 );
 646                 }
 647
 648                 foreach( explode( ',', $param ) as $key ) {
 649                         $key = trim( $key );
 650                         if( isset( $constants[$key] ) ) {
 651                                 $ns = $constants[$key];
 652                                 $this->namespaces[$ns] = true;
 653                         } elseif( is_numeric( $key ) ) {
 654                                 $ns = intval( $key );
 655                                 $this->namespaces[$ns] = true;
 656                         } else {
 657                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
 658                         }
 659                 }
 660         }
 661
 662         function pass( $page ) {
 663                 $match = isset( $this->namespaces[$page->page_namespace] );
 664                 return $this->invert xor $match;
 665         }
 666 }
 667
 668
 669 /**
 670  * Dump output filter to include only the last revision in each page sequence.
 671  */
 672 class DumpLatestFilter extends DumpFilter {
 673         var $page, $pageString, $rev, $revString;
 674
 675         function writeOpenPage( $page, $string ) {
 676                 $this->page = $page;
 677                 $this->pageString = $string;
 678         }
 679
 680         function writeClosePage( $string ) {
 681                 if( $this->rev ) {
 682                         $this->sink->writeOpenPage( $this->page, $this->pageString );
 683                         $this->sink->writeRevision( $this->rev, $this->revString );
 684                         $this->sink->writeClosePage( $string );
 685                 }
 686                 $this->rev = null;
 687                 $this->revString = null;
 688                 $this->page = null;
 689                 $this->pageString = null;
 690         }
 691
 692         function writeRevision( $rev, $string ) {
 693                 if( $rev->rev_id == $this->page->page_latest ) {
 694                         $this->rev = $rev;
 695                         $this->revString = $string;
 696                 }
 697         }
 698 }
 699
 700 /**
 701  * Base class for output stream; prints to stdout or buffer or whereever.
 702  */
 703 class DumpMultiWriter {
 704         function DumpMultiWriter( $sinks ) {
 705                 $this->sinks = $sinks;
 706                 $this->count = count( $sinks );
 707         }
 708
 709         function writeOpenStream( $string ) {
 710                 for( $i = 0; $i < $this->count; $i++ ) {
 711                         $this->sinks[$i]->writeOpenStream( $string );
 712                 }
 713         }
 714
 715         function writeCloseStream( $string ) {
 716                 for( $i = 0; $i < $this->count; $i++ ) {
 717                         $this->sinks[$i]->writeCloseStream( $string );
 718                 }
 719         }
 720
 721         function writeOpenPage( $page, $string ) {
 722                 for( $i = 0; $i < $this->count; $i++ ) {
 723                         $this->sinks[$i]->writeOpenPage( $page, $string );
 724                 }
 725         }
 726
 727         function writeClosePage( $string ) {
 728                 for( $i = 0; $i < $this->count; $i++ ) {
 729                         $this->sinks[$i]->writeClosePage( $string );
 730                 }
 731         }
 732
 733         function writeRevision( $rev, $string ) {
 734                 for( $i = 0; $i < $this->count; $i++ ) {
 735                         $this->sinks[$i]->writeRevision( $rev, $string );
 736                 }
 737         }
 738 }
 739
 740 function xmlsafe( $string ) {
 741         $fname = 'xmlsafe';
 742         wfProfileIn( $fname );
 743
 744         /**
 745          * The page may contain old data which has not been properly normalized.
 746          * Invalid UTF-8 sequences or forbidden control characters will make our
 747          * XML output invalid, so be sure to strip them out.
 748          */
 749         $string = UtfNormal::cleanUp( $string );
 750
 751         $string = htmlspecialchars( $string );
 752         wfProfileOut( $fname );
 753         return $string;
 754 }
 755
 756 ?>