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