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  * @package MediaWiki
  22  * @subpackage SpecialPage
  23  */
  24
  25 /** */
  26
  27 define( 'MW_EXPORT_FULL',     0 );
  28 define( 'MW_EXPORT_CURRENT',  1 );
  29
  30 define( 'MW_EXPORT_BUFFER',   0 );
  31 define( 'MW_EXPORT_STREAM',   1 );
  32
  33 define( 'MW_EXPORT_TEXT',     0 );
  34 define( 'MW_EXPORT_STUB',     1 );
  35
  36
  37 /**
  38  * @package MediaWiki
  39  * @subpackage SpecialPage
  40  */
  41 class WikiExporter {
  42
  43         var $list_authors = false ; # Return distinct author list (when not returning full history)
  44         var $author_list = "" ;
  45
  46         /**
  47          * If using MW_EXPORT_STREAM to stream a large amount of data,
  48          * provide a database connection which is not managed by
  49          * LoadBalancer to read from: some history blob types will
  50          * make additional queries to pull source data while the
  51          * main query is still running.
  52          *
  53          * @param Database $db
  54          * @param mixed $history one of MW_EXPORT_FULL or MW_EXPORT_CURRENT, or an
  55          *                       associative array:
  56          *                         offset: non-inclusive offset at which to start the query
  57          *                         limit: maximum number of rows to return
  58          *                         dir: "asc" or "desc" timestamp order
  59          * @param int $buffer one of MW_EXPORT_BUFFER or MW_EXPORT_STREAM
  60          */
  61         function WikiExporter( &$db, $history = MW_EXPORT_CURRENT,
  62                         $buffer = MW_EXPORT_BUFFER, $text = MW_EXPORT_TEXT ) {
  63                 $this->db =& $db;
  64                 $this->history = $history;
  65                 $this->buffer  = $buffer;
  66                 $this->writer  = new XmlDumpWriter();
  67                 $this->sink    = new DumpOutput();
  68                 $this->text    = $text;
  69         }
  70
  71         /**
  72          * Set the DumpOutput or DumpFilter object which will receive
  73          * various row objects and XML output for filtering. Filters
  74          * can be chained or used as callbacks.
  75          *
  76          * @param mixed $callback
  77          */
  78         function setOutputSink( &$sink ) {
  79                 $this->sink =& $sink;
  80         }
  81
  82         function openStream() {
  83                 $output = $this->writer->openStream();
  84                 $this->sink->writeOpenStream( $output );
  85         }
  86
  87         function closeStream() {
  88                 $output = $this->writer->closeStream();
  89                 $this->sink->writeCloseStream( $output );
  90         }
  91
  92         /**
  93          * Dumps a series of page and revision records for all pages
  94          * in the database, either including complete history or only
  95          * the most recent version.
  96          */
  97         function allPages() {
  98                 return $this->dumpFrom( '' );
  99         }
 100
 101         /**
 102          * Dumps a series of page and revision records for those pages
 103          * in the database falling within the page_id range given.
 104          * @param int $start Inclusive lower limit (this id is included)
 105          * @param int $end   Exclusive upper limit (this id is not included)
 106          *                   If 0, no upper limit.
 107          */
 108         function pagesByRange( $start, $end ) {
 109                 $condition = 'page_id >= ' . intval( $start );
 110                 if( $end ) {
 111                         $condition .= ' AND page_id < ' . intval( $end );
 112                 }
 113                 return $this->dumpFrom( $condition );
 114         }
 115
 116         /**
 117          * @param Title $title
 118          */
 119         function pageByTitle( $title ) {
 120                 return $this->dumpFrom(
 121                         'page_namespace=' . $title->getNamespace() .
 122                         ' AND page_title=' . $this->db->addQuotes( $title->getDbKey() ) );
 123         }
 124
 125         function pageByName( $name ) {
 126                 $title = Title::newFromText( $name );
 127                 if( is_null( $title ) ) {
 128                         return new WikiError( "Can't export invalid title" );
 129                 } else {
 130                         return $this->pageByTitle( $title );
 131                 }
 132         }
 133
 134         function pagesByName( $names ) {
 135                 foreach( $names as $name ) {
 136                         $this->pageByName( $name );
 137                 }
 138         }
 139
 140
 141         // -------------------- private implementation below --------------------
 142
 143         # Generates the distinct list of authors of an article
 144         # Not called by default (depends on $this->list_authors)
 145         # Can be set by Special:Export when not exporting whole history
 146         function do_list_authors ( $page , $revision , $cond ) {
 147                 $fname = "do_list_authors" ;
 148                 wfProfileIn( $fname );
 149                 $this->author_list = "<contributors>";
 150                 $sql = "SELECT DISTINCT rev_user_text,rev_user FROM {$page},{$revision} WHERE page_id=rev_page AND " . $cond ;
 151                 $result = $this->db->query( $sql, $fname );
 152                 $resultset = $this->db->resultObject( $result );
 153                 while( $row = $resultset->fetchObject() ) {
 154                         $this->author_list .= "<contributor>" .
 155                                 "<username>" .
 156                                 htmlentities( $row->rev_user_text )  .
 157                                 "</username>" .
 158                                 "<id>" .
 159                                 $row->rev_user .
 160                                 "</id>" .
 161                                 "</contributor>";
 162                 }
 163                 wfProfileOut( $fname );
 164                 $this->author_list .= "</contributors>";
 165         }
 166
 167         function dumpFrom( $cond = '' ) {
 168                 $fname = 'WikiExporter::dumpFrom';
 169                 wfProfileIn( $fname );
 170
 171                 $page     = $this->db->tableName( 'page' );
 172                 $revision = $this->db->tableName( 'revision' );
 173                 $text     = $this->db->tableName( 'text' );
 174
 175                 $order = 'ORDER BY page_id';
 176                 $limit = '';
 177
 178                 if( $this->history == MW_EXPORT_FULL ) {
 179                         $join = 'page_id=rev_page';
 180                 } elseif( $this->history == MW_EXPORT_CURRENT ) {
 181                         if ( $this->list_authors && $cond != '' )  { // List authors, if so desired
 182                                 $this->do_list_authors ( $page , $revision , $cond );
 183                         }
 184                         $join = 'page_id=rev_page AND page_latest=rev_id';
 185                 } elseif ( is_array( $this->history ) ) {
 186                         $join = 'page_id=rev_page';
 187                         if ( $this->history['dir'] == 'asc' ) {
 188                                 $op = '>';
 189                                 $order .= ', rev_timestamp';
 190                         } else {
 191                                 $op = '<';
 192                                 $order .= ', rev_timestamp DESC';
 193                         }
 194                         if ( !empty( $this->history['offset'] ) ) {
 195                                 $join .= " AND rev_timestamp $op " . $this->db->addQuotes(
 196                                         $this->db->timestamp( $this->history['offset'] ) );
 197                         }
 198                         if ( !empty( $this->history['limit'] ) ) {
 199                                 $limitNum = intval( $this->history['limit'] );
 200                                 if ( $limitNum > 0 ) {
 201                                         $limit = "LIMIT $limitNum";
 202                                 }
 203                         }
 204                 } else {
 205                         wfProfileOut( $fname );
 206                         return new WikiError( "$fname given invalid history dump type." );
 207                 }
 208                 $where = ( $cond == '' ) ? '' : "$cond AND";
 209
 210                 if( $this->buffer == MW_EXPORT_STREAM ) {
 211                         $prev = $this->db->bufferResults( false );
 212                 }
 213                 if( $cond == '' ) {
 214                         // Optimization hack for full-database dump
 215                         $revindex = $pageindex = $this->db->useIndexClause("PRIMARY");
 216                         $straight = ' /*! STRAIGHT_JOIN */ ';
 217                 } else {
 218                         $pageindex = '';
 219                         $revindex = '';
 220                         $straight = '';
 221                 }
 222                 if( $this->text == MW_EXPORT_STUB ) {
 223                         $sql = "SELECT $straight * FROM
 224                                         $page $pageindex,
 225                                         $revision $revindex
 226                                         WHERE $where $join
 227                                         $order $limit";
 228                 } else {
 229                         $sql = "SELECT $straight * FROM
 230                                         $page $pageindex,
 231                                         $revision $revindex,
 232                                         $text
 233                                         WHERE $where $join AND rev_text_id=old_id
 234                                         $order $limit";
 235                 }
 236                 $result = $this->db->query( $sql, $fname );
 237                 $wrapper = $this->db->resultObject( $result );
 238                 $this->outputStream( $wrapper );
 239
 240                 if ( $this->list_authors ) {
 241                         $this->outputStream( $wrapper );
 242                 }
 243
 244                 if( $this->buffer == MW_EXPORT_STREAM ) {
 245                         $this->db->bufferResults( $prev );
 246                 }
 247
 248                 wfProfileOut( $fname );
 249         }
 250
 251         /**
 252          * Runs through a query result set dumping page and revision records.
 253          * The result set should be sorted/grouped by page to avoid duplicate
 254          * page records in the output.
 255          *
 256          * The result set will be freed once complete. Should be safe for
 257          * streaming (non-buffered) queries, as long as it was made on a
 258          * separate database connection not managed by LoadBalancer; some
 259          * blob storage types will make queries to pull source data.
 260          *
 261          * @param ResultWrapper $resultset
 262          * @access private
 263          */
 264         function outputStream( $resultset ) {
 265                 $last = null;
 266                 while( $row = $resultset->fetchObject() ) {
 267                         if( is_null( $last ) ||
 268                                 $last->page_namespace != $row->page_namespace ||
 269                                 $last->page_title     != $row->page_title ) {
 270                                 if( isset( $last ) ) {
 271                                         $output = $this->writer->closePage();
 272                                         $this->sink->writeClosePage( $output );
 273                                 }
 274                                 $output = $this->writer->openPage( $row );
 275                                 $this->sink->writeOpenPage( $row, $output );
 276                                 $last = $row;
 277                         }
 278                         $output = $this->writer->writeRevision( $row );
 279                         $this->sink->writeRevision( $row, $output );
 280                 }
 281                 if( isset( $last ) ) {
 282                         $output = $this->author_list . $this->writer->closePage();
 283                         $this->sink->writeClosePage( $output );
 284                 }
 285                 $resultset->free();
 286         }
 287 }
 288
 289 class XmlDumpWriter {
 290
 291         /**
 292          * Returns the export schema version.
 293          * @return string
 294          */
 295         function schemaVersion() {
 296                 return "0.3"; // FIXME: upgrade to 0.4 when updated XSD is ready, for the revision deletion bits
 297         }
 298
 299         /**
 300          * Opens the XML output stream's root <mediawiki> element.
 301          * This does not include an xml directive, so is safe to include
 302          * as a subelement in a larger XML stream. Namespace and XML Schema
 303          * references are included.
 304          *
 305          * Output will be encoded in UTF-8.
 306          *
 307          * @return string
 308          */
 309         function openStream() {
 310                 global $wgContLanguageCode;
 311                 $ver = $this->schemaVersion();
 312                 return wfElement( 'mediawiki', array(
 313                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 314                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 315                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 316                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 317                         'version'            => $ver,
 318                         'xml:lang'           => $wgContLanguageCode ),
 319                         null ) .
 320                         "\n" .
 321                         $this->siteInfo();
 322         }
 323
 324         function siteInfo() {
 325                 $info = array(
 326                         $this->sitename(),
 327                         $this->homelink(),
 328                         $this->generator(),
 329                         $this->caseSetting(),
 330                         $this->namespaces() );
 331                 return "  <siteinfo>\n    " .
 332                         implode( "\n    ", $info ) .
 333                         "\n  </siteinfo>\n";
 334         }
 335
 336         function sitename() {
 337                 global $wgSitename;
 338                 return wfElement( 'sitename', array(), $wgSitename );
 339         }
 340
 341         function generator() {
 342                 global $wgVersion;
 343                 return wfElement( 'generator', array(), "MediaWiki $wgVersion" );
 344         }
 345
 346         function homelink() {
 347                 $page = Title::newFromText( wfMsgForContent( 'mainpage' ) );
 348                 return wfElement( 'base', array(), $page->getFullUrl() );
 349         }
 350
 351         function caseSetting() {
 352                 global $wgCapitalLinks;
 353                 // "case-insensitive" option is reserved for future
 354                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 355                 return wfElement( 'case', array(), $sensitivity );
 356         }
 357
 358         function namespaces() {
 359                 global $wgContLang;
 360                 $spaces = "  <namespaces>\n";
 361                 foreach( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 362                         $spaces .= '      ' . wfElement( 'namespace', array( 'key' => $ns ), $title ) . "\n";
 363                 }
 364                 $spaces .= "    </namespaces>";
 365                 return $spaces;
 366         }
 367
 368         /**
 369          * Closes the output stream with the closing root element.
 370          * Call when finished dumping things.
 371          */
 372         function closeStream() {
 373                 return "</mediawiki>\n";
 374         }
 375
 376
 377         /**
 378          * Opens a <page> section on the output stream, with data
 379          * from the given database row.
 380          *
 381          * @param object $row
 382          * @return string
 383          * @access private
 384          */
 385         function openPage( $row ) {
 386                 $out = "  <page>\n";
 387                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 388                 $out .= '    ' . wfElementClean( 'title', array(), $title->getPrefixedText() ) . "\n";
 389                 $out .= '    ' . wfElement( 'id', array(), strval( $row->page_id ) ) . "\n";
 390                 if( '' != $row->page_restrictions ) {
 391                         $out .= '    ' . wfElement( 'restrictions', array(),
 392                                 strval( $row->page_restrictions ) ) . "\n";
 393                 }
 394                 return $out;
 395         }
 396
 397         /**
 398          * Closes a <page> section on the output stream.
 399          *
 400          * @access private
 401          */
 402         function closePage() {
 403                 return "  </page>\n";
 404         }
 405
 406         /**
 407          * Dumps a <revision> section on the output stream, with
 408          * data filled in from the given database row.
 409          *
 410          * @param object $row
 411          * @return string
 412          * @access private
 413          */
 414         function writeRevision( $row ) {
 415                 $fname = 'WikiExporter::dumpRev';
 416                 wfProfileIn( $fname );
 417
 418                 $out  = "    <revision>\n";
 419                 $out .= "      " . wfElement( 'id', null, strval( $row->rev_id ) ) . "\n";
 420
 421                 $ts = wfTimestamp( TS_ISO_8601, $row->rev_timestamp );
 422                 $out .= "      " . wfElement( 'timestamp', null, $ts ) . "\n";
 423
 424                 if( $row->rev_deleted & Revision::DELETED_USER ) {
 425                         $out .= "      " . wfElement( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 426                 } else {
 427                         $out .= "      <contributor>\n";
 428                         if( $row->rev_user ) {
 429                                 $out .= "        " . wfElementClean( 'username', null, strval( $row->rev_user_text ) ) . "\n";
 430                                 $out .= "        " . wfElement( 'id', null, strval( $row->rev_user ) ) . "\n";
 431                         } else {
 432                                 $out .= "        " . wfElementClean( 'ip', null, strval( $row->rev_user_text ) ) . "\n";
 433                         }
 434                         $out .= "      </contributor>\n";
 435                 }
 436
 437                 if( $row->rev_minor_edit ) {
 438                         $out .=  "      <minor/>\n";
 439                 }
 440                 if( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 441                         $out .= "      " . wfElement( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 442                 } elseif( $row->rev_comment != '' ) {
 443                         $out .= "      " . wfElementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 444                 }
 445
 446                 if( $row->rev_deleted & Revision::DELETED_TEXT ) {
 447                         $out .= "      " . wfElement( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 448                 } elseif( isset( $row->old_text ) ) {
 449                         // Raw text from the database may have invalid chars
 450                         $text = strval( Revision::getRevisionText( $row ) );
 451                         $out .= "      " . wfElementClean( 'text',
 452                                 array( 'xml:space' => 'preserve' ),
 453                                 strval( $text ) ) . "\n";
 454                 } else {
 455                         // Stub output
 456                         $out .= "      " . wfElement( 'text',
 457                                 array( 'id' => $row->rev_text_id ),
 458                                 "" ) . "\n";
 459                 }
 460
 461                 $out .= "    </revision>\n";
 462
 463                 wfProfileOut( $fname );
 464                 return $out;
 465         }
 466
 467 }
 468
 469
 470 /**
 471  * Base class for output stream; prints to stdout or buffer or whereever.
 472  */
 473 class DumpOutput {
 474         function writeOpenStream( $string ) {
 475                 $this->write( $string );
 476         }
 477
 478         function writeCloseStream( $string ) {
 479                 $this->write( $string );
 480         }
 481
 482         function writeOpenPage( $page, $string ) {
 483                 $this->write( $string );
 484         }
 485
 486         function writeClosePage( $string ) {
 487                 $this->write( $string );
 488         }
 489
 490         function writeRevision( $rev, $string ) {
 491                 $this->write( $string );
 492         }
 493
 494         /**
 495          * Override to write to a different stream type.
 496          * @return bool
 497          */
 498         function write( $string ) {
 499                 print $string;
 500         }
 501 }
 502
 503 /**
 504  * Stream outputter to send data to a file.
 505  */
 506 class DumpFileOutput extends DumpOutput {
 507         var $handle;
 508
 509         function DumpFileOutput( $file ) {
 510                 $this->handle = fopen( $file, "wt" );
 511         }
 512
 513         function write( $string ) {
 514                 fputs( $this->handle, $string );
 515         }
 516 }
 517
 518 /**
 519  * Stream outputter to send data to a file via some filter program.
 520  * Even if compression is available in a library, using a separate
 521  * program can allow us to make use of a multi-processor system.
 522  */
 523 class DumpPipeOutput extends DumpFileOutput {
 524         function DumpPipeOutput( $command, $file = null ) {
 525                 if( !is_null( $file ) ) {
 526                         $command .=  " > " . wfEscapeShellArg( $file );
 527                 }
 528                 $this->handle = popen( $command, "w" );
 529         }
 530 }
 531
 532 /**
 533  * Sends dump output via the gzip compressor.
 534  */
 535 class DumpGZipOutput extends DumpPipeOutput {
 536         function DumpGZipOutput( $file ) {
 537                 parent::DumpPipeOutput( "gzip", $file );
 538         }
 539 }
 540
 541 /**
 542  * Sends dump output via the bgzip2 compressor.
 543  */
 544 class DumpBZip2Output extends DumpPipeOutput {
 545         function DumpBZip2Output( $file ) {
 546                 parent::DumpPipeOutput( "bzip2", $file );
 547         }
 548 }
 549
 550 /**
 551  * Sends dump output via the p7zip compressor.
 552  */
 553 class Dump7ZipOutput extends DumpPipeOutput {
 554         function Dump7ZipOutput( $file ) {
 555                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 556                 // Suppress annoying useless crap from p7zip
 557                 // Unfortunately this could suppress real error messages too
 558                 $command .= " >/dev/null 2>&1";
 559                 parent::DumpPipeOutput( $command );
 560         }
 561 }
 562
 563
 564
 565 /**
 566  * Dump output filter class.
 567  * This just does output filtering and streaming; XML formatting is done
 568  * higher up, so be careful in what you do.
 569  */
 570 class DumpFilter {
 571         function DumpFilter( &$sink ) {
 572                 $this->sink =& $sink;
 573         }
 574
 575         function writeOpenStream( $string ) {
 576                 $this->sink->writeOpenStream( $string );
 577         }
 578
 579         function writeCloseStream( $string ) {
 580                 $this->sink->writeCloseStream( $string );
 581         }
 582
 583         function writeOpenPage( $page, $string ) {
 584                 $this->sendingThisPage = $this->pass( $page, $string );
 585                 if( $this->sendingThisPage ) {
 586                         $this->sink->writeOpenPage( $page, $string );
 587                 }
 588         }
 589
 590         function writeClosePage( $string ) {
 591                 if( $this->sendingThisPage ) {
 592                         $this->sink->writeClosePage( $string );
 593                         $this->sendingThisPage = false;
 594                 }
 595         }
 596
 597         function writeRevision( $rev, $string ) {
 598                 if( $this->sendingThisPage ) {
 599                         $this->sink->writeRevision( $rev, $string );
 600                 }
 601         }
 602
 603         /**
 604          * Override for page-based filter types.
 605          * @return bool
 606          */
 607         function pass( $page, $string ) {
 608                 return true;
 609         }
 610 }
 611
 612 /**
 613  * Simple dump output filter to exclude all talk pages.
 614  */
 615 class DumpNotalkFilter extends DumpFilter {
 616         function pass( $page ) {
 617                 return !Namespace::isTalk( $page->page_namespace );
 618         }
 619 }
 620
 621 /**
 622  * Dump output filter to include or exclude pages in a given set of namespaces.
 623  */
 624 class DumpNamespaceFilter extends DumpFilter {
 625         var $invert = false;
 626         var $namespaces = array();
 627
 628         function DumpNamespaceFilter( &$sink, $param ) {
 629                 parent::DumpFilter( $sink );
 630
 631                 $constants = array(
 632                         "NS_MAIN"           => NS_MAIN,
 633                         "NS_TALK"           => NS_TALK,
 634                         "NS_USER"           => NS_USER,
 635                         "NS_USER_TALK"      => NS_USER_TALK,
 636                         "NS_PROJECT"        => NS_PROJECT,
 637                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
 638                         "NS_IMAGE"          => NS_IMAGE,
 639                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
 640                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
 641                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
 642                         "NS_TEMPLATE"       => NS_TEMPLATE,
 643                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
 644                         "NS_HELP"           => NS_HELP,
 645                         "NS_HELP_TALK"      => NS_HELP_TALK,
 646                         "NS_CATEGORY"       => NS_CATEGORY,
 647                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
 648
 649                 if( $param{0} == '!' ) {
 650                         $this->invert = true;
 651                         $param = substr( $param, 1 );
 652                 }
 653
 654                 foreach( explode( ',', $param ) as $key ) {
 655                         $key = trim( $key );
 656                         if( isset( $constants[$key] ) ) {
 657                                 $ns = $constants[$key];
 658                                 $this->namespaces[$ns] = true;
 659                         } elseif( is_numeric( $key ) ) {
 660                                 $ns = intval( $key );
 661                                 $this->namespaces[$ns] = true;
 662                         } else {
 663                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
 664                         }
 665                 }
 666         }
 667
 668         function pass( $page ) {
 669                 $match = isset( $this->namespaces[$page->page_namespace] );
 670                 return $this->invert xor $match;
 671         }
 672 }
 673
 674
 675 /**
 676  * Dump output filter to include only the last revision in each page sequence.
 677  */
 678 class DumpLatestFilter extends DumpFilter {
 679         var $page, $pageString, $rev, $revString;
 680
 681         function writeOpenPage( $page, $string ) {
 682                 $this->page = $page;
 683                 $this->pageString = $string;
 684         }
 685
 686         function writeClosePage( $string ) {
 687                 if( $this->rev ) {
 688                         $this->sink->writeOpenPage( $this->page, $this->pageString );
 689                         $this->sink->writeRevision( $this->rev, $this->revString );
 690                         $this->sink->writeClosePage( $string );
 691                 }
 692                 $this->rev = null;
 693                 $this->revString = null;
 694                 $this->page = null;
 695                 $this->pageString = null;
 696         }
 697
 698         function writeRevision( $rev, $string ) {
 699                 if( $rev->rev_id == $this->page->page_latest ) {
 700                         $this->rev = $rev;
 701                         $this->revString = $string;
 702                 }
 703         }
 704 }
 705
 706 /**
 707  * Base class for output stream; prints to stdout or buffer or whereever.
 708  */
 709 class DumpMultiWriter {
 710         function DumpMultiWriter( $sinks ) {
 711                 $this->sinks = $sinks;
 712                 $this->count = count( $sinks );
 713         }
 714
 715         function writeOpenStream( $string ) {
 716                 for( $i = 0; $i < $this->count; $i++ ) {
 717                         $this->sinks[$i]->writeOpenStream( $string );
 718                 }
 719         }
 720
 721         function writeCloseStream( $string ) {
 722                 for( $i = 0; $i < $this->count; $i++ ) {
 723                         $this->sinks[$i]->writeCloseStream( $string );
 724                 }
 725         }
 726
 727         function writeOpenPage( $page, $string ) {
 728                 for( $i = 0; $i < $this->count; $i++ ) {
 729                         $this->sinks[$i]->writeOpenPage( $page, $string );
 730                 }
 731         }
 732
 733         function writeClosePage( $string ) {
 734                 for( $i = 0; $i < $this->count; $i++ ) {
 735                         $this->sinks[$i]->writeClosePage( $string );
 736                 }
 737         }
 738
 739         function writeRevision( $rev, $string ) {
 740                 for( $i = 0; $i < $this->count; $i++ ) {
 741                         $this->sinks[$i]->writeRevision( $rev, $string );
 742                 }
 743         }
 744 }
 745
 746 function xmlsafe( $string ) {
 747         $fname = 'xmlsafe';
 748         wfProfileIn( $fname );
 749
 750         /**
 751          * The page may contain old data which has not been properly normalized.
 752          * Invalid UTF-8 sequences or forbidden control characters will make our
 753          * XML output invalid, so be sure to strip them out.
 754          */
 755         $string = UtfNormal::cleanUp( $string );
 756
 757         $string = htmlspecialchars( $string );
 758         wfProfileOut( $fname );
 759         return $string;
 760 }
 761
 762 ?>