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         var $dumpUploads = false;
  30
  31         const FULL = 0;
  32         const CURRENT = 1;
  33
  34         const BUFFER = 0;
  35         const STREAM = 1;
  36
  37         const TEXT = 0;
  38         const STUB = 1;
  39
  40         /**
  41          * If using WikiExporter::STREAM to stream a large amount of data,
  42          * provide a database connection which is not managed by
  43          * LoadBalancer to read from: some history blob types will
  44          * make additional queries to pull source data while the
  45          * main query is still running.
  46          *
  47          * @param Database $db
  48          * @param mixed $history one of WikiExporter::FULL or WikiExporter::CURRENT, or an
  49          *                       associative array:
  50          *                         offset: non-inclusive offset at which to start the query
  51          *                         limit: maximum number of rows to return
  52          *                         dir: "asc" or "desc" timestamp order
  53          * @param int $buffer one of WikiExporter::BUFFER or WikiExporter::STREAM
  54          */
  55         function __construct( &$db, $history = WikiExporter::CURRENT,
  56                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  57                 $this->db =& $db;
  58                 $this->history = $history;
  59                 $this->buffer  = $buffer;
  60                 $this->writer  = new XmlDumpWriter();
  61                 $this->sink    = new DumpOutput();
  62                 $this->text    = $text;
  63         }
  64
  65         /**
  66          * Set the DumpOutput or DumpFilter object which will receive
  67          * various row objects and XML output for filtering. Filters
  68          * can be chained or used as callbacks.
  69          *
  70          * @param mixed $callback
  71          */
  72         function setOutputSink( &$sink ) {
  73                 $this->sink =& $sink;
  74         }
  75
  76         function openStream() {
  77                 $output = $this->writer->openStream();
  78                 $this->sink->writeOpenStream( $output );
  79         }
  80
  81         function closeStream() {
  82                 $output = $this->writer->closeStream();
  83                 $this->sink->writeCloseStream( $output );
  84         }
  85
  86         /**
  87          * Dumps a series of page and revision records for all pages
  88          * in the database, either including complete history or only
  89          * the most recent version.
  90          */
  91         function allPages() {
  92                 return $this->dumpFrom( '' );
  93         }
  94
  95         /**
  96          * Dumps a series of page and revision records for those pages
  97          * in the database falling within the page_id range given.
  98          * @param int $start Inclusive lower limit (this id is included)
  99          * @param int $end   Exclusive upper limit (this id is not included)
 100          *                   If 0, no upper limit.
 101          */
 102         function pagesByRange( $start, $end ) {
 103                 $condition = 'page_id >= ' . intval( $start );
 104                 if( $end ) {
 105                         $condition .= ' AND page_id < ' . intval( $end );
 106                 }
 107                 return $this->dumpFrom( $condition );
 108         }
 109
 110         /**
 111          * @param Title $title
 112          */
 113         function pageByTitle( $title ) {
 114                 return $this->dumpFrom(
 115                         'page_namespace=' . $title->getNamespace() .
 116                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 117         }
 118
 119         function pageByName( $name ) {
 120                 $title = Title::newFromText( $name );
 121                 if( is_null( $title ) ) {
 122                         return new WikiError( "Can't export invalid title" );
 123                 } else {
 124                         return $this->pageByTitle( $title );
 125                 }
 126         }
 127
 128         function pagesByName( $names ) {
 129                 foreach( $names as $name ) {
 130                         $this->pageByName( $name );
 131                 }
 132         }
 133
 134
 135         // -------------------- private implementation below --------------------
 136
 137         # Generates the distinct list of authors of an article
 138         # Not called by default (depends on $this->list_authors)
 139         # Can be set by Special:Export when not exporting whole history
 140         function do_list_authors ( $page , $revision , $cond ) {
 141                 $fname = "do_list_authors" ;
 142                 wfProfileIn( $fname );
 143                 $this->author_list = "<contributors>";
 144                 //rev_deleted
 145                 $nothidden = '(rev_deleted & '.Revision::DELETED_USER.') = 0';
 146
 147                 $sql = "SELECT DISTINCT rev_user_text,rev_user FROM {$page},{$revision} WHERE page_id=rev_page AND $nothidden AND " . $cond ;
 148                 $result = $this->db->query( $sql, $fname );
 149                 $resultset = $this->db->resultObject( $result );
 150                 while( $row = $resultset->fetchObject() ) {
 151                         $this->author_list .= "<contributor>" .
 152                                 "<username>" .
 153                                 htmlentities( $row->rev_user_text )  .
 154                                 "</username>" .
 155                                 "<id>" .
 156                                 $row->rev_user .
 157                                 "</id>" .
 158                                 "</contributor>";
 159                 }
 160                 wfProfileOut( $fname );
 161                 $this->author_list .= "</contributors>";
 162         }
 163
 164         function dumpFrom( $cond = '' ) {
 165                 $fname = 'WikiExporter::dumpFrom';
 166                 wfProfileIn( $fname );
 167
 168                 $page     = $this->db->tableName( 'page' );
 169                 $revision = $this->db->tableName( 'revision' );
 170                 $text     = $this->db->tableName( 'text' );
 171
 172                 $order = 'ORDER BY page_id';
 173                 $limit = '';
 174
 175                 if( $this->history == WikiExporter::FULL ) {
 176                         $join = 'page_id=rev_page';
 177                 } elseif( $this->history == WikiExporter::CURRENT ) {
 178                         if ( $this->list_authors && $cond != '' )  { // List authors, if so desired
 179                                 $this->do_list_authors ( $page , $revision , $cond );
 180                         }
 181                         $join = 'page_id=rev_page AND page_latest=rev_id';
 182                 } elseif ( is_array( $this->history ) ) {
 183                         $join = 'page_id=rev_page';
 184                         if ( $this->history['dir'] == 'asc' ) {
 185                                 $op = '>';
 186                                 $order .= ', rev_timestamp';
 187                         } else {
 188                                 $op = '<';
 189                                 $order .= ', rev_timestamp DESC';
 190                         }
 191                         if ( !empty( $this->history['offset'] ) ) {
 192                                 $join .= " AND rev_timestamp $op " . $this->db->addQuotes(
 193                                         $this->db->timestamp( $this->history['offset'] ) );
 194                         }
 195                         if ( !empty( $this->history['limit'] ) ) {
 196                                 $limitNum = intval( $this->history['limit'] );
 197                                 if ( $limitNum > 0 ) {
 198                                         $limit = "LIMIT $limitNum";
 199                                 }
 200                         }
 201                 } else {
 202                         wfProfileOut( $fname );
 203                         return new WikiError( "$fname given invalid history dump type." );
 204                 }
 205                 $where = ( $cond == '' ) ? '' : "$cond AND";
 206
 207                 if( $this->buffer == WikiExporter::STREAM ) {
 208                         $prev = $this->db->bufferResults( false );
 209                 }
 210                 if( $cond == '' ) {
 211                         // Optimization hack for full-database dump
 212                         $revindex = $pageindex = $this->db->useIndexClause("PRIMARY");
 213                         $straight = ' /*! STRAIGHT_JOIN */ ';
 214                 } else {
 215                         $pageindex = '';
 216                         $revindex = '';
 217                         $straight = '';
 218                 }
 219                 if( $this->text == WikiExporter::STUB ) {
 220                         $sql = "SELECT $straight * FROM
 221                                         $page $pageindex,
 222                                         $revision $revindex
 223                                         WHERE $where $join
 224                                         $order $limit";
 225                 } else {
 226                         $sql = "SELECT $straight * FROM
 227                                         $page $pageindex,
 228                                         $revision $revindex,
 229                                         $text
 230                                         WHERE $where $join AND rev_text_id=old_id
 231                                         $order $limit";
 232                 }
 233                 $result = $this->db->query( $sql, $fname );
 234                 $wrapper = $this->db->resultObject( $result );
 235                 $this->outputStream( $wrapper );
 236
 237                 if ( $this->list_authors ) {
 238                         $this->outputStream( $wrapper );
 239                 }
 240
 241                 if( $this->buffer == WikiExporter::STREAM ) {
 242                         $this->db->bufferResults( $prev );
 243                 }
 244
 245                 wfProfileOut( $fname );
 246         }
 247
 248         /**
 249          * Runs through a query result set dumping page and revision records.
 250          * The result set should be sorted/grouped by page to avoid duplicate
 251          * page records in the output.
 252          *
 253          * The result set will be freed once complete. Should be safe for
 254          * streaming (non-buffered) queries, as long as it was made on a
 255          * separate database connection not managed by LoadBalancer; some
 256          * blob storage types will make queries to pull source data.
 257          *
 258          * @param ResultWrapper $resultset
 259          * @access private
 260          */
 261         function outputStream( $resultset ) {
 262                 $last = null;
 263                 while( $row = $resultset->fetchObject() ) {
 264                         if( is_null( $last ) ||
 265                                 $last->page_namespace != $row->page_namespace ||
 266                                 $last->page_title     != $row->page_title ) {
 267                                 if( isset( $last ) ) {
 268                                         $output = '';
 269                                         if( $this->dumpUploads ) {
 270                                                 $output .= $this->writer->writeUploads( $last );
 271                                         }
 272                                         $output .= $this->writer->closePage();
 273                                         $this->sink->writeClosePage( $output );
 274                                 }
 275                                 $output = $this->writer->openPage( $row );
 276                                 $this->sink->writeOpenPage( $row, $output );
 277                                 $last = $row;
 278                         }
 279                         $output = $this->writer->writeRevision( $row );
 280                         $this->sink->writeRevision( $row, $output );
 281                 }
 282                 if( isset( $last ) ) {
 283                         $output = '';
 284                         if( $this->dumpUploads ) {
 285                                 $output .= $this->writer->writeUploads( $last );
 286                         }
 287                         $output .= $this->author_list;
 288                         $output .= $this->writer->closePage();
 289                         $this->sink->writeClosePage( $output );
 290                 }
 291                 $resultset->free();
 292         }
 293 }
 294
 295 /**
 296  * @addtogroup Dump
 297  */
 298 class XmlDumpWriter {
 299
 300         /**
 301          * Returns the export schema version.
 302          * @return string
 303          */
 304         function schemaVersion() {
 305                 return "0.3"; // FIXME: upgrade to 0.4 when updated XSD is ready, for the revision deletion bits
 306         }
 307
 308         /**
 309          * Opens the XML output stream's root <mediawiki> element.
 310          * This does not include an xml directive, so is safe to include
 311          * as a subelement in a larger XML stream. Namespace and XML Schema
 312          * references are included.
 313          *
 314          * Output will be encoded in UTF-8.
 315          *
 316          * @return string
 317          */
 318         function openStream() {
 319                 global $wgContLanguageCode;
 320                 $ver = $this->schemaVersion();
 321                 return wfElement( 'mediawiki', array(
 322                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 323                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 324                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 325                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 326                         'version'            => $ver,
 327                         'xml:lang'           => $wgContLanguageCode ),
 328                         null ) .
 329                         "\n" .
 330                         $this->siteInfo();
 331         }
 332
 333         function siteInfo() {
 334                 $info = array(
 335                         $this->sitename(),
 336                         $this->homelink(),
 337                         $this->generator(),
 338                         $this->caseSetting(),
 339                         $this->namespaces() );
 340                 return "  <siteinfo>\n    " .
 341                         implode( "\n    ", $info ) .
 342                         "\n  </siteinfo>\n";
 343         }
 344
 345         function sitename() {
 346                 global $wgSitename;
 347                 return wfElement( 'sitename', array(), $wgSitename );
 348         }
 349
 350         function generator() {
 351                 global $wgVersion;
 352                 return wfElement( 'generator', array(), "MediaWiki $wgVersion" );
 353         }
 354
 355         function homelink() {
 356                 return wfElement( 'base', array(), Title::newMainPage()->getFullUrl() );
 357         }
 358
 359         function caseSetting() {
 360                 global $wgCapitalLinks;
 361                 // "case-insensitive" option is reserved for future
 362                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 363                 return wfElement( 'case', array(), $sensitivity );
 364         }
 365
 366         function namespaces() {
 367                 global $wgContLang;
 368                 $spaces = "  <namespaces>\n";
 369                 foreach( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 370                         $spaces .= '      ' . wfElement( 'namespace', array( 'key' => $ns ), $title ) . "\n";
 371                 }
 372                 $spaces .= "    </namespaces>";
 373                 return $spaces;
 374         }
 375
 376         /**
 377          * Closes the output stream with the closing root element.
 378          * Call when finished dumping things.
 379          */
 380         function closeStream() {
 381                 return "</mediawiki>\n";
 382         }
 383
 384
 385         /**
 386          * Opens a <page> section on the output stream, with data
 387          * from the given database row.
 388          *
 389          * @param object $row
 390          * @return string
 391          * @access private
 392          */
 393         function openPage( $row ) {
 394                 $out = "  <page>\n";
 395                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 396                 $out .= '    ' . wfElementClean( 'title', array(), $title->getPrefixedText() ) . "\n";
 397                 $out .= '    ' . wfElement( 'id', array(), strval( $row->page_id ) ) . "\n";
 398                 if( '' != $row->page_restrictions ) {
 399                         $out .= '    ' . wfElement( 'restrictions', array(),
 400                                 strval( $row->page_restrictions ) ) . "\n";
 401                 }
 402                 return $out;
 403         }
 404
 405         /**
 406          * Closes a <page> section on the output stream.
 407          *
 408          * @access private
 409          */
 410         function closePage() {
 411                 return "  </page>\n";
 412         }
 413
 414         /**
 415          * Dumps a <revision> section on the output stream, with
 416          * data filled in from the given database row.
 417          *
 418          * @param object $row
 419          * @return string
 420          * @access private
 421          */
 422         function writeRevision( $row ) {
 423                 $fname = 'WikiExporter::dumpRev';
 424                 wfProfileIn( $fname );
 425
 426                 $out  = "    <revision>\n";
 427                 $out .= "      " . wfElement( 'id', null, strval( $row->rev_id ) ) . "\n";
 428
 429                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 430
 431                 if( $row->rev_deleted & Revision::DELETED_USER ) {
 432                         $out .= "      " . wfElement( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 433                 } else {
 434                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 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         function writeTimestamp( $timestamp ) {
 468                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 469                 return "      " . wfElement( 'timestamp', null, $ts ) . "\n";
 470         }
 471
 472         function writeContributor( $id, $text ) {
 473                 $out = "      <contributor>\n";
 474                 if( $id ) {
 475                         $out .= "        " . wfElementClean( 'username', null, strval( $text ) ) . "\n";
 476                         $out .= "        " . wfElement( 'id', null, strval( $id ) ) . "\n";
 477                 } else {
 478                         $out .= "        " . wfElementClean( 'ip', null, strval( $text ) ) . "\n";
 479                 }
 480                 $out .= "      </contributor>\n";
 481                 return $out;
 482         }
 483
 484         /**
 485          * Warning! This data is potentially inconsistent. :(
 486          */
 487         function writeUploads( $row ) {
 488                 if( $row->page_namespace == NS_IMAGE ) {
 489                         $img = wfFindFile( $row->page_title );
 490                         if( $img ) {
 491                                 $out = '';
 492                                 foreach( array_reverse( $img->getHistory() ) as $ver ) {
 493                                         $out .= $this->writeUpload( $ver );
 494                                 }
 495                                 $out .= $this->writeUpload( $img );
 496                                 return $out;
 497                         }
 498                 }
 499                 return '';
 500         }
 501
 502         function writeUpload( $file ) {
 503                 return "    <upload>\n" .
 504                         $this->writeTimestamp( $file->getTimestamp() ) .
 505                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 506                         "      " . wfElementClean( 'comment', null, $file->getDescription() ) . "\n" .
 507                         "      " . wfElement( 'filename', null, $file->getName() ) . "\n" .
 508                         "      " . wfElement( 'src', null, $file->getFullUrl() ) . "\n" .
 509                         "      " . wfElement( 'size', null, $file->getSize() ) . "\n" .
 510                         "    </upload>\n";
 511         }
 512
 513 }
 514
 515
 516 /**
 517  * Base class for output stream; prints to stdout or buffer or whereever.
 518  * @addtogroup Dump
 519  */
 520 class DumpOutput {
 521         function writeOpenStream( $string ) {
 522                 $this->write( $string );
 523         }
 524
 525         function writeCloseStream( $string ) {
 526                 $this->write( $string );
 527         }
 528
 529         function writeOpenPage( $page, $string ) {
 530                 $this->write( $string );
 531         }
 532
 533         function writeClosePage( $string ) {
 534                 $this->write( $string );
 535         }
 536
 537         function writeRevision( $rev, $string ) {
 538                 $this->write( $string );
 539         }
 540
 541         /**
 542          * Override to write to a different stream type.
 543          * @return bool
 544          */
 545         function write( $string ) {
 546                 print $string;
 547         }
 548 }
 549
 550 /**
 551  * Stream outputter to send data to a file.
 552  * @addtogroup Dump
 553  */
 554 class DumpFileOutput extends DumpOutput {
 555         var $handle;
 556
 557         function DumpFileOutput( $file ) {
 558                 $this->handle = fopen( $file, "wt" );
 559         }
 560
 561         function write( $string ) {
 562                 fputs( $this->handle, $string );
 563         }
 564 }
 565
 566 /**
 567  * Stream outputter to send data to a file via some filter program.
 568  * Even if compression is available in a library, using a separate
 569  * program can allow us to make use of a multi-processor system.
 570  * @addtogroup Dump
 571  */
 572 class DumpPipeOutput extends DumpFileOutput {
 573         function DumpPipeOutput( $command, $file = null ) {
 574                 if( !is_null( $file ) ) {
 575                         $command .=  " > " . wfEscapeShellArg( $file );
 576                 }
 577                 $this->handle = popen( $command, "w" );
 578         }
 579 }
 580
 581 /**
 582  * Sends dump output via the gzip compressor.
 583  * @addtogroup Dump
 584  */
 585 class DumpGZipOutput extends DumpPipeOutput {
 586         function DumpGZipOutput( $file ) {
 587                 parent::DumpPipeOutput( "gzip", $file );
 588         }
 589 }
 590
 591 /**
 592  * Sends dump output via the bgzip2 compressor.
 593  * @addtogroup Dump
 594  */
 595 class DumpBZip2Output extends DumpPipeOutput {
 596         function DumpBZip2Output( $file ) {
 597                 parent::DumpPipeOutput( "bzip2", $file );
 598         }
 599 }
 600
 601 /**
 602  * Sends dump output via the p7zip compressor.
 603  * @addtogroup Dump
 604  */
 605 class Dump7ZipOutput extends DumpPipeOutput {
 606         function Dump7ZipOutput( $file ) {
 607                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 608                 // Suppress annoying useless crap from p7zip
 609                 // Unfortunately this could suppress real error messages too
 610                 $command .= ' >' . wfGetNull() . ' 2>&1';
 611                 parent::DumpPipeOutput( $command );
 612         }
 613 }
 614
 615
 616
 617 /**
 618  * Dump output filter class.
 619  * This just does output filtering and streaming; XML formatting is done
 620  * higher up, so be careful in what you do.
 621  * @addtogroup Dump
 622  */
 623 class DumpFilter {
 624         function DumpFilter( &$sink ) {
 625                 $this->sink =& $sink;
 626         }
 627
 628         function writeOpenStream( $string ) {
 629                 $this->sink->writeOpenStream( $string );
 630         }
 631
 632         function writeCloseStream( $string ) {
 633                 $this->sink->writeCloseStream( $string );
 634         }
 635
 636         function writeOpenPage( $page, $string ) {
 637                 $this->sendingThisPage = $this->pass( $page, $string );
 638                 if( $this->sendingThisPage ) {
 639                         $this->sink->writeOpenPage( $page, $string );
 640                 }
 641         }
 642
 643         function writeClosePage( $string ) {
 644                 if( $this->sendingThisPage ) {
 645                         $this->sink->writeClosePage( $string );
 646                         $this->sendingThisPage = false;
 647                 }
 648         }
 649
 650         function writeRevision( $rev, $string ) {
 651                 if( $this->sendingThisPage ) {
 652                         $this->sink->writeRevision( $rev, $string );
 653                 }
 654         }
 655
 656         /**
 657          * Override for page-based filter types.
 658          * @return bool
 659          */
 660         function pass( $page ) {
 661                 return true;
 662         }
 663 }
 664
 665 /**
 666  * Simple dump output filter to exclude all talk pages.
 667  * @addtogroup Dump
 668  */
 669 class DumpNotalkFilter extends DumpFilter {
 670         function pass( $page ) {
 671                 return !MWNamespace::isTalk( $page->page_namespace );
 672         }
 673 }
 674
 675 /**
 676  * Dump output filter to include or exclude pages in a given set of namespaces.
 677  * @addtogroup Dump
 678  */
 679 class DumpNamespaceFilter extends DumpFilter {
 680         var $invert = false;
 681         var $namespaces = array();
 682
 683         function DumpNamespaceFilter( &$sink, $param ) {
 684                 parent::DumpFilter( $sink );
 685
 686                 $constants = array(
 687                         "NS_MAIN"           => NS_MAIN,
 688                         "NS_TALK"           => NS_TALK,
 689                         "NS_USER"           => NS_USER,
 690                         "NS_USER_TALK"      => NS_USER_TALK,
 691                         "NS_PROJECT"        => NS_PROJECT,
 692                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
 693                         "NS_IMAGE"          => NS_IMAGE,
 694                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
 695                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
 696                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
 697                         "NS_TEMPLATE"       => NS_TEMPLATE,
 698                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
 699                         "NS_HELP"           => NS_HELP,
 700                         "NS_HELP_TALK"      => NS_HELP_TALK,
 701                         "NS_CATEGORY"       => NS_CATEGORY,
 702                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
 703
 704                 if( $param{0} == '!' ) {
 705                         $this->invert = true;
 706                         $param = substr( $param, 1 );
 707                 }
 708
 709                 foreach( explode( ',', $param ) as $key ) {
 710                         $key = trim( $key );
 711                         if( isset( $constants[$key] ) ) {
 712                                 $ns = $constants[$key];
 713                                 $this->namespaces[$ns] = true;
 714                         } elseif( is_numeric( $key ) ) {
 715                                 $ns = intval( $key );
 716                                 $this->namespaces[$ns] = true;
 717                         } else {
 718                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
 719                         }
 720                 }
 721         }
 722
 723         function pass( $page ) {
 724                 $match = isset( $this->namespaces[$page->page_namespace] );
 725                 return $this->invert xor $match;
 726         }
 727 }
 728
 729
 730 /**
 731  * Dump output filter to include only the last revision in each page sequence.
 732  * @addtogroup Dump
 733  */
 734 class DumpLatestFilter extends DumpFilter {
 735         var $page, $pageString, $rev, $revString;
 736
 737         function writeOpenPage( $page, $string ) {
 738                 $this->page = $page;
 739                 $this->pageString = $string;
 740         }
 741
 742         function writeClosePage( $string ) {
 743                 if( $this->rev ) {
 744                         $this->sink->writeOpenPage( $this->page, $this->pageString );
 745                         $this->sink->writeRevision( $this->rev, $this->revString );
 746                         $this->sink->writeClosePage( $string );
 747                 }
 748                 $this->rev = null;
 749                 $this->revString = null;
 750                 $this->page = null;
 751                 $this->pageString = null;
 752         }
 753
 754         function writeRevision( $rev, $string ) {
 755                 if( $rev->rev_id == $this->page->page_latest ) {
 756                         $this->rev = $rev;
 757                         $this->revString = $string;
 758                 }
 759         }
 760 }
 761
 762 /**
 763  * Base class for output stream; prints to stdout or buffer or whereever.
 764  * @addtogroup Dump
 765  */
 766 class DumpMultiWriter {
 767         function DumpMultiWriter( $sinks ) {
 768                 $this->sinks = $sinks;
 769                 $this->count = count( $sinks );
 770         }
 771
 772         function writeOpenStream( $string ) {
 773                 for( $i = 0; $i < $this->count; $i++ ) {
 774                         $this->sinks[$i]->writeOpenStream( $string );
 775                 }
 776         }
 777
 778         function writeCloseStream( $string ) {
 779                 for( $i = 0; $i < $this->count; $i++ ) {
 780                         $this->sinks[$i]->writeCloseStream( $string );
 781                 }
 782         }
 783
 784         function writeOpenPage( $page, $string ) {
 785                 for( $i = 0; $i < $this->count; $i++ ) {
 786                         $this->sinks[$i]->writeOpenPage( $page, $string );
 787                 }
 788         }
 789
 790         function writeClosePage( $string ) {
 791                 for( $i = 0; $i < $this->count; $i++ ) {
 792                         $this->sinks[$i]->writeClosePage( $string );
 793                 }
 794         }
 795
 796         function writeRevision( $rev, $string ) {
 797                 for( $i = 0; $i < $this->count; $i++ ) {
 798                         $this->sinks[$i]->writeRevision( $rev, $string );
 799                 }
 800         }
 801 }
 802
 803 function xmlsafe( $string ) {
 804         $fname = 'xmlsafe';
 805         wfProfileIn( $fname );
 806
 807         /**
 808          * The page may contain old data which has not been properly normalized.
 809          * Invalid UTF-8 sequences or forbidden control characters will make our
 810          * XML output invalid, so be sure to strip them out.
 811          */
 812         $string = UtfNormal::cleanUp( $string );
 813
 814         $string = htmlspecialchars( $string );
 815         wfProfileOut( $fname );
 816         return $string;
 817 }
 818
 819