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  * @defgroup Dump Dump
  22  */
  23
  24 /**
  25  * @ingroup SpecialPage Dump
  26  */
  27 class WikiExporter {
  28         var $list_authors = false ; # Return distinct author list (when not returning full history)
  29         var $author_list = "" ;
  30
  31         var $dumpUploads = false;
  32
  33         const FULL = 0;
  34         const CURRENT = 1;
  35
  36         const BUFFER = 0;
  37         const STREAM = 1;
  38
  39         const TEXT = 0;
  40         const STUB = 1;
  41
  42         /**
  43          * If using WikiExporter::STREAM to stream a large amount of data,
  44          * provide a database connection which is not managed by
  45          * LoadBalancer to read from: some history blob types will
  46          * make additional queries to pull source data while the
  47          * main query is still running.
  48          *
  49          * @param $db Database
  50          * @param $history Mixed: one of WikiExporter::FULL or WikiExporter::CURRENT,
  51          *                 or an associative array:
  52          *                   offset: non-inclusive offset at which to start the query
  53          *                   limit: maximum number of rows to return
  54          *                   dir: "asc" or "desc" timestamp order
  55          * @param $buffer Int: one of WikiExporter::BUFFER or WikiExporter::STREAM
  56          */
  57         function __construct( &$db, $history = WikiExporter::CURRENT,
  58                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  59                 $this->db =& $db;
  60                 $this->history = $history;
  61                 $this->buffer  = $buffer;
  62                 $this->writer  = new XmlDumpWriter();
  63                 $this->sink    = new DumpOutput();
  64                 $this->text    = $text;
  65         }
  66
  67         /**
  68          * Set the DumpOutput or DumpFilter object which will receive
  69          * various row objects and XML output for filtering. Filters
  70          * can be chained or used as callbacks.
  71          *
  72          * @param $sink mixed
  73          */
  74         function setOutputSink( &$sink ) {
  75                 $this->sink =& $sink;
  76         }
  77
  78         function openStream() {
  79                 $output = $this->writer->openStream();
  80                 $this->sink->writeOpenStream( $output );
  81         }
  82
  83         function closeStream() {
  84                 $output = $this->writer->closeStream();
  85                 $this->sink->writeCloseStream( $output );
  86         }
  87
  88         /**
  89          * Dumps a series of page and revision records for all pages
  90          * in the database, either including complete history or only
  91          * the most recent version.
  92          */
  93         function allPages() {
  94                 return $this->dumpFrom( '' );
  95         }
  96
  97         /**
  98          * Dumps a series of page and revision records for those pages
  99          * in the database falling within the page_id range given.
 100          * @param $start Int: inclusive lower limit (this id is included)
 101          * @param $end   Int: Exclusive upper limit (this id is not included)
 102          *                   If 0, no upper limit.
 103          */
 104         function pagesByRange( $start, $end ) {
 105                 $condition = 'page_id >= ' . intval( $start );
 106                 if( $end ) {
 107                         $condition .= ' AND page_id < ' . intval( $end );
 108                 }
 109                 return $this->dumpFrom( $condition );
 110         }
 111
 112         /**
 113          * @param $title Title
 114          */
 115         function pageByTitle( $title ) {
 116                 return $this->dumpFrom(
 117                         'page_namespace=' . $title->getNamespace() .
 118                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 119         }
 120
 121         function pageByName( $name ) {
 122                 $title = Title::newFromText( $name );
 123                 if( is_null( $title ) ) {
 124                         return new WikiError( "Can't export invalid title" );
 125                 } else {
 126                         return $this->pageByTitle( $title );
 127                 }
 128         }
 129
 130         function pagesByName( $names ) {
 131                 foreach( $names as $name ) {
 132                         $this->pageByName( $name );
 133                 }
 134         }
 135
 136
 137         // -------------------- private implementation below --------------------
 138
 139         # Generates the distinct list of authors of an article
 140         # Not called by default (depends on $this->list_authors)
 141         # Can be set by Special:Export when not exporting whole history
 142         function do_list_authors ( $page , $revision , $cond ) {
 143                 $fname = "do_list_authors" ;
 144                 wfProfileIn( $fname );
 145                 $this->author_list = "<contributors>";
 146                 //rev_deleted
 147                 $nothidden = '(rev_deleted & '.Revision::DELETED_USER.') = 0';
 148
 149                 $sql = "SELECT DISTINCT rev_user_text,rev_user FROM {$page},{$revision} WHERE page_id=rev_page AND $nothidden AND " . $cond ;
 150                 $result = $this->db->query( $sql, $fname );
 151                 $resultset = $this->db->resultObject( $result );
 152                 while( $row = $resultset->fetchObject() ) {
 153                         $this->author_list .= "<contributor>" .
 154                                 "<username>" .
 155                                 htmlentities( $row->rev_user_text )  .
 156                                 "</username>" .
 157                                 "<id>" .
 158                                 $row->rev_user .
 159                                 "</id>" .
 160                                 "</contributor>";
 161                 }
 162                 wfProfileOut( $fname );
 163                 $this->author_list .= "</contributors>";
 164         }
 165
 166         function dumpFrom( $cond = '' ) {
 167                 $fname = 'WikiExporter::dumpFrom';
 168                 wfProfileIn( $fname );
 169
 170                 $page     = $this->db->tableName( 'page' );
 171                 $revision = $this->db->tableName( 'revision' );
 172                 $text     = $this->db->tableName( 'text' );
 173
 174                 $order = 'ORDER BY page_id';
 175                 $limit = '';
 176
 177                 if( $this->history == WikiExporter::FULL ) {
 178                         $join = 'page_id=rev_page';
 179                 } elseif( $this->history == WikiExporter::CURRENT ) {
 180                         if ( $this->list_authors && $cond != '' )  { // List authors, if so desired
 181                                 $this->do_list_authors ( $page , $revision , $cond );
 182                         }
 183                         $join = 'page_id=rev_page AND page_latest=rev_id';
 184                 } elseif ( is_array( $this->history ) ) {
 185                         $join = 'page_id=rev_page';
 186                         if ( $this->history['dir'] == 'asc' ) {
 187                                 $op = '>';
 188                                 $order .= ', rev_timestamp';
 189                         } else {
 190                                 $op = '<';
 191                                 $order .= ', rev_timestamp DESC';
 192                         }
 193                         if ( !empty( $this->history['offset'] ) ) {
 194                                 $join .= " AND rev_timestamp $op " . $this->db->addQuotes(
 195                                         $this->db->timestamp( $this->history['offset'] ) );
 196                         }
 197                         if ( !empty( $this->history['limit'] ) ) {
 198                                 $limitNum = intval( $this->history['limit'] );
 199                                 if ( $limitNum > 0 ) {
 200                                         $limit = "LIMIT $limitNum";
 201                                 }
 202                         }
 203                 } else {
 204                         wfProfileOut( $fname );
 205                         return new WikiError( "$fname given invalid history dump type." );
 206                 }
 207                 $where = ( $cond == '' ) ? '' : "$cond AND";
 208
 209                 if( $this->buffer == WikiExporter::STREAM ) {
 210                         $prev = $this->db->bufferResults( false );
 211                 }
 212                 if( $cond == '' ) {
 213                         // Optimization hack for full-database dump
 214                         $revindex = $pageindex = $this->db->useIndexClause("PRIMARY");
 215                         $straight = ' /*! STRAIGHT_JOIN */ ';
 216                 } else {
 217                         $pageindex = '';
 218                         $revindex = '';
 219                         $straight = '';
 220                 }
 221                 if( $this->text == WikiExporter::STUB ) {
 222                         $sql = "SELECT $straight * FROM
 223                                         $page $pageindex,
 224                                         $revision $revindex
 225                                         WHERE $where $join
 226                                         $order $limit";
 227                 } else {
 228                         $sql = "SELECT $straight * FROM
 229                                         $page $pageindex,
 230                                         $revision $revindex,
 231                                         $text
 232                                         WHERE $where $join AND rev_text_id=old_id
 233                                         $order $limit";
 234                 }
 235                 $result = $this->db->query( $sql, $fname );
 236                 $wrapper = $this->db->resultObject( $result );
 237                 $this->outputStream( $wrapper );
 238
 239                 if ( $this->list_authors ) {
 240                         $this->outputStream( $wrapper );
 241                 }
 242
 243                 if( $this->buffer == WikiExporter::STREAM ) {
 244                         $this->db->bufferResults( $prev );
 245                 }
 246
 247                 wfProfileOut( $fname );
 248         }
 249
 250         /**
 251          * Runs through a query result set dumping page and revision records.
 252          * The result set should be sorted/grouped by page to avoid duplicate
 253          * page records in the output.
 254          *
 255          * The result set will be freed once complete. Should be safe for
 256          * streaming (non-buffered) queries, as long as it was made on a
 257          * separate database connection not managed by LoadBalancer; some
 258          * blob storage types will make queries to pull source data.
 259          *
 260          * @param $resultset ResultWrapper
 261          * @access private
 262          */
 263         function outputStream( $resultset ) {
 264                 $last = null;
 265                 while( $row = $resultset->fetchObject() ) {
 266                         if( is_null( $last ) ||
 267                                 $last->page_namespace != $row->page_namespace ||
 268                                 $last->page_title     != $row->page_title ) {
 269                                 if( isset( $last ) ) {
 270                                         $output = '';
 271                                         if( $this->dumpUploads ) {
 272                                                 $output .= $this->writer->writeUploads( $last );
 273                                         }
 274                                         $output .= $this->writer->closePage();
 275                                         $this->sink->writeClosePage( $output );
 276                                 }
 277                                 $output = $this->writer->openPage( $row );
 278                                 $this->sink->writeOpenPage( $row, $output );
 279                                 $last = $row;
 280                         }
 281                         $output = $this->writer->writeRevision( $row );
 282                         $this->sink->writeRevision( $row, $output );
 283                 }
 284                 if( isset( $last ) ) {
 285                         $output = '';
 286                         if( $this->dumpUploads ) {
 287                                 $output .= $this->writer->writeUploads( $last );
 288                         }
 289                         $output .= $this->author_list;
 290                         $output .= $this->writer->closePage();
 291                         $this->sink->writeClosePage( $output );
 292                 }
 293                 $resultset->free();
 294         }
 295 }
 296
 297 /**
 298  * @ingroup Dump
 299  */
 300 class XmlDumpWriter {
 301
 302         /**
 303          * Returns the export schema version.
 304          * @return string
 305          */
 306         function schemaVersion() {
 307                 return "0.3"; // FIXME: upgrade to 0.4 when updated XSD is ready, for the revision deletion bits
 308         }
 309
 310         /**
 311          * Opens the XML output stream's root <mediawiki> element.
 312          * This does not include an xml directive, so is safe to include
 313          * as a subelement in a larger XML stream. Namespace and XML Schema
 314          * references are included.
 315          *
 316          * Output will be encoded in UTF-8.
 317          *
 318          * @return string
 319          */
 320         function openStream() {
 321                 global $wgContLanguageCode;
 322                 $ver = $this->schemaVersion();
 323                 return wfElement( 'mediawiki', array(
 324                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 325                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 326                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 327                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 328                         'version'            => $ver,
 329                         'xml:lang'           => $wgContLanguageCode ),
 330                         null ) .
 331                         "\n" .
 332                         $this->siteInfo();
 333         }
 334
 335         function siteInfo() {
 336                 $info = array(
 337                         $this->sitename(),
 338                         $this->homelink(),
 339                         $this->generator(),
 340                         $this->caseSetting(),
 341                         $this->namespaces() );
 342                 return "  <siteinfo>\n    " .
 343                         implode( "\n    ", $info ) .
 344                         "\n  </siteinfo>\n";
 345         }
 346
 347         function sitename() {
 348                 global $wgSitename;
 349                 return wfElement( 'sitename', array(), $wgSitename );
 350         }
 351
 352         function generator() {
 353                 global $wgVersion;
 354                 return wfElement( 'generator', array(), "MediaWiki $wgVersion" );
 355         }
 356
 357         function homelink() {
 358                 return wfElement( 'base', array(), Title::newMainPage()->getFullUrl() );
 359         }
 360
 361         function caseSetting() {
 362                 global $wgCapitalLinks;
 363                 // "case-insensitive" option is reserved for future
 364                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 365                 return wfElement( 'case', array(), $sensitivity );
 366         }
 367
 368         function namespaces() {
 369                 global $wgContLang;
 370                 $spaces = "  <namespaces>\n";
 371                 foreach( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 372                         $spaces .= '      ' . wfElement( 'namespace', array( 'key' => $ns ), $title ) . "\n";
 373                 }
 374                 $spaces .= "    </namespaces>";
 375                 return $spaces;
 376         }
 377
 378         /**
 379          * Closes the output stream with the closing root element.
 380          * Call when finished dumping things.
 381          */
 382         function closeStream() {
 383                 return "</mediawiki>\n";
 384         }
 385
 386
 387         /**
 388          * Opens a <page> section on the output stream, with data
 389          * from the given database row.
 390          *
 391          * @param $row object
 392          * @return string
 393          * @access private
 394          */
 395         function openPage( $row ) {
 396                 $out = "  <page>\n";
 397                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 398                 $out .= '    ' . wfElementClean( 'title', array(), $title->getPrefixedText() ) . "\n";
 399                 $out .= '    ' . wfElement( 'id', array(), strval( $row->page_id ) ) . "\n";
 400                 if( '' != $row->page_restrictions ) {
 401                         $out .= '    ' . wfElement( 'restrictions', array(),
 402                                 strval( $row->page_restrictions ) ) . "\n";
 403                 }
 404                 return $out;
 405         }
 406
 407         /**
 408          * Closes a <page> section on the output stream.
 409          *
 410          * @access private
 411          */
 412         function closePage() {
 413                 return "  </page>\n";
 414         }
 415
 416         /**
 417          * Dumps a <revision> section on the output stream, with
 418          * data filled in from the given database row.
 419          *
 420          * @param $row object
 421          * @return string
 422          * @access private
 423          */
 424         function writeRevision( $row ) {
 425                 $fname = 'WikiExporter::dumpRev';
 426                 wfProfileIn( $fname );
 427
 428                 $out  = "    <revision>\n";
 429                 $out .= "      " . wfElement( 'id', null, strval( $row->rev_id ) ) . "\n";
 430
 431                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 432
 433                 if( $row->rev_deleted & Revision::DELETED_USER ) {
 434                         $out .= "      " . wfElement( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 435                 } else {
 436                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 437                 }
 438
 439                 if( $row->rev_minor_edit ) {
 440                         $out .=  "      <minor/>\n";
 441                 }
 442                 if( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 443                         $out .= "      " . wfElement( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 444                 } elseif( $row->rev_comment != '' ) {
 445                         $out .= "      " . wfElementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 446                 }
 447
 448                 if( $row->rev_deleted & Revision::DELETED_TEXT ) {
 449                         $out .= "      " . wfElement( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 450                 } elseif( isset( $row->old_text ) ) {
 451                         // Raw text from the database may have invalid chars
 452                         $text = strval( Revision::getRevisionText( $row ) );
 453                         $out .= "      " . wfElementClean( 'text',
 454                                 array( 'xml:space' => 'preserve' ),
 455                                 strval( $text ) ) . "\n";
 456                 } else {
 457                         // Stub output
 458                         $out .= "      " . wfElement( 'text',
 459                                 array( 'id' => $row->rev_text_id ),
 460                                 "" ) . "\n";
 461                 }
 462
 463                 $out .= "    </revision>\n";
 464
 465                 wfProfileOut( $fname );
 466                 return $out;
 467         }
 468
 469         function writeTimestamp( $timestamp ) {
 470                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 471                 return "      " . wfElement( 'timestamp', null, $ts ) . "\n";
 472         }
 473
 474         function writeContributor( $id, $text ) {
 475                 $out = "      <contributor>\n";
 476                 if( $id ) {
 477                         $out .= "        " . wfElementClean( 'username', null, strval( $text ) ) . "\n";
 478                         $out .= "        " . wfElement( 'id', null, strval( $id ) ) . "\n";
 479                 } else {
 480                         $out .= "        " . wfElementClean( 'ip', null, strval( $text ) ) . "\n";
 481                 }
 482                 $out .= "      </contributor>\n";
 483                 return $out;
 484         }
 485
 486         /**
 487          * Warning! This data is potentially inconsistent. :(
 488          */
 489         function writeUploads( $row ) {
 490                 if( $row->page_namespace == NS_IMAGE ) {
 491                         $img = wfFindFile( $row->page_title );
 492                         if( $img ) {
 493                                 $out = '';
 494                                 foreach( array_reverse( $img->getHistory() ) as $ver ) {
 495                                         $out .= $this->writeUpload( $ver );
 496                                 }
 497                                 $out .= $this->writeUpload( $img );
 498                                 return $out;
 499                         }
 500                 }
 501                 return '';
 502         }
 503
 504         function writeUpload( $file ) {
 505                 return "    <upload>\n" .
 506                         $this->writeTimestamp( $file->getTimestamp() ) .
 507                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 508                         "      " . wfElementClean( 'comment', null, $file->getDescription() ) . "\n" .
 509                         "      " . wfElement( 'filename', null, $file->getName() ) . "\n" .
 510                         "      " . wfElement( 'src', null, $file->getFullUrl() ) . "\n" .
 511                         "      " . wfElement( 'size', null, $file->getSize() ) . "\n" .
 512                         "    </upload>\n";
 513         }
 514
 515 }
 516
 517
 518 /**
 519  * Base class for output stream; prints to stdout or buffer or whereever.
 520  * @ingroup Dump
 521  */
 522 class DumpOutput {
 523         function writeOpenStream( $string ) {
 524                 $this->write( $string );
 525         }
 526
 527         function writeCloseStream( $string ) {
 528                 $this->write( $string );
 529         }
 530
 531         function writeOpenPage( $page, $string ) {
 532                 $this->write( $string );
 533         }
 534
 535         function writeClosePage( $string ) {
 536                 $this->write( $string );
 537         }
 538
 539         function writeRevision( $rev, $string ) {
 540                 $this->write( $string );
 541         }
 542
 543         /**
 544          * Override to write to a different stream type.
 545          * @return bool
 546          */
 547         function write( $string ) {
 548                 print $string;
 549         }
 550 }
 551
 552 /**
 553  * Stream outputter to send data to a file.
 554  * @ingroup Dump
 555  */
 556 class DumpFileOutput extends DumpOutput {
 557         var $handle;
 558
 559         function DumpFileOutput( $file ) {
 560                 $this->handle = fopen( $file, "wt" );
 561         }
 562
 563         function write( $string ) {
 564                 fputs( $this->handle, $string );
 565         }
 566 }
 567
 568 /**
 569  * Stream outputter to send data to a file via some filter program.
 570  * Even if compression is available in a library, using a separate
 571  * program can allow us to make use of a multi-processor system.
 572  * @ingroup Dump
 573  */
 574 class DumpPipeOutput extends DumpFileOutput {
 575         function DumpPipeOutput( $command, $file = null ) {
 576                 if( !is_null( $file ) ) {
 577                         $command .=  " > " . wfEscapeShellArg( $file );
 578                 }
 579                 $this->handle = popen( $command, "w" );
 580         }
 581 }
 582
 583 /**
 584  * Sends dump output via the gzip compressor.
 585  * @ingroup Dump
 586  */
 587 class DumpGZipOutput extends DumpPipeOutput {
 588         function DumpGZipOutput( $file ) {
 589                 parent::DumpPipeOutput( "gzip", $file );
 590         }
 591 }
 592
 593 /**
 594  * Sends dump output via the bgzip2 compressor.
 595  * @ingroup Dump
 596  */
 597 class DumpBZip2Output extends DumpPipeOutput {
 598         function DumpBZip2Output( $file ) {
 599                 parent::DumpPipeOutput( "bzip2", $file );
 600         }
 601 }
 602
 603 /**
 604  * Sends dump output via the p7zip compressor.
 605  * @ingroup Dump
 606  */
 607 class Dump7ZipOutput extends DumpPipeOutput {
 608         function Dump7ZipOutput( $file ) {
 609                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 610                 // Suppress annoying useless crap from p7zip
 611                 // Unfortunately this could suppress real error messages too
 612                 $command .= ' >' . wfGetNull() . ' 2>&1';
 613                 parent::DumpPipeOutput( $command );
 614         }
 615 }
 616
 617
 618
 619 /**
 620  * Dump output filter class.
 621  * This just does output filtering and streaming; XML formatting is done
 622  * higher up, so be careful in what you do.
 623  * @ingroup Dump
 624  */
 625 class DumpFilter {
 626         function DumpFilter( &$sink ) {
 627                 $this->sink =& $sink;
 628         }
 629
 630         function writeOpenStream( $string ) {
 631                 $this->sink->writeOpenStream( $string );
 632         }
 633
 634         function writeCloseStream( $string ) {
 635                 $this->sink->writeCloseStream( $string );
 636         }
 637
 638         function writeOpenPage( $page, $string ) {
 639                 $this->sendingThisPage = $this->pass( $page, $string );
 640                 if( $this->sendingThisPage ) {
 641                         $this->sink->writeOpenPage( $page, $string );
 642                 }
 643         }
 644
 645         function writeClosePage( $string ) {
 646                 if( $this->sendingThisPage ) {
 647                         $this->sink->writeClosePage( $string );
 648                         $this->sendingThisPage = false;
 649                 }
 650         }
 651
 652         function writeRevision( $rev, $string ) {
 653                 if( $this->sendingThisPage ) {
 654                         $this->sink->writeRevision( $rev, $string );
 655                 }
 656         }
 657
 658         /**
 659          * Override for page-based filter types.
 660          * @return bool
 661          */
 662         function pass( $page ) {
 663                 return true;
 664         }
 665 }
 666
 667 /**
 668  * Simple dump output filter to exclude all talk pages.
 669  * @ingroup Dump
 670  */
 671 class DumpNotalkFilter extends DumpFilter {
 672         function pass( $page ) {
 673                 return !MWNamespace::isTalk( $page->page_namespace );
 674         }
 675 }
 676
 677 /**
 678  * Dump output filter to include or exclude pages in a given set of namespaces.
 679  * @ingroup Dump
 680  */
 681 class DumpNamespaceFilter extends DumpFilter {
 682         var $invert = false;
 683         var $namespaces = array();
 684
 685         function DumpNamespaceFilter( &$sink, $param ) {
 686                 parent::DumpFilter( $sink );
 687
 688                 $constants = array(
 689                         "NS_MAIN"           => NS_MAIN,
 690                         "NS_TALK"           => NS_TALK,
 691                         "NS_USER"           => NS_USER,
 692                         "NS_USER_TALK"      => NS_USER_TALK,
 693                         "NS_PROJECT"        => NS_PROJECT,
 694                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
 695                         "NS_IMAGE"          => NS_IMAGE,
 696                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
 697                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
 698                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
 699                         "NS_TEMPLATE"       => NS_TEMPLATE,
 700                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
 701                         "NS_HELP"           => NS_HELP,
 702                         "NS_HELP_TALK"      => NS_HELP_TALK,
 703                         "NS_CATEGORY"       => NS_CATEGORY,
 704                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
 705
 706                 if( $param{0} == '!' ) {
 707                         $this->invert = true;
 708                         $param = substr( $param, 1 );
 709                 }
 710
 711                 foreach( explode( ',', $param ) as $key ) {
 712                         $key = trim( $key );
 713                         if( isset( $constants[$key] ) ) {
 714                                 $ns = $constants[$key];
 715                                 $this->namespaces[$ns] = true;
 716                         } elseif( is_numeric( $key ) ) {
 717                                 $ns = intval( $key );
 718                                 $this->namespaces[$ns] = true;
 719                         } else {
 720                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
 721                         }
 722                 }
 723         }
 724
 725         function pass( $page ) {
 726                 $match = isset( $this->namespaces[$page->page_namespace] );
 727                 return $this->invert xor $match;
 728         }
 729 }
 730
 731
 732 /**
 733  * Dump output filter to include only the last revision in each page sequence.
 734  * @ingroup Dump
 735  */
 736 class DumpLatestFilter extends DumpFilter {
 737         var $page, $pageString, $rev, $revString;
 738
 739         function writeOpenPage( $page, $string ) {
 740                 $this->page = $page;
 741                 $this->pageString = $string;
 742         }
 743
 744         function writeClosePage( $string ) {
 745                 if( $this->rev ) {
 746                         $this->sink->writeOpenPage( $this->page, $this->pageString );
 747                         $this->sink->writeRevision( $this->rev, $this->revString );
 748                         $this->sink->writeClosePage( $string );
 749                 }
 750                 $this->rev = null;
 751                 $this->revString = null;
 752                 $this->page = null;
 753                 $this->pageString = null;
 754         }
 755
 756         function writeRevision( $rev, $string ) {
 757                 if( $rev->rev_id == $this->page->page_latest ) {
 758                         $this->rev = $rev;
 759                         $this->revString = $string;
 760                 }
 761         }
 762 }
 763
 764 /**
 765  * Base class for output stream; prints to stdout or buffer or whereever.
 766  * @ingroup Dump
 767  */
 768 class DumpMultiWriter {
 769         function DumpMultiWriter( $sinks ) {
 770                 $this->sinks = $sinks;
 771                 $this->count = count( $sinks );
 772         }
 773
 774         function writeOpenStream( $string ) {
 775                 for( $i = 0; $i < $this->count; $i++ ) {
 776                         $this->sinks[$i]->writeOpenStream( $string );
 777                 }
 778         }
 779
 780         function writeCloseStream( $string ) {
 781                 for( $i = 0; $i < $this->count; $i++ ) {
 782                         $this->sinks[$i]->writeCloseStream( $string );
 783                 }
 784         }
 785
 786         function writeOpenPage( $page, $string ) {
 787                 for( $i = 0; $i < $this->count; $i++ ) {
 788                         $this->sinks[$i]->writeOpenPage( $page, $string );
 789                 }
 790         }
 791
 792         function writeClosePage( $string ) {
 793                 for( $i = 0; $i < $this->count; $i++ ) {
 794                         $this->sinks[$i]->writeClosePage( $string );
 795                 }
 796         }
 797
 798         function writeRevision( $rev, $string ) {
 799                 for( $i = 0; $i < $this->count; $i++ ) {
 800                         $this->sinks[$i]->writeRevision( $rev, $string );
 801                 }
 802         }
 803 }
 804
 805 function xmlsafe( $string ) {
 806         $fname = 'xmlsafe';
 807         wfProfileIn( $fname );
 808
 809         /**
 810          * The page may contain old data which has not been properly normalized.
 811          * Invalid UTF-8 sequences or forbidden control characters will make our
 812          * XML output invalid, so be sure to strip them out.
 813          */
 814         $string = UtfNormal::cleanUp( $string );
 815
 816         $string = htmlspecialchars( $string );
 817         wfProfileOut( $fname );
 818         return $string;
 819 }