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         const LOGS = 2;
  36
  37         const BUFFER = 0;
  38         const STREAM = 1;
  39
  40         const TEXT = 0;
  41         const STUB = 1;
  42
  43         /**
  44          * If using WikiExporter::STREAM to stream a large amount of data,
  45          * provide a database connection which is not managed by
  46          * LoadBalancer to read from: some history blob types will
  47          * make additional queries to pull source data while the
  48          * main query is still running.
  49          *
  50          * @param $db Database
  51          * @param $history Mixed: one of WikiExporter::FULL or WikiExporter::CURRENT,
  52          *                 or an associative array:
  53          *                   offset: non-inclusive offset at which to start the query
  54          *                   limit: maximum number of rows to return
  55          *                   dir: "asc" or "desc" timestamp order
  56          * @param $buffer Int: one of WikiExporter::BUFFER or WikiExporter::STREAM
  57          */
  58         function __construct( &$db, $history = WikiExporter::CURRENT,
  59                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  60                 $this->db =& $db;
  61                 $this->history = $history;
  62                 $this->buffer  = $buffer;
  63                 $this->writer  = new XmlDumpWriter();
  64                 $this->sink    = new DumpOutput();
  65                 $this->text    = $text;
  66         }
  67
  68         /**
  69          * Set the DumpOutput or DumpFilter object which will receive
  70          * various row objects and XML output for filtering. Filters
  71          * can be chained or used as callbacks.
  72          *
  73          * @param $sink mixed
  74          */
  75         public function setOutputSink( &$sink ) {
  76                 $this->sink =& $sink;
  77         }
  78
  79         public function openStream() {
  80                 $output = $this->writer->openStream();
  81                 $this->sink->writeOpenStream( $output );
  82         }
  83
  84         public function closeStream() {
  85                 $output = $this->writer->closeStream();
  86                 $this->sink->writeCloseStream( $output );
  87         }
  88
  89         /**
  90          * Dumps a series of page and revision records for all pages
  91          * in the database, either including complete history or only
  92          * the most recent version.
  93          */
  94         public function allPages() {
  95                 return $this->dumpFrom( '' );
  96         }
  97
  98         /**
  99          * Dumps a series of page and revision records for those pages
 100          * in the database falling within the page_id range given.
 101          * @param $start Int: inclusive lower limit (this id is included)
 102          * @param $end   Int: Exclusive upper limit (this id is not included)
 103          *                   If 0, no upper limit.
 104          */
 105         public function pagesByRange( $start, $end ) {
 106                 $condition = 'page_id >= ' . intval( $start );
 107                 if( $end ) {
 108                         $condition .= ' AND page_id < ' . intval( $end );
 109                 }
 110                 return $this->dumpFrom( $condition );
 111         }
 112
 113         /**
 114          * @param $title Title
 115          */
 116         public function pageByTitle( $title ) {
 117                 return $this->dumpFrom(
 118                         'page_namespace=' . $title->getNamespace() .
 119                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 120         }
 121
 122         public function pageByName( $name ) {
 123                 $title = Title::newFromText( $name );
 124                 if( is_null( $title ) ) {
 125                         return new WikiError( "Can't export invalid title" );
 126                 } else {
 127                         return $this->pageByTitle( $title );
 128                 }
 129         }
 130
 131         public function pagesByName( $names ) {
 132                 foreach( $names as $name ) {
 133                         $this->pageByName( $name );
 134                 }
 135         }
 136
 137         public function allLogs() {
 138                 return $this->dumpFrom( '' );
 139         }
 140
 141         public function logsByRange( $start, $end ) {
 142                 $condition = 'log_id >= ' . intval( $start );
 143                 if( $end ) {
 144                         $condition .= ' AND log_id < ' . intval( $end );
 145                 }
 146                 return $this->dumpFrom( $condition );
 147         }
 148
 149         # Generates the distinct list of authors of an article
 150         # Not called by default (depends on $this->list_authors)
 151         # Can be set by Special:Export when not exporting whole history
 152         protected function do_list_authors( $page , $revision , $cond ) {
 153                 $fname = "do_list_authors" ;
 154                 wfProfileIn( $fname );
 155                 $this->author_list = "<contributors>";
 156                 //rev_deleted
 157                 $nothidden = '(rev_deleted & '.Revision::DELETED_USER.') = 0';
 158
 159                 $sql = "SELECT DISTINCT rev_user_text,rev_user FROM {$page},{$revision}
 160                 WHERE page_id=rev_page AND $nothidden AND " . $cond ;
 161                 $result = $this->db->query( $sql, $fname );
 162                 $resultset = $this->db->resultObject( $result );
 163                 while( $row = $resultset->fetchObject() ) {
 164                         $this->author_list .= "<contributor>" .
 165                                 "<username>" .
 166                                 htmlentities( $row->rev_user_text )  .
 167                                 "</username>" .
 168                                 "<id>" .
 169                                 $row->rev_user .
 170                                 "</id>" .
 171                                 "</contributor>";
 172                 }
 173                 wfProfileOut( $fname );
 174                 $this->author_list .= "</contributors>";
 175         }
 176
 177         protected function dumpFrom( $cond = '' ) {
 178                 $fname = 'WikiExporter::dumpFrom';
 179                 wfProfileIn( $fname );
 180
 181                 # For logs dumps...
 182                 if( $this->history & self::LOGS ) {
 183                         $where = array( 'user_id = log_user' );
 184                         # Hide private logs
 185                         $where[] = LogEventsList::getExcludeClause( $this->db );
 186                         if( $cond ) $where[] = $cond;
 187                         $result = $this->db->select( array('logging','user'),
 188                                 '*',
 189                                 $where,
 190                                 $fname,
 191                                 array( 'ORDER BY' => 'log_id', 'USE INDEX' => array('logging' => 'PRIMARY') )
 192                         );
 193                         $wrapper = $this->db->resultObject( $result );
 194                         $this->outputLogStream( $wrapper );
 195                 # For page dumps...
 196                 } else {
 197                         list($page,$revision,$text) = $this->db->tableNamesN('page','revision','text');
 198
 199                         $order = 'ORDER BY page_id';
 200                         $limit = '';
 201
 202                         if( $this->history == WikiExporter::FULL ) {
 203                                 $join = 'page_id=rev_page';
 204                         } elseif( $this->history == WikiExporter::CURRENT ) {
 205                                 if ( $this->list_authors && $cond != '' )  { // List authors, if so desired
 206                                         $this->do_list_authors ( $page , $revision , $cond );
 207                                 }
 208                                 $join = 'page_id=rev_page AND page_latest=rev_id';
 209                         } elseif ( is_array( $this->history ) ) {
 210                                 $join = 'page_id=rev_page';
 211                                 if ( $this->history['dir'] == 'asc' ) {
 212                                         $op = '>';
 213                                         $order .= ', rev_timestamp';
 214                                 } else {
 215                                         $op = '<';
 216                                         $order .= ', rev_timestamp DESC';
 217                                 }
 218                                 if ( !empty( $this->history['offset'] ) ) {
 219                                         $join .= " AND rev_timestamp $op " . $this->db->addQuotes(
 220                                                 $this->db->timestamp( $this->history['offset'] ) );
 221                                 }
 222                                 if ( !empty( $this->history['limit'] ) ) {
 223                                         $limitNum = intval( $this->history['limit'] );
 224                                         if ( $limitNum > 0 ) {
 225                                                 $limit = "LIMIT $limitNum";
 226                                         }
 227                                 }
 228                         } else {
 229                                 wfProfileOut( $fname );
 230                                 return new WikiError( "$fname given invalid history dump type." );
 231                         }
 232                         $where = ( $cond == '' ) ? '' : "$cond AND";
 233
 234                         if( $this->buffer == WikiExporter::STREAM ) {
 235                                 $prev = $this->db->bufferResults( false );
 236                         }
 237                         if( $cond == '' ) {
 238                                 // Optimization hack for full-database dump
 239                                 $revindex = $pageindex = $this->db->useIndexClause("PRIMARY");
 240                                 $straight = ' /*! STRAIGHT_JOIN */ ';
 241                         } else {
 242                                 $pageindex = '';
 243                                 $revindex = '';
 244                                 $straight = '';
 245                         }
 246                         if( $this->text == WikiExporter::STUB ) {
 247                                 $sql = "SELECT $straight * FROM
 248                                         $page $pageindex,
 249                                         $revision $revindex
 250                                         WHERE $where $join
 251                                         $order $limit";
 252                         } else {
 253                                 $sql = "SELECT $straight * FROM
 254                                         $page $pageindex,
 255                                         $revision $revindex,
 256                                         $text
 257                                         WHERE $where $join AND rev_text_id=old_id
 258                                         $order $limit";
 259                         }
 260                         $result = $this->db->query( $sql, $fname );
 261                         $wrapper = $this->db->resultObject( $result );
 262                         $this->outputPageStream( $wrapper );
 263
 264                         if ( $this->list_authors ) {
 265                                 $this->outputPageStream( $wrapper );
 266                         }
 267
 268                         if( $this->buffer == WikiExporter::STREAM ) {
 269                                 $this->db->bufferResults( $prev );
 270                         }
 271                 }
 272                 wfProfileOut( $fname );
 273         }
 274
 275         /**
 276          * Runs through a query result set dumping page and revision records.
 277          * The result set should be sorted/grouped by page to avoid duplicate
 278          * page records in the output.
 279          *
 280          * The result set will be freed once complete. Should be safe for
 281          * streaming (non-buffered) queries, as long as it was made on a
 282          * separate database connection not managed by LoadBalancer; some
 283          * blob storage types will make queries to pull source data.
 284          *
 285          * @param $resultset ResultWrapper
 286          */
 287         protected function outputPageStream( $resultset ) {
 288                 $last = null;
 289                 while( $row = $resultset->fetchObject() ) {
 290                         if( is_null( $last ) ||
 291                                 $last->page_namespace != $row->page_namespace ||
 292                                 $last->page_title     != $row->page_title ) {
 293                                 if( isset( $last ) ) {
 294                                         $output = '';
 295                                         if( $this->dumpUploads ) {
 296                                                 $output .= $this->writer->writeUploads( $last );
 297                                         }
 298                                         $output .= $this->writer->closePage();
 299                                         $this->sink->writeClosePage( $output );
 300                                 }
 301                                 $output = $this->writer->openPage( $row );
 302                                 $this->sink->writeOpenPage( $row, $output );
 303                                 $last = $row;
 304                         }
 305                         $output = $this->writer->writeRevision( $row );
 306                         $this->sink->writeRevision( $row, $output );
 307                 }
 308                 if( isset( $last ) ) {
 309                         $output = '';
 310                         if( $this->dumpUploads ) {
 311                                 $output .= $this->writer->writeUploads( $last );
 312                         }
 313                         $output .= $this->author_list;
 314                         $output .= $this->writer->closePage();
 315                         $this->sink->writeClosePage( $output );
 316                 }
 317                 $resultset->free();
 318         }
 319
 320         protected function outputLogStream( $resultset ) {
 321                 while( $row = $resultset->fetchObject() ) {
 322                         $output = $this->writer->writeLogItem( $row );
 323                         $this->sink->writeLogItem( $row, $output );
 324                 }
 325                 $resultset->free();
 326         }
 327 }
 328
 329 /**
 330  * @ingroup Dump
 331  */
 332 class XmlDumpWriter {
 333
 334         /**
 335          * Returns the export schema version.
 336          * @return string
 337          */
 338         function schemaVersion() {
 339                 return "0.3"; // FIXME: upgrade to 0.4 when updated XSD is ready, for the revision deletion bits
 340         }
 341
 342         /**
 343          * Opens the XML output stream's root <mediawiki> element.
 344          * This does not include an xml directive, so is safe to include
 345          * as a subelement in a larger XML stream. Namespace and XML Schema
 346          * references are included.
 347          *
 348          * Output will be encoded in UTF-8.
 349          *
 350          * @return string
 351          */
 352         function openStream() {
 353                 global $wgContLanguageCode;
 354                 $ver = $this->schemaVersion();
 355                 return wfElement( 'mediawiki', array(
 356                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 357                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 358                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 359                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 360                         'version'            => $ver,
 361                         'xml:lang'           => $wgContLanguageCode ),
 362                         null ) .
 363                         "\n" .
 364                         $this->siteInfo();
 365         }
 366
 367         function siteInfo() {
 368                 $info = array(
 369                         $this->sitename(),
 370                         $this->homelink(),
 371                         $this->generator(),
 372                         $this->caseSetting(),
 373                         $this->namespaces() );
 374                 return "  <siteinfo>\n    " .
 375                         implode( "\n    ", $info ) .
 376                         "\n  </siteinfo>\n";
 377         }
 378
 379         function sitename() {
 380                 global $wgSitename;
 381                 return wfElement( 'sitename', array(), $wgSitename );
 382         }
 383
 384         function generator() {
 385                 global $wgVersion;
 386                 return wfElement( 'generator', array(), "MediaWiki $wgVersion" );
 387         }
 388
 389         function homelink() {
 390                 return wfElement( 'base', array(), Title::newMainPage()->getFullUrl() );
 391         }
 392
 393         function caseSetting() {
 394                 global $wgCapitalLinks;
 395                 // "case-insensitive" option is reserved for future
 396                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 397                 return wfElement( 'case', array(), $sensitivity );
 398         }
 399
 400         function namespaces() {
 401                 global $wgContLang;
 402                 $spaces = "  <namespaces>\n";
 403                 foreach( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 404                         $spaces .= '      ' . wfElement( 'namespace', array( 'key' => $ns ), $title ) . "\n";
 405                 }
 406                 $spaces .= "    </namespaces>";
 407                 return $spaces;
 408         }
 409
 410         /**
 411          * Closes the output stream with the closing root element.
 412          * Call when finished dumping things.
 413          */
 414         function closeStream() {
 415                 return "</mediawiki>\n";
 416         }
 417
 418
 419         /**
 420          * Opens a <page> section on the output stream, with data
 421          * from the given database row.
 422          *
 423          * @param $row object
 424          * @return string
 425          * @access private
 426          */
 427         function openPage( $row ) {
 428                 $out = "  <page>\n";
 429                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 430                 $out .= '    ' . wfElementClean( 'title', array(), $title->getPrefixedText() ) . "\n";
 431                 $out .= '    ' . wfElement( 'id', array(), strval( $row->page_id ) ) . "\n";
 432                 if( '' != $row->page_restrictions ) {
 433                         $out .= '    ' . wfElement( 'restrictions', array(),
 434                                 strval( $row->page_restrictions ) ) . "\n";
 435                 }
 436                 return $out;
 437         }
 438
 439         /**
 440          * Closes a <page> section on the output stream.
 441          *
 442          * @access private
 443          */
 444         function closePage() {
 445                 return "  </page>\n";
 446         }
 447
 448         /**
 449          * Dumps a <revision> section on the output stream, with
 450          * data filled in from the given database row.
 451          *
 452          * @param $row object
 453          * @return string
 454          * @access private
 455          */
 456         function writeRevision( $row ) {
 457                 $fname = 'WikiExporter::dumpRev';
 458                 wfProfileIn( $fname );
 459
 460                 $out  = "    <revision>\n";
 461                 $out .= "      " . wfElement( 'id', null, strval( $row->rev_id ) ) . "\n";
 462
 463                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 464
 465                 if( $row->rev_deleted & Revision::DELETED_USER ) {
 466                         $out .= "      " . wfElement( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 467                 } else {
 468                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 469                 }
 470
 471                 if( $row->rev_minor_edit ) {
 472                         $out .=  "      <minor/>\n";
 473                 }
 474                 if( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 475                         $out .= "      " . wfElement( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 476                 } elseif( $row->rev_comment != '' ) {
 477                         $out .= "      " . wfElementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 478                 }
 479
 480                 if( $row->rev_deleted & Revision::DELETED_TEXT ) {
 481                         $out .= "      " . wfElement( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 482                 } elseif( isset( $row->old_text ) ) {
 483                         // Raw text from the database may have invalid chars
 484                         $text = strval( Revision::getRevisionText( $row ) );
 485                         $out .= "      " . wfElementClean( 'text',
 486                                 array( 'xml:space' => 'preserve' ),
 487                                 strval( $text ) ) . "\n";
 488                 } else {
 489                         // Stub output
 490                         $out .= "      " . wfElement( 'text',
 491                                 array( 'id' => $row->rev_text_id ),
 492                                 "" ) . "\n";
 493                 }
 494
 495                 $out .= "    </revision>\n";
 496
 497                 wfProfileOut( $fname );
 498                 return $out;
 499         }
 500
 501         /**
 502          * Dumps a <logitem> section on the output stream, with
 503          * data filled in from the given database row.
 504          *
 505          * @param $row object
 506          * @return string
 507          * @access private
 508          */
 509         function writeLogItem( $row ) {
 510                 $fname = 'WikiExporter::writeLogItem';
 511                 wfProfileIn( $fname );
 512
 513                 $out  = "    <logitem>\n";
 514                 $out .= "      " . wfElement( 'id', null, strval( $row->log_id ) ) . "\n";
 515
 516                 $out .= $this->writeTimestamp( $row->log_timestamp );
 517
 518                 if( $row->log_deleted & LogPage::DELETED_USER ) {
 519                         $out .= "      " . wfElement( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 520                 } else {
 521                         $out .= $this->writeContributor( $row->log_user, $row->user_name );
 522                 }
 523
 524                 if( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 525                         $out .= "      " . wfElement( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 526                 } elseif( $row->log_comment != '' ) {
 527                         $out .= "      " . wfElementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 528                 }
 529
 530                 $out .= "      " . wfElement( 'type', null, strval( $row->log_type ) ) . "\n";
 531                 $out .= "      " . wfElement( 'action', null, strval( $row->log_action ) ) . "\n";
 532
 533                 if( $row->log_deleted & LogPage::DELETED_ACTION ) {
 534                         $out .= "      " . wfElement( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 535                 } else {
 536                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 537                         $out .= "      " . wfElementClean( 'logtitle', null, $title->getPrefixedText() ) . "\n";
 538                         $out .= "      " . wfElementClean( 'params',
 539                                 array( 'xml:space' => 'preserve' ),
 540                                 strval( $row->log_params ) ) . "\n";
 541                 }
 542
 543                 $out .= "    </logitem>\n";
 544
 545                 wfProfileOut( $fname );
 546                 return $out;
 547         }
 548
 549         function writeTimestamp( $timestamp ) {
 550                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 551                 return "      " . wfElement( 'timestamp', null, $ts ) . "\n";
 552         }
 553
 554         function writeContributor( $id, $text ) {
 555                 $out = "      <contributor>\n";
 556                 if( $id ) {
 557                         $out .= "        " . wfElementClean( 'username', null, strval( $text ) ) . "\n";
 558                         $out .= "        " . wfElement( 'id', null, strval( $id ) ) . "\n";
 559                 } else {
 560                         $out .= "        " . wfElementClean( 'ip', null, strval( $text ) ) . "\n";
 561                 }
 562                 $out .= "      </contributor>\n";
 563                 return $out;
 564         }
 565
 566         /**
 567          * Warning! This data is potentially inconsistent. :(
 568          */
 569         function writeUploads( $row ) {
 570                 if( $row->page_namespace == NS_IMAGE ) {
 571                         $img = wfFindFile( $row->page_title );
 572                         if( $img ) {
 573                                 $out = '';
 574                                 foreach( array_reverse( $img->getHistory() ) as $ver ) {
 575                                         $out .= $this->writeUpload( $ver );
 576                                 }
 577                                 $out .= $this->writeUpload( $img );
 578                                 return $out;
 579                         }
 580                 }
 581                 return '';
 582         }
 583
 584         function writeUpload( $file ) {
 585                 return "    <upload>\n" .
 586                         $this->writeTimestamp( $file->getTimestamp() ) .
 587                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 588                         "      " . wfElementClean( 'comment', null, $file->getDescription() ) . "\n" .
 589                         "      " . wfElement( 'filename', null, $file->getName() ) . "\n" .
 590                         "      " . wfElement( 'src', null, $file->getFullUrl() ) . "\n" .
 591                         "      " . wfElement( 'size', null, $file->getSize() ) . "\n" .
 592                         "    </upload>\n";
 593         }
 594
 595 }
 596
 597
 598 /**
 599  * Base class for output stream; prints to stdout or buffer or whereever.
 600  * @ingroup Dump
 601  */
 602 class DumpOutput {
 603         function writeOpenStream( $string ) {
 604                 $this->write( $string );
 605         }
 606
 607         function writeCloseStream( $string ) {
 608                 $this->write( $string );
 609         }
 610
 611         function writeOpenPage( $page, $string ) {
 612                 $this->write( $string );
 613         }
 614
 615         function writeClosePage( $string ) {
 616                 $this->write( $string );
 617         }
 618
 619         function writeRevision( $rev, $string ) {
 620                 $this->write( $string );
 621         }
 622
 623         function writeLogItem( $rev, $string ) {
 624                 $this->write( $string );
 625         }
 626
 627         /**
 628          * Override to write to a different stream type.
 629          * @return bool
 630          */
 631         function write( $string ) {
 632                 print $string;
 633         }
 634 }
 635
 636 /**
 637  * Stream outputter to send data to a file.
 638  * @ingroup Dump
 639  */
 640 class DumpFileOutput extends DumpOutput {
 641         var $handle;
 642
 643         function DumpFileOutput( $file ) {
 644                 $this->handle = fopen( $file, "wt" );
 645         }
 646
 647         function write( $string ) {
 648                 fputs( $this->handle, $string );
 649         }
 650 }
 651
 652 /**
 653  * Stream outputter to send data to a file via some filter program.
 654  * Even if compression is available in a library, using a separate
 655  * program can allow us to make use of a multi-processor system.
 656  * @ingroup Dump
 657  */
 658 class DumpPipeOutput extends DumpFileOutput {
 659         function DumpPipeOutput( $command, $file = null ) {
 660                 if( !is_null( $file ) ) {
 661                         $command .=  " > " . wfEscapeShellArg( $file );
 662                 }
 663                 $this->handle = popen( $command, "w" );
 664         }
 665 }
 666
 667 /**
 668  * Sends dump output via the gzip compressor.
 669  * @ingroup Dump
 670  */
 671 class DumpGZipOutput extends DumpPipeOutput {
 672         function DumpGZipOutput( $file ) {
 673                 parent::DumpPipeOutput( "gzip", $file );
 674         }
 675 }
 676
 677 /**
 678  * Sends dump output via the bgzip2 compressor.
 679  * @ingroup Dump
 680  */
 681 class DumpBZip2Output extends DumpPipeOutput {
 682         function DumpBZip2Output( $file ) {
 683                 parent::DumpPipeOutput( "bzip2", $file );
 684         }
 685 }
 686
 687 /**
 688  * Sends dump output via the p7zip compressor.
 689  * @ingroup Dump
 690  */
 691 class Dump7ZipOutput extends DumpPipeOutput {
 692         function Dump7ZipOutput( $file ) {
 693                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 694                 // Suppress annoying useless crap from p7zip
 695                 // Unfortunately this could suppress real error messages too
 696                 $command .= ' >' . wfGetNull() . ' 2>&1';
 697                 parent::DumpPipeOutput( $command );
 698         }
 699 }
 700
 701
 702
 703 /**
 704  * Dump output filter class.
 705  * This just does output filtering and streaming; XML formatting is done
 706  * higher up, so be careful in what you do.
 707  * @ingroup Dump
 708  */
 709 class DumpFilter {
 710         function DumpFilter( &$sink ) {
 711                 $this->sink =& $sink;
 712         }
 713
 714         function writeOpenStream( $string ) {
 715                 $this->sink->writeOpenStream( $string );
 716         }
 717
 718         function writeCloseStream( $string ) {
 719                 $this->sink->writeCloseStream( $string );
 720         }
 721
 722         function writeOpenPage( $page, $string ) {
 723                 $this->sendingThisPage = $this->pass( $page, $string );
 724                 if( $this->sendingThisPage ) {
 725                         $this->sink->writeOpenPage( $page, $string );
 726                 }
 727         }
 728
 729         function writeClosePage( $string ) {
 730                 if( $this->sendingThisPage ) {
 731                         $this->sink->writeClosePage( $string );
 732                         $this->sendingThisPage = false;
 733                 }
 734         }
 735
 736         function writeRevision( $rev, $string ) {
 737                 if( $this->sendingThisPage ) {
 738                         $this->sink->writeRevision( $rev, $string );
 739                 }
 740         }
 741
 742         function writeLogItem( $rev, $string ) {
 743                 $this->sink->writeRevision( $rev, $string );
 744         }
 745
 746         /**
 747          * Override for page-based filter types.
 748          * @return bool
 749          */
 750         function pass( $page ) {
 751                 return true;
 752         }
 753 }
 754
 755 /**
 756  * Simple dump output filter to exclude all talk pages.
 757  * @ingroup Dump
 758  */
 759 class DumpNotalkFilter extends DumpFilter {
 760         function pass( $page ) {
 761                 return !MWNamespace::isTalk( $page->page_namespace );
 762         }
 763 }
 764
 765 /**
 766  * Dump output filter to include or exclude pages in a given set of namespaces.
 767  * @ingroup Dump
 768  */
 769 class DumpNamespaceFilter extends DumpFilter {
 770         var $invert = false;
 771         var $namespaces = array();
 772
 773         function DumpNamespaceFilter( &$sink, $param ) {
 774                 parent::DumpFilter( $sink );
 775
 776                 $constants = array(
 777                         "NS_MAIN"           => NS_MAIN,
 778                         "NS_TALK"           => NS_TALK,
 779                         "NS_USER"           => NS_USER,
 780                         "NS_USER_TALK"      => NS_USER_TALK,
 781                         "NS_PROJECT"        => NS_PROJECT,
 782                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
 783                         "NS_FILE"           => NS_FILE,
 784                         "NS_FILE_TALK"      => NS_FILE_TALK,
 785                         "NS_IMAGE"          => NS_IMAGE,  // NS_IMAGE is an alias for NS_FILE
 786                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
 787                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
 788                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
 789                         "NS_TEMPLATE"       => NS_TEMPLATE,
 790                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
 791                         "NS_HELP"           => NS_HELP,
 792                         "NS_HELP_TALK"      => NS_HELP_TALK,
 793                         "NS_CATEGORY"       => NS_CATEGORY,
 794                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
 795
 796                 if( $param{0} == '!' ) {
 797                         $this->invert = true;
 798                         $param = substr( $param, 1 );
 799                 }
 800
 801                 foreach( explode( ',', $param ) as $key ) {
 802                         $key = trim( $key );
 803                         if( isset( $constants[$key] ) ) {
 804                                 $ns = $constants[$key];
 805                                 $this->namespaces[$ns] = true;
 806                         } elseif( is_numeric( $key ) ) {
 807                                 $ns = intval( $key );
 808                                 $this->namespaces[$ns] = true;
 809                         } else {
 810                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
 811                         }
 812                 }
 813         }
 814
 815         function pass( $page ) {
 816                 $match = isset( $this->namespaces[$page->page_namespace] );
 817                 return $this->invert xor $match;
 818         }
 819 }
 820
 821
 822 /**
 823  * Dump output filter to include only the last revision in each page sequence.
 824  * @ingroup Dump
 825  */
 826 class DumpLatestFilter extends DumpFilter {
 827         var $page, $pageString, $rev, $revString;
 828
 829         function writeOpenPage( $page, $string ) {
 830                 $this->page = $page;
 831                 $this->pageString = $string;
 832         }
 833
 834         function writeClosePage( $string ) {
 835                 if( $this->rev ) {
 836                         $this->sink->writeOpenPage( $this->page, $this->pageString );
 837                         $this->sink->writeRevision( $this->rev, $this->revString );
 838                         $this->sink->writeClosePage( $string );
 839                 }
 840                 $this->rev = null;
 841                 $this->revString = null;
 842                 $this->page = null;
 843                 $this->pageString = null;
 844         }
 845
 846         function writeRevision( $rev, $string ) {
 847                 if( $rev->rev_id == $this->page->page_latest ) {
 848                         $this->rev = $rev;
 849                         $this->revString = $string;
 850                 }
 851         }
 852 }
 853
 854 /**
 855  * Base class for output stream; prints to stdout or buffer or whereever.
 856  * @ingroup Dump
 857  */
 858 class DumpMultiWriter {
 859         function DumpMultiWriter( $sinks ) {
 860                 $this->sinks = $sinks;
 861                 $this->count = count( $sinks );
 862         }
 863
 864         function writeOpenStream( $string ) {
 865                 for( $i = 0; $i < $this->count; $i++ ) {
 866                         $this->sinks[$i]->writeOpenStream( $string );
 867                 }
 868         }
 869
 870         function writeCloseStream( $string ) {
 871                 for( $i = 0; $i < $this->count; $i++ ) {
 872                         $this->sinks[$i]->writeCloseStream( $string );
 873                 }
 874         }
 875
 876         function writeOpenPage( $page, $string ) {
 877                 for( $i = 0; $i < $this->count; $i++ ) {
 878                         $this->sinks[$i]->writeOpenPage( $page, $string );
 879                 }
 880         }
 881
 882         function writeClosePage( $string ) {
 883                 for( $i = 0; $i < $this->count; $i++ ) {
 884                         $this->sinks[$i]->writeClosePage( $string );
 885                 }
 886         }
 887
 888         function writeRevision( $rev, $string ) {
 889                 for( $i = 0; $i < $this->count; $i++ ) {
 890                         $this->sinks[$i]->writeRevision( $rev, $string );
 891                 }
 892         }
 893 }
 894
 895 function xmlsafe( $string ) {
 896         $fname = 'xmlsafe';
 897         wfProfileIn( $fname );
 898
 899         /**
 900          * The page may contain old data which has not been properly normalized.
 901          * Invalid UTF-8 sequences or forbidden control characters will make our
 902          * XML output invalid, so be sure to strip them out.
 903          */
 904         $string = UtfNormal::cleanUp( $string );
 905
 906         $string = htmlspecialchars( $string );
 907         wfProfileOut( $fname );
 908         return $string;
 909 }