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 = 1;
  34         const CURRENT = 2;
  35         const STABLE = 4; // extension defined
  36         const LOGS = 8;
  37
  38         const BUFFER = 0;
  39         const STREAM = 1;
  40
  41         const TEXT = 0;
  42         const STUB = 1;
  43
  44         /**
  45          * If using WikiExporter::STREAM to stream a large amount of data,
  46          * provide a database connection which is not managed by
  47          * LoadBalancer to read from: some history blob types will
  48          * make additional queries to pull source data while the
  49          * main query is still running.
  50          *
  51          * @param $db Database
  52          * @param $history Mixed: one of WikiExporter::FULL or WikiExporter::CURRENT,
  53          *                 or an associative array:
  54          *                   offset: non-inclusive offset at which to start the query
  55          *                   limit: maximum number of rows to return
  56          *                   dir: "asc" or "desc" timestamp order
  57          * @param $buffer Int: one of WikiExporter::BUFFER or WikiExporter::STREAM
  58          */
  59         function __construct( &$db, $history = WikiExporter::CURRENT,
  60                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  61                 $this->db =& $db;
  62                 $this->history = $history;
  63                 $this->buffer  = $buffer;
  64                 $this->writer  = new XmlDumpWriter();
  65                 $this->sink    = new DumpOutput();
  66                 $this->text    = $text;
  67         }
  68
  69         /**
  70          * Set the DumpOutput or DumpFilter object which will receive
  71          * various row objects and XML output for filtering. Filters
  72          * can be chained or used as callbacks.
  73          *
  74          * @param $sink mixed
  75          */
  76         public function setOutputSink( &$sink ) {
  77                 $this->sink =& $sink;
  78         }
  79
  80         public function openStream() {
  81                 $output = $this->writer->openStream();
  82                 $this->sink->writeOpenStream( $output );
  83         }
  84
  85         public function closeStream() {
  86                 $output = $this->writer->closeStream();
  87                 $this->sink->writeCloseStream( $output );
  88         }
  89
  90         /**
  91          * Dumps a series of page and revision records for all pages
  92          * in the database, either including complete history or only
  93          * the most recent version.
  94          */
  95         public function allPages() {
  96                 return $this->dumpFrom( '' );
  97         }
  98
  99         /**
 100          * Dumps a series of page and revision records for those pages
 101          * in the database falling within the page_id range given.
 102          * @param $start Int: inclusive lower limit (this id is included)
 103          * @param $end   Int: Exclusive upper limit (this id is not included)
 104          *                   If 0, no upper limit.
 105          */
 106         public function pagesByRange( $start, $end ) {
 107                 $condition = 'page_id >= ' . intval( $start );
 108                 if( $end ) {
 109                         $condition .= ' AND page_id < ' . intval( $end );
 110                 }
 111                 return $this->dumpFrom( $condition );
 112         }
 113
 114         /**
 115          * @param $title Title
 116          */
 117         public function pageByTitle( $title ) {
 118                 return $this->dumpFrom(
 119                         'page_namespace=' . $title->getNamespace() .
 120                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 121         }
 122
 123         public function pageByName( $name ) {
 124                 $title = Title::newFromText( $name );
 125                 if( is_null( $title ) ) {
 126                         return new WikiError( "Can't export invalid title" );
 127                 } else {
 128                         return $this->pageByTitle( $title );
 129                 }
 130         }
 131
 132         public function pagesByName( $names ) {
 133                 foreach( $names as $name ) {
 134                         $this->pageByName( $name );
 135                 }
 136         }
 137
 138         public function allLogs() {
 139                 return $this->dumpFrom( '' );
 140         }
 141
 142         public function logsByRange( $start, $end ) {
 143                 $condition = 'log_id >= ' . intval( $start );
 144                 if( $end ) {
 145                         $condition .= ' AND log_id < ' . intval( $end );
 146                 }
 147                 return $this->dumpFrom( $condition );
 148         }
 149
 150         # Generates the distinct list of authors of an article
 151         # Not called by default (depends on $this->list_authors)
 152         # Can be set by Special:Export when not exporting whole history
 153         protected function do_list_authors( $page , $revision , $cond ) {
 154                 $fname = "do_list_authors" ;
 155                 wfProfileIn( $fname );
 156                 $this->author_list = "<contributors>";
 157                 //rev_deleted
 158                 $nothidden = '(rev_deleted & '.Revision::DELETED_USER.') = 0';
 159
 160                 $sql = "SELECT DISTINCT rev_user_text,rev_user FROM {$page},{$revision}
 161                 WHERE page_id=rev_page AND $nothidden AND " . $cond ;
 162                 $result = $this->db->query( $sql, $fname );
 163                 $resultset = $this->db->resultObject( $result );
 164                 while( $row = $resultset->fetchObject() ) {
 165                         $this->author_list .= "<contributor>" .
 166                                 "<username>" .
 167                                 htmlentities( $row->rev_user_text )  .
 168                                 "</username>" .
 169                                 "<id>" .
 170                                 $row->rev_user .
 171                                 "</id>" .
 172                                 "</contributor>";
 173                 }
 174                 wfProfileOut( $fname );
 175                 $this->author_list .= "</contributors>";
 176         }
 177
 178         protected function dumpFrom( $cond = '' ) {
 179                 wfProfileIn( __METHOD__ );
 180                 # For logging dumps...
 181                 if( $this->history & self::LOGS ) {
 182                         if( $this->buffer == WikiExporter::STREAM ) {
 183                                 $prev = $this->db->bufferResults( false );
 184                         }
 185                         $where = array( 'user_id = log_user' );
 186                         # Hide private logs
 187                         $hideLogs = LogEventsList::getExcludeClause( $this->db );
 188                         if( $hideLogs ) $where[] = $hideLogs;
 189                         # Add on any caller specified conditions
 190                         if( $cond ) $where[] = $cond;
 191                         # Get logging table name for logging.* clause
 192                         $logging = $this->db->tableName('logging');
 193                         $result = $this->db->select( array('logging','user'),
 194                                 array( "{$logging}.*", 'user_name' ), // grab the user name
 195                                 $where,
 196                                 __METHOD__,
 197                                 array( 'ORDER BY' => 'log_id', 'USE INDEX' => array('logging' => 'PRIMARY') )
 198                         );
 199                         $wrapper = $this->db->resultObject( $result );
 200                         if( $this->buffer == WikiExporter::STREAM ) {
 201                                 $this->db->bufferResults( $prev );
 202                         }
 203                         $this->outputLogStream( $wrapper );
 204                 # For page dumps...
 205                 } else {
 206                         $tables = array( 'page', 'revision' );
 207                         $opts = array( 'ORDER BY' => 'page_id ASC' );
 208                         $opts['USE INDEX'] = array();
 209                         $join = array();
 210                         # Full history dumps...
 211                         if( $this->history & WikiExporter::FULL ) {
 212                                 $join['revision'] = array('INNER JOIN','page_id=rev_page');
 213                         # Latest revision dumps...
 214                         } elseif( $this->history & WikiExporter::CURRENT ) {
 215                                 if( $this->list_authors && $cond != '' )  { // List authors, if so desired
 216                                         list($page,$revision) = $this->db->tableNamesN('page','revision');
 217                                         $this->do_list_authors( $page, $revision, $cond );
 218                                 }
 219                                 $join['revision'] = array('INNER JOIN','page_id=rev_page AND page_latest=rev_id');
 220                         # "Stable" revision dumps...
 221                         } elseif( $this->history & WikiExporter::STABLE ) {
 222                                 # Default JOIN, to be overridden...
 223                                 $join['revision'] = array('INNER JOIN','page_id=rev_page AND page_latest=rev_id');
 224                                 # One, and only one hook should set this, and return false
 225                                 if( wfRunHooks( 'WikiExporter::dumpStableQuery', array(&$tables,&$opts,&$join) ) ) {
 226                                         wfProfileOut( __METHOD__ );
 227                                         return new WikiError( __METHOD__." given invalid history dump type." );
 228                                 }
 229                         # Time offset/limit for all pages/history...
 230                         } elseif( is_array( $this->history ) ) {
 231                                 $revJoin = 'page_id=rev_page';
 232                                 # Set time order
 233                                 if( $this->history['dir'] == 'asc' ) {
 234                                         $op = '>';
 235                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 236                                 } else {
 237                                         $op = '<';
 238                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 239                                 }
 240                                 # Set offset
 241                                 if( !empty( $this->history['offset'] ) ) {
 242                                         $revJoin .= " AND rev_timestamp $op " .
 243                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 244                                 }
 245                                 $join['revision'] = array('INNER JOIN',$revJoin);
 246                                 # Set query limit
 247                                 if( !empty( $this->history['limit'] ) ) {
 248                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 249                                 }
 250                         # Uknown history specification parameter?
 251                         } else {
 252                                 wfProfileOut( __METHOD__ );
 253                                 return new WikiError( __METHOD__." given invalid history dump type." );
 254                         }
 255                         # Query optimization hacks
 256                         if( $cond == '' ) {
 257                                 $opts[] = 'STRAIGHT_JOIN';
 258                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 259                         }
 260                         # Build text join options
 261                         if( $this->text != WikiExporter::STUB ) { // 1-pass
 262                                 $tables[] = 'text';
 263                                 $join['text'] = array('INNER JOIN','rev_text_id=old_id');
 264                         }
 265
 266                         if( $this->buffer == WikiExporter::STREAM ) {
 267                                 $prev = $this->db->bufferResults( false );
 268                         }
 269
 270                         # Do the query!
 271                         $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 272                         $wrapper = $this->db->resultObject( $result );
 273                         # Output dump results
 274                         $this->outputPageStream( $wrapper );
 275                         if( $this->list_authors ) {
 276                                 $this->outputPageStream( $wrapper );
 277                         }
 278
 279                         if( $this->buffer == WikiExporter::STREAM ) {
 280                                 $this->db->bufferResults( $prev );
 281                         }
 282                 }
 283                 wfProfileOut( __METHOD__ );
 284         }
 285
 286         /**
 287          * Runs through a query result set dumping page and revision records.
 288          * The result set should be sorted/grouped by page to avoid duplicate
 289          * page records in the output.
 290          *
 291          * The result set will be freed once complete. Should be safe for
 292          * streaming (non-buffered) queries, as long as it was made on a
 293          * separate database connection not managed by LoadBalancer; some
 294          * blob storage types will make queries to pull source data.
 295          *
 296          * @param $resultset ResultWrapper
 297          */
 298         protected function outputPageStream( $resultset ) {
 299                 $last = null;
 300                 while( $row = $resultset->fetchObject() ) {
 301                         if( is_null( $last ) ||
 302                                 $last->page_namespace != $row->page_namespace ||
 303                                 $last->page_title     != $row->page_title ) {
 304                                 if( isset( $last ) ) {
 305                                         $output = '';
 306                                         if( $this->dumpUploads ) {
 307                                                 $output .= $this->writer->writeUploads( $last );
 308                                         }
 309                                         $output .= $this->writer->closePage();
 310                                         $this->sink->writeClosePage( $output );
 311                                 }
 312                                 $output = $this->writer->openPage( $row );
 313                                 $this->sink->writeOpenPage( $row, $output );
 314                                 $last = $row;
 315                         }
 316                         $output = $this->writer->writeRevision( $row );
 317                         $this->sink->writeRevision( $row, $output );
 318                 }
 319                 if( isset( $last ) ) {
 320                         $output = '';
 321                         if( $this->dumpUploads ) {
 322                                 $output .= $this->writer->writeUploads( $last );
 323                         }
 324                         $output .= $this->author_list;
 325                         $output .= $this->writer->closePage();
 326                         $this->sink->writeClosePage( $output );
 327                 }
 328                 $resultset->free();
 329         }
 330
 331         protected function outputLogStream( $resultset ) {
 332                 while( $row = $resultset->fetchObject() ) {
 333                         $output = $this->writer->writeLogItem( $row );
 334                         $this->sink->writeLogItem( $row, $output );
 335                 }
 336                 $resultset->free();
 337         }
 338 }
 339
 340 /**
 341  * @ingroup Dump
 342  */
 343 class XmlDumpWriter {
 344
 345         /**
 346          * Returns the export schema version.
 347          * @return string
 348          */
 349         function schemaVersion() {
 350                 return "0.3"; // FIXME: upgrade to 0.4 when updated XSD is ready, for the revision deletion bits
 351         }
 352
 353         /**
 354          * Opens the XML output stream's root <mediawiki> element.
 355          * This does not include an xml directive, so is safe to include
 356          * as a subelement in a larger XML stream. Namespace and XML Schema
 357          * references are included.
 358          *
 359          * Output will be encoded in UTF-8.
 360          *
 361          * @return string
 362          */
 363         function openStream() {
 364                 global $wgContLanguageCode;
 365                 $ver = $this->schemaVersion();
 366                 return Xml::element( 'mediawiki', array(
 367                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 368                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 369                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 370                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 371                         'version'            => $ver,
 372                         'xml:lang'           => $wgContLanguageCode ),
 373                         null ) .
 374                         "\n" .
 375                         $this->siteInfo();
 376         }
 377
 378         function siteInfo() {
 379                 $info = array(
 380                         $this->sitename(),
 381                         $this->homelink(),
 382                         $this->generator(),
 383                         $this->caseSetting(),
 384                         $this->namespaces() );
 385                 return "  <siteinfo>\n    " .
 386                         implode( "\n    ", $info ) .
 387                         "\n  </siteinfo>\n";
 388         }
 389
 390         function sitename() {
 391                 global $wgSitename;
 392                 return Xml::element( 'sitename', array(), $wgSitename );
 393         }
 394
 395         function generator() {
 396                 global $wgVersion;
 397                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 398         }
 399
 400         function homelink() {
 401                 return Xml::element( 'base', array(), Title::newMainPage()->getFullUrl() );
 402         }
 403
 404         function caseSetting() {
 405                 global $wgCapitalLinks;
 406                 // "case-insensitive" option is reserved for future
 407                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 408                 return Xml::element( 'case', array(), $sensitivity );
 409         }
 410
 411         function namespaces() {
 412                 global $wgContLang;
 413                 $spaces = "<namespaces>\n";
 414                 foreach( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 415                         $spaces .= '      ' . Xml::element( 'namespace', array( 'key' => $ns ), $title ) . "\n";
 416                 }
 417                 $spaces .= "    </namespaces>";
 418                 return $spaces;
 419         }
 420
 421         /**
 422          * Closes the output stream with the closing root element.
 423          * Call when finished dumping things.
 424          */
 425         function closeStream() {
 426                 return "</mediawiki>\n";
 427         }
 428
 429
 430         /**
 431          * Opens a <page> section on the output stream, with data
 432          * from the given database row.
 433          *
 434          * @param $row object
 435          * @return string
 436          * @access private
 437          */
 438         function openPage( $row ) {
 439                 $out = "  <page>\n";
 440                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 441                 $out .= '    ' . Xml::elementClean( 'title', array(), $title->getPrefixedText() ) . "\n";
 442                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 443                 if( '' != $row->page_restrictions ) {
 444                         $out .= '    ' . Xml::element( 'restrictions', array(),
 445                                 strval( $row->page_restrictions ) ) . "\n";
 446                 }
 447                 return $out;
 448         }
 449
 450         /**
 451          * Closes a <page> section on the output stream.
 452          *
 453          * @access private
 454          */
 455         function closePage() {
 456                 return "  </page>\n";
 457         }
 458
 459         /**
 460          * Dumps a <revision> section on the output stream, with
 461          * data filled in from the given database row.
 462          *
 463          * @param $row object
 464          * @return string
 465          * @access private
 466          */
 467         function writeRevision( $row ) {
 468                 $fname = 'WikiExporter::dumpRev';
 469                 wfProfileIn( $fname );
 470
 471                 $out  = "    <revision>\n";
 472                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 473
 474                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 475
 476                 if( $row->rev_deleted & Revision::DELETED_USER ) {
 477                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 478                 } else {
 479                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 480                 }
 481
 482                 if( $row->rev_minor_edit ) {
 483                         $out .=  "      <minor/>\n";
 484                 }
 485                 if( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 486                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 487                 } elseif( $row->rev_comment != '' ) {
 488                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 489                 }
 490
 491                 if( $row->rev_deleted & Revision::DELETED_TEXT ) {
 492                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 493                 } elseif( isset( $row->old_text ) ) {
 494                         // Raw text from the database may have invalid chars
 495                         $text = strval( Revision::getRevisionText( $row ) );
 496                         $out .= "      " . Xml::elementClean( 'text',
 497                                 array( 'xml:space' => 'preserve' ),
 498                                 strval( $text ) ) . "\n";
 499                 } else {
 500                         // Stub output
 501                         $out .= "      " . Xml::element( 'text',
 502                                 array( 'id' => $row->rev_text_id ),
 503                                 "" ) . "\n";
 504                 }
 505
 506                 $out .= "    </revision>\n";
 507
 508                 wfProfileOut( $fname );
 509                 return $out;
 510         }
 511
 512         /**
 513          * Dumps a <logitem> section on the output stream, with
 514          * data filled in from the given database row.
 515          *
 516          * @param $row object
 517          * @return string
 518          * @access private
 519          */
 520         function writeLogItem( $row ) {
 521                 $fname = 'WikiExporter::writeLogItem';
 522                 wfProfileIn( $fname );
 523
 524                 $out  = "    <logitem>\n";
 525                 $out .= "      " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 526
 527                 $out .= $this->writeTimestamp( $row->log_timestamp );
 528
 529                 if( $row->log_deleted & LogPage::DELETED_USER ) {
 530                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 531                 } else {
 532                         $out .= $this->writeContributor( $row->log_user, $row->user_name );
 533                 }
 534
 535                 if( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 536                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 537                 } elseif( $row->log_comment != '' ) {
 538                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 539                 }
 540
 541                 $out .= "      " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 542                 $out .= "      " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 543
 544                 if( $row->log_deleted & LogPage::DELETED_ACTION ) {
 545                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 546                 } else {
 547                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 548                         $out .= "      " . Xml::elementClean( 'logtitle', null, $title->getPrefixedText() ) . "\n";
 549                         $out .= "      " . Xml::elementClean( 'params',
 550                                 array( 'xml:space' => 'preserve' ),
 551                                 strval( $row->log_params ) ) . "\n";
 552                 }
 553
 554                 $out .= "    </logitem>\n";
 555
 556                 wfProfileOut( $fname );
 557                 return $out;
 558         }
 559
 560         function writeTimestamp( $timestamp ) {
 561                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 562                 return "      " . Xml::element( 'timestamp', null, $ts ) . "\n";
 563         }
 564
 565         function writeContributor( $id, $text ) {
 566                 $out = "      <contributor>\n";
 567                 if( $id ) {
 568                         $out .= "        " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 569                         $out .= "        " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 570                 } else {
 571                         $out .= "        " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 572                 }
 573                 $out .= "      </contributor>\n";
 574                 return $out;
 575         }
 576
 577         /**
 578          * Warning! This data is potentially inconsistent. :(
 579          */
 580         function writeUploads( $row ) {
 581                 if( $row->page_namespace == NS_IMAGE ) {
 582                         $img = wfFindFile( $row->page_title );
 583                         if( $img ) {
 584                                 $out = '';
 585                                 foreach( array_reverse( $img->getHistory() ) as $ver ) {
 586                                         $out .= $this->writeUpload( $ver );
 587                                 }
 588                                 $out .= $this->writeUpload( $img );
 589                                 return $out;
 590                         }
 591                 }
 592                 return '';
 593         }
 594
 595         function writeUpload( $file ) {
 596                 return "    <upload>\n" .
 597                         $this->writeTimestamp( $file->getTimestamp() ) .
 598                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 599                         "      " . Xml::elementClean( 'comment', null, $file->getDescription() ) . "\n" .
 600                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 601                         "      " . Xml::element( 'src', null, $file->getFullUrl() ) . "\n" .
 602                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 603                         "    </upload>\n";
 604         }
 605
 606 }
 607
 608
 609 /**
 610  * Base class for output stream; prints to stdout or buffer or whereever.
 611  * @ingroup Dump
 612  */
 613 class DumpOutput {
 614         function writeOpenStream( $string ) {
 615                 $this->write( $string );
 616         }
 617
 618         function writeCloseStream( $string ) {
 619                 $this->write( $string );
 620         }
 621
 622         function writeOpenPage( $page, $string ) {
 623                 $this->write( $string );
 624         }
 625
 626         function writeClosePage( $string ) {
 627                 $this->write( $string );
 628         }
 629
 630         function writeRevision( $rev, $string ) {
 631                 $this->write( $string );
 632         }
 633
 634         function writeLogItem( $rev, $string ) {
 635                 $this->write( $string );
 636         }
 637
 638         /**
 639          * Override to write to a different stream type.
 640          * @return bool
 641          */
 642         function write( $string ) {
 643                 print $string;
 644         }
 645 }
 646
 647 /**
 648  * Stream outputter to send data to a file.
 649  * @ingroup Dump
 650  */
 651 class DumpFileOutput extends DumpOutput {
 652         var $handle;
 653
 654         function DumpFileOutput( $file ) {
 655                 $this->handle = fopen( $file, "wt" );
 656         }
 657
 658         function write( $string ) {
 659                 fputs( $this->handle, $string );
 660         }
 661 }
 662
 663 /**
 664  * Stream outputter to send data to a file via some filter program.
 665  * Even if compression is available in a library, using a separate
 666  * program can allow us to make use of a multi-processor system.
 667  * @ingroup Dump
 668  */
 669 class DumpPipeOutput extends DumpFileOutput {
 670         function DumpPipeOutput( $command, $file = null ) {
 671                 if( !is_null( $file ) ) {
 672                         $command .=  " > " . wfEscapeShellArg( $file );
 673                 }
 674                 $this->handle = popen( $command, "w" );
 675         }
 676 }
 677
 678 /**
 679  * Sends dump output via the gzip compressor.
 680  * @ingroup Dump
 681  */
 682 class DumpGZipOutput extends DumpPipeOutput {
 683         function DumpGZipOutput( $file ) {
 684                 parent::DumpPipeOutput( "gzip", $file );
 685         }
 686 }
 687
 688 /**
 689  * Sends dump output via the bgzip2 compressor.
 690  * @ingroup Dump
 691  */
 692 class DumpBZip2Output extends DumpPipeOutput {
 693         function DumpBZip2Output( $file ) {
 694                 parent::DumpPipeOutput( "bzip2", $file );
 695         }
 696 }
 697
 698 /**
 699  * Sends dump output via the p7zip compressor.
 700  * @ingroup Dump
 701  */
 702 class Dump7ZipOutput extends DumpPipeOutput {
 703         function Dump7ZipOutput( $file ) {
 704                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 705                 // Suppress annoying useless crap from p7zip
 706                 // Unfortunately this could suppress real error messages too
 707                 $command .= ' >' . wfGetNull() . ' 2>&1';
 708                 parent::DumpPipeOutput( $command );
 709         }
 710 }
 711
 712
 713
 714 /**
 715  * Dump output filter class.
 716  * This just does output filtering and streaming; XML formatting is done
 717  * higher up, so be careful in what you do.
 718  * @ingroup Dump
 719  */
 720 class DumpFilter {
 721         function DumpFilter( &$sink ) {
 722                 $this->sink =& $sink;
 723         }
 724
 725         function writeOpenStream( $string ) {
 726                 $this->sink->writeOpenStream( $string );
 727         }
 728
 729         function writeCloseStream( $string ) {
 730                 $this->sink->writeCloseStream( $string );
 731         }
 732
 733         function writeOpenPage( $page, $string ) {
 734                 $this->sendingThisPage = $this->pass( $page, $string );
 735                 if( $this->sendingThisPage ) {
 736                         $this->sink->writeOpenPage( $page, $string );
 737                 }
 738         }
 739
 740         function writeClosePage( $string ) {
 741                 if( $this->sendingThisPage ) {
 742                         $this->sink->writeClosePage( $string );
 743                         $this->sendingThisPage = false;
 744                 }
 745         }
 746
 747         function writeRevision( $rev, $string ) {
 748                 if( $this->sendingThisPage ) {
 749                         $this->sink->writeRevision( $rev, $string );
 750                 }
 751         }
 752
 753         function writeLogItem( $rev, $string ) {
 754                 $this->sink->writeRevision( $rev, $string );
 755         }
 756
 757         /**
 758          * Override for page-based filter types.
 759          * @return bool
 760          */
 761         function pass( $page ) {
 762                 return true;
 763         }
 764 }
 765
 766 /**
 767  * Simple dump output filter to exclude all talk pages.
 768  * @ingroup Dump
 769  */
 770 class DumpNotalkFilter extends DumpFilter {
 771         function pass( $page ) {
 772                 return !MWNamespace::isTalk( $page->page_namespace );
 773         }
 774 }
 775
 776 /**
 777  * Dump output filter to include or exclude pages in a given set of namespaces.
 778  * @ingroup Dump
 779  */
 780 class DumpNamespaceFilter extends DumpFilter {
 781         var $invert = false;
 782         var $namespaces = array();
 783
 784         function DumpNamespaceFilter( &$sink, $param ) {
 785                 parent::DumpFilter( $sink );
 786
 787                 $constants = array(
 788                         "NS_MAIN"           => NS_MAIN,
 789                         "NS_TALK"           => NS_TALK,
 790                         "NS_USER"           => NS_USER,
 791                         "NS_USER_TALK"      => NS_USER_TALK,
 792                         "NS_PROJECT"        => NS_PROJECT,
 793                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
 794                         "NS_FILE"           => NS_FILE,
 795                         "NS_FILE_TALK"      => NS_FILE_TALK,
 796                         "NS_IMAGE"          => NS_IMAGE,  // NS_IMAGE is an alias for NS_FILE
 797                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
 798                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
 799                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
 800                         "NS_TEMPLATE"       => NS_TEMPLATE,
 801                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
 802                         "NS_HELP"           => NS_HELP,
 803                         "NS_HELP_TALK"      => NS_HELP_TALK,
 804                         "NS_CATEGORY"       => NS_CATEGORY,
 805                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
 806
 807                 if( $param{0} == '!' ) {
 808                         $this->invert = true;
 809                         $param = substr( $param, 1 );
 810                 }
 811
 812                 foreach( explode( ',', $param ) as $key ) {
 813                         $key = trim( $key );
 814                         if( isset( $constants[$key] ) ) {
 815                                 $ns = $constants[$key];
 816                                 $this->namespaces[$ns] = true;
 817                         } elseif( is_numeric( $key ) ) {
 818                                 $ns = intval( $key );
 819                                 $this->namespaces[$ns] = true;
 820                         } else {
 821                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
 822                         }
 823                 }
 824         }
 825
 826         function pass( $page ) {
 827                 $match = isset( $this->namespaces[$page->page_namespace] );
 828                 return $this->invert xor $match;
 829         }
 830 }
 831
 832
 833 /**
 834  * Dump output filter to include only the last revision in each page sequence.
 835  * @ingroup Dump
 836  */
 837 class DumpLatestFilter extends DumpFilter {
 838         var $page, $pageString, $rev, $revString;
 839
 840         function writeOpenPage( $page, $string ) {
 841                 $this->page = $page;
 842                 $this->pageString = $string;
 843         }
 844
 845         function writeClosePage( $string ) {
 846                 if( $this->rev ) {
 847                         $this->sink->writeOpenPage( $this->page, $this->pageString );
 848                         $this->sink->writeRevision( $this->rev, $this->revString );
 849                         $this->sink->writeClosePage( $string );
 850                 }
 851                 $this->rev = null;
 852                 $this->revString = null;
 853                 $this->page = null;
 854                 $this->pageString = null;
 855         }
 856
 857         function writeRevision( $rev, $string ) {
 858                 if( $rev->rev_id == $this->page->page_latest ) {
 859                         $this->rev = $rev;
 860                         $this->revString = $string;
 861                 }
 862         }
 863 }
 864
 865 /**
 866  * Base class for output stream; prints to stdout or buffer or whereever.
 867  * @ingroup Dump
 868  */
 869 class DumpMultiWriter {
 870         function DumpMultiWriter( $sinks ) {
 871                 $this->sinks = $sinks;
 872                 $this->count = count( $sinks );
 873         }
 874
 875         function writeOpenStream( $string ) {
 876                 for( $i = 0; $i < $this->count; $i++ ) {
 877                         $this->sinks[$i]->writeOpenStream( $string );
 878                 }
 879         }
 880
 881         function writeCloseStream( $string ) {
 882                 for( $i = 0; $i < $this->count; $i++ ) {
 883                         $this->sinks[$i]->writeCloseStream( $string );
 884                 }
 885         }
 886
 887         function writeOpenPage( $page, $string ) {
 888                 for( $i = 0; $i < $this->count; $i++ ) {
 889                         $this->sinks[$i]->writeOpenPage( $page, $string );
 890                 }
 891         }
 892
 893         function writeClosePage( $string ) {
 894                 for( $i = 0; $i < $this->count; $i++ ) {
 895                         $this->sinks[$i]->writeClosePage( $string );
 896                 }
 897         }
 898
 899         function writeRevision( $rev, $string ) {
 900                 for( $i = 0; $i < $this->count; $i++ ) {
 901                         $this->sinks[$i]->writeRevision( $rev, $string );
 902                 }
 903         }
 904 }
 905
 906 function xmlsafe( $string ) {
 907         $fname = 'xmlsafe';
 908         wfProfileIn( $fname );
 909
 910         /**
 911          * The page may contain old data which has not been properly normalized.
 912          * Invalid UTF-8 sequences or forbidden control characters will make our
 913          * XML output invalid, so be sure to strip them out.
 914          */
 915         $string = UtfNormal::cleanUp( $string );
 916
 917         $string = htmlspecialchars( $string );
 918         wfProfileOut( $fname );
 919         return $string;
 920 }