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          * @param $text Int: one of WikiExporter::TEXT or WikiExporter::STUB
  59          */
  60         function __construct( &$db, $history = WikiExporter::CURRENT,
  61                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  62                 $this->db =& $db;
  63                 $this->history = $history;
  64                 $this->buffer  = $buffer;
  65                 $this->writer  = new XmlDumpWriter();
  66                 $this->sink    = new DumpOutput();
  67                 $this->text    = $text;
  68         }
  69
  70         /**
  71          * Set the DumpOutput or DumpFilter object which will receive
  72          * various row objects and XML output for filtering. Filters
  73          * can be chained or used as callbacks.
  74          *
  75          * @param $sink mixed
  76          */
  77         public function setOutputSink( &$sink ) {
  78                 $this->sink =& $sink;
  79         }
  80
  81         public function openStream() {
  82                 $output = $this->writer->openStream();
  83                 $this->sink->writeOpenStream( $output );
  84         }
  85
  86         public function closeStream() {
  87                 $output = $this->writer->closeStream();
  88                 $this->sink->writeCloseStream( $output );
  89         }
  90
  91         /**
  92          * Dumps a series of page and revision records for all pages
  93          * in the database, either including complete history or only
  94          * the most recent version.
  95          */
  96         public function allPages() {
  97                 return $this->dumpFrom( '' );
  98         }
  99
 100         /**
 101          * Dumps a series of page and revision records for those pages
 102          * in the database falling within the page_id range given.
 103          * @param $start Int: inclusive lower limit (this id is included)
 104          * @param $end   Int: Exclusive upper limit (this id is not included)
 105          *                   If 0, no upper limit.
 106          */
 107         public function pagesByRange( $start, $end ) {
 108                 $condition = 'page_id >= ' . intval( $start );
 109                 if( $end ) {
 110                         $condition .= ' AND page_id < ' . intval( $end );
 111                 }
 112                 return $this->dumpFrom( $condition );
 113         }
 114
 115         /**
 116          * @param $title Title
 117          */
 118         public function pageByTitle( $title ) {
 119                 return $this->dumpFrom(
 120                         'page_namespace=' . $title->getNamespace() .
 121                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 122         }
 123
 124         public function pageByName( $name ) {
 125                 $title = Title::newFromText( $name );
 126                 if( is_null( $title ) ) {
 127                         return new WikiError( "Can't export invalid title" );
 128                 } else {
 129                         return $this->pageByTitle( $title );
 130                 }
 131         }
 132
 133         public function pagesByName( $names ) {
 134                 foreach( $names as $name ) {
 135                         $this->pageByName( $name );
 136                 }
 137         }
 138
 139         public function allLogs() {
 140                 return $this->dumpFrom( '' );
 141         }
 142
 143         public function logsByRange( $start, $end ) {
 144                 $condition = 'log_id >= ' . intval( $start );
 145                 if( $end ) {
 146                         $condition .= ' AND log_id < ' . intval( $end );
 147                 }
 148                 return $this->dumpFrom( $condition );
 149         }
 150
 151         # Generates the distinct list of authors of an article
 152         # Not called by default (depends on $this->list_authors)
 153         # Can be set by Special:Export when not exporting whole history
 154         protected function do_list_authors( $page , $revision , $cond ) {
 155                 wfProfileIn( __METHOD__ );
 156                 $this->author_list = "<contributors>";
 157                 //rev_deleted
 158                 $nothidden = '('.$this->db->bitAnd('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, __METHOD__ );
 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( __METHOD__ );
 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                         $this->outputLogStream( $wrapper );
 201                         if( $this->buffer == WikiExporter::STREAM ) {
 202                                 $this->db->bufferResults( $prev );
 203                         }
 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                         if( is_array( $this->history ) ) {
 211                                 # Time offset/limit for all pages/history...
 212                                 $revJoin = 'page_id=rev_page';
 213                                 # Set time order
 214                                 if( $this->history['dir'] == 'asc' ) {
 215                                         $op = '>';
 216                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 217                                 } else {
 218                                         $op = '<';
 219                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 220                                 }
 221                                 # Set offset
 222                                 if( !empty( $this->history['offset'] ) ) {
 223                                         $revJoin .= " AND rev_timestamp $op " .
 224                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 225                                 }
 226                                 $join['revision'] = array('INNER JOIN',$revJoin);
 227                                 # Set query limit
 228                                 if( !empty( $this->history['limit'] ) ) {
 229                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 230                                 }
 231                         } elseif( $this->history & WikiExporter::FULL ) {
 232                                 # Full history dumps...
 233                                 $join['revision'] = array('INNER JOIN','page_id=rev_page');
 234                         } elseif( $this->history & WikiExporter::CURRENT ) {
 235                                 # Latest revision dumps...
 236                                 if( $this->list_authors && $cond != '' )  { // List authors, if so desired
 237                                         list($page,$revision) = $this->db->tableNamesN('page','revision');
 238                                         $this->do_list_authors( $page, $revision, $cond );
 239                                 }
 240                                 $join['revision'] = array('INNER JOIN','page_id=rev_page AND page_latest=rev_id');
 241                         } elseif( $this->history & WikiExporter::STABLE ) {
 242                                 # "Stable" revision dumps...
 243                                 # Default JOIN, to be overridden...
 244                                 $join['revision'] = array('INNER JOIN','page_id=rev_page AND page_latest=rev_id');
 245                                 # One, and only one hook should set this, and return false
 246                                 if( wfRunHooks( 'WikiExporter::dumpStableQuery', array(&$tables,&$opts,&$join) ) ) {
 247                                         wfProfileOut( __METHOD__ );
 248                                         return new WikiError( __METHOD__." given invalid history dump type." );
 249                                 }
 250                         } else {
 251                                 # Uknown history specification parameter?
 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                         wfRunHooks( 'ModifyExportQuery',
 271                                                 array( $this->db, &$tables, &$cond, &$opts, &$join ) );
 272
 273                         # Do the query!
 274                         $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 275                         $wrapper = $this->db->resultObject( $result );
 276                         # Output dump results
 277                         $this->outputPageStream( $wrapper );
 278                         if( $this->list_authors ) {
 279                                 $this->outputPageStream( $wrapper );
 280                         }
 281
 282                         if( $this->buffer == WikiExporter::STREAM ) {
 283                                 $this->db->bufferResults( $prev );
 284                         }
 285                 }
 286                 wfProfileOut( __METHOD__ );
 287         }
 288
 289         /**
 290          * Runs through a query result set dumping page and revision records.
 291          * The result set should be sorted/grouped by page to avoid duplicate
 292          * page records in the output.
 293          *
 294          * The result set will be freed once complete. Should be safe for
 295          * streaming (non-buffered) queries, as long as it was made on a
 296          * separate database connection not managed by LoadBalancer; some
 297          * blob storage types will make queries to pull source data.
 298          *
 299          * @param $resultset ResultWrapper
 300          */
 301         protected function outputPageStream( $resultset ) {
 302                 $last = null;
 303                 while( $row = $resultset->fetchObject() ) {
 304                         if( is_null( $last ) ||
 305                                 $last->page_namespace != $row->page_namespace ||
 306                                 $last->page_title     != $row->page_title ) {
 307                                 if( isset( $last ) ) {
 308                                         $output = '';
 309                                         if( $this->dumpUploads ) {
 310                                                 $output .= $this->writer->writeUploads( $last );
 311                                         }
 312                                         $output .= $this->writer->closePage();
 313                                         $this->sink->writeClosePage( $output );
 314                                 }
 315                                 $output = $this->writer->openPage( $row );
 316                                 $this->sink->writeOpenPage( $row, $output );
 317                                 $last = $row;
 318                         }
 319                         $output = $this->writer->writeRevision( $row );
 320                         $this->sink->writeRevision( $row, $output );
 321                 }
 322                 if( isset( $last ) ) {
 323                         $output = '';
 324                         if( $this->dumpUploads ) {
 325                                 $output .= $this->writer->writeUploads( $last );
 326                         }
 327                         $output .= $this->author_list;
 328                         $output .= $this->writer->closePage();
 329                         $this->sink->writeClosePage( $output );
 330                 }
 331         }
 332
 333         protected function outputLogStream( $resultset ) {
 334                 while( $row = $resultset->fetchObject() ) {
 335                         $output = $this->writer->writeLogItem( $row );
 336                         $this->sink->writeLogItem( $row, $output );
 337                 }
 338         }
 339 }
 340
 341 /**
 342  * @ingroup Dump
 343  */
 344 class XmlDumpWriter {
 345
 346         /**
 347          * Returns the export schema version.
 348          * @return string
 349          */
 350         function schemaVersion() {
 351                 return "0.4";
 352         }
 353
 354         /**
 355          * Opens the XML output stream's root <mediawiki> element.
 356          * This does not include an xml directive, so is safe to include
 357          * as a subelement in a larger XML stream. Namespace and XML Schema
 358          * references are included.
 359          *
 360          * Output will be encoded in UTF-8.
 361          *
 362          * @return string
 363          */
 364         function openStream() {
 365                 global $wgContLanguageCode;
 366                 $ver = $this->schemaVersion();
 367                 return Xml::element( 'mediawiki', array(
 368                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 369                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 370                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 371                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 372                         'version'            => $ver,
 373                         'xml:lang'           => $wgContLanguageCode ),
 374                         null ) .
 375                         "\n" .
 376                         $this->siteInfo();
 377         }
 378
 379         function siteInfo() {
 380                 $info = array(
 381                         $this->sitename(),
 382                         $this->homelink(),
 383                         $this->generator(),
 384                         $this->caseSetting(),
 385                         $this->namespaces() );
 386                 return "  <siteinfo>\n    " .
 387                         implode( "\n    ", $info ) .
 388                         "\n  </siteinfo>\n";
 389         }
 390
 391         function sitename() {
 392                 global $wgSitename;
 393                 return Xml::element( 'sitename', array(), $wgSitename );
 394         }
 395
 396         function generator() {
 397                 global $wgVersion;
 398                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 399         }
 400
 401         function homelink() {
 402                 return Xml::element( 'base', array(), Title::newMainPage()->getFullUrl() );
 403         }
 404
 405         function caseSetting() {
 406                 global $wgCapitalLinks;
 407                 // "case-insensitive" option is reserved for future
 408                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 409                 return Xml::element( 'case', array(), $sensitivity );
 410         }
 411
 412         function namespaces() {
 413                 global $wgContLang;
 414                 $spaces = "<namespaces>\n";
 415                 foreach( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 416                         $spaces .= '      ' .
 417                                 Xml::element( 'namespace',
 418                                         array(  'key' => $ns,
 419                                                         'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 420                                         ), $title ) . "\n";
 421                 }
 422                 $spaces .= "    </namespaces>";
 423                 return $spaces;
 424         }
 425
 426         /**
 427          * Closes the output stream with the closing root element.
 428          * Call when finished dumping things.
 429          */
 430         function closeStream() {
 431                 return "</mediawiki>\n";
 432         }
 433
 434
 435         /**
 436          * Opens a <page> section on the output stream, with data
 437          * from the given database row.
 438          *
 439          * @param $row object
 440          * @return string
 441          * @access private
 442          */
 443         function openPage( $row ) {
 444                 $out = "  <page>\n";
 445                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 446                 $out .= '    ' . Xml::elementClean( 'title', array(), $title->getPrefixedText() ) . "\n";
 447                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 448                 if( $row->page_is_redirect ) {
 449                         $out .= '    ' . Xml::element( 'redirect', array() ) . "\n";
 450                 }
 451                 if( $row->page_restrictions != '' ) {
 452                         $out .= '    ' . Xml::element( 'restrictions', array(),
 453                                 strval( $row->page_restrictions ) ) . "\n";
 454                 }
 455
 456                 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 457
 458                 return $out;
 459         }
 460
 461         /**
 462          * Closes a <page> section on the output stream.
 463          *
 464          * @access private
 465          */
 466         function closePage() {
 467                 return "  </page>\n";
 468         }
 469
 470         /**
 471          * Dumps a <revision> section on the output stream, with
 472          * data filled in from the given database row.
 473          *
 474          * @param $row object
 475          * @return string
 476          * @access private
 477          */
 478         function writeRevision( $row ) {
 479                 wfProfileIn( __METHOD__ );
 480
 481                 $out  = "    <revision>\n";
 482                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 483
 484                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 485
 486                 if( $row->rev_deleted & Revision::DELETED_USER ) {
 487                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 488                 } else {
 489                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 490                 }
 491
 492                 if( $row->rev_minor_edit ) {
 493                         $out .=  "      <minor/>\n";
 494                 }
 495                 if( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 496                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 497                 } elseif( $row->rev_comment != '' ) {
 498                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 499                 }
 500
 501                 $text = '';
 502                 if( $row->rev_deleted & Revision::DELETED_TEXT ) {
 503                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 504                 } elseif( isset( $row->old_text ) ) {
 505                         // Raw text from the database may have invalid chars
 506                         $text = strval( Revision::getRevisionText( $row ) );
 507                         $out .= "      " . Xml::elementClean( 'text',
 508                                 array( 'xml:space' => 'preserve' ),
 509                                 strval( $text ) ) . "\n";
 510                 } else {
 511                         // Stub output
 512                         $out .= "      " . Xml::element( 'text',
 513                                 array( 'id' => $row->rev_text_id ),
 514                                 "" ) . "\n";
 515                 }
 516
 517                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 518
 519                 $out .= "    </revision>\n";
 520
 521                 wfProfileOut( __METHOD__ );
 522                 return $out;
 523         }
 524
 525         /**
 526          * Dumps a <logitem> section on the output stream, with
 527          * data filled in from the given database row.
 528          *
 529          * @param $row object
 530          * @return string
 531          * @access private
 532          */
 533         function writeLogItem( $row ) {
 534                 wfProfileIn( __METHOD__ );
 535
 536                 $out  = "    <logitem>\n";
 537                 $out .= "      " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 538
 539                 $out .= $this->writeTimestamp( $row->log_timestamp );
 540
 541                 if( $row->log_deleted & LogPage::DELETED_USER ) {
 542                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 543                 } else {
 544                         $out .= $this->writeContributor( $row->log_user, $row->user_name );
 545                 }
 546
 547                 if( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 548                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 549                 } elseif( $row->log_comment != '' ) {
 550                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 551                 }
 552
 553                 $out .= "      " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 554                 $out .= "      " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 555
 556                 if( $row->log_deleted & LogPage::DELETED_ACTION ) {
 557                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 558                 } else {
 559                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 560                         $out .= "      " . Xml::elementClean( 'logtitle', null, $title->getPrefixedText() ) . "\n";
 561                         $out .= "      " . Xml::elementClean( 'params',
 562                                 array( 'xml:space' => 'preserve' ),
 563                                 strval( $row->log_params ) ) . "\n";
 564                 }
 565
 566                 $out .= "    </logitem>\n";
 567
 568                 wfProfileOut( __METHOD__ );
 569                 return $out;
 570         }
 571
 572         function writeTimestamp( $timestamp ) {
 573                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 574                 return "      " . Xml::element( 'timestamp', null, $ts ) . "\n";
 575         }
 576
 577         function writeContributor( $id, $text ) {
 578                 $out = "      <contributor>\n";
 579                 if( $id ) {
 580                         $out .= "        " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 581                         $out .= "        " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 582                 } else {
 583                         $out .= "        " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 584                 }
 585                 $out .= "      </contributor>\n";
 586                 return $out;
 587         }
 588
 589         /**
 590          * Warning! This data is potentially inconsistent. :(
 591          */
 592         function writeUploads( $row ) {
 593                 if( $row->page_namespace == NS_IMAGE ) {
 594                         $img = wfFindFile( $row->page_title );
 595                         if( $img ) {
 596                                 $out = '';
 597                                 foreach( array_reverse( $img->getHistory() ) as $ver ) {
 598                                         $out .= $this->writeUpload( $ver );
 599                                 }
 600                                 $out .= $this->writeUpload( $img );
 601                                 return $out;
 602                         }
 603                 }
 604                 return '';
 605         }
 606
 607         function writeUpload( $file ) {
 608                 return "    <upload>\n" .
 609                         $this->writeTimestamp( $file->getTimestamp() ) .
 610                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 611                         "      " . Xml::elementClean( 'comment', null, $file->getDescription() ) . "\n" .
 612                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 613                         "      " . Xml::element( 'src', null, $file->getFullUrl() ) . "\n" .
 614                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 615                         "    </upload>\n";
 616         }
 617
 618 }
 619
 620
 621 /**
 622  * Base class for output stream; prints to stdout or buffer or whereever.
 623  * @ingroup Dump
 624  */
 625 class DumpOutput {
 626         function writeOpenStream( $string ) {
 627                 $this->write( $string );
 628         }
 629
 630         function writeCloseStream( $string ) {
 631                 $this->write( $string );
 632         }
 633
 634         function writeOpenPage( $page, $string ) {
 635                 $this->write( $string );
 636         }
 637
 638         function writeClosePage( $string ) {
 639                 $this->write( $string );
 640         }
 641
 642         function writeRevision( $rev, $string ) {
 643                 $this->write( $string );
 644         }
 645
 646         function writeLogItem( $rev, $string ) {
 647                 $this->write( $string );
 648         }
 649
 650         /**
 651          * Override to write to a different stream type.
 652          * @return bool
 653          */
 654         function write( $string ) {
 655                 print $string;
 656         }
 657 }
 658
 659 /**
 660  * Stream outputter to send data to a file.
 661  * @ingroup Dump
 662  */
 663 class DumpFileOutput extends DumpOutput {
 664         var $handle;
 665
 666         function DumpFileOutput( $file ) {
 667                 $this->handle = fopen( $file, "wt" );
 668         }
 669
 670         function write( $string ) {
 671                 fputs( $this->handle, $string );
 672         }
 673 }
 674
 675 /**
 676  * Stream outputter to send data to a file via some filter program.
 677  * Even if compression is available in a library, using a separate
 678  * program can allow us to make use of a multi-processor system.
 679  * @ingroup Dump
 680  */
 681 class DumpPipeOutput extends DumpFileOutput {
 682         function DumpPipeOutput( $command, $file = null ) {
 683                 if( !is_null( $file ) ) {
 684                         $command .=  " > " . wfEscapeShellArg( $file );
 685                 }
 686                 $this->handle = popen( $command, "w" );
 687         }
 688 }
 689
 690 /**
 691  * Sends dump output via the gzip compressor.
 692  * @ingroup Dump
 693  */
 694 class DumpGZipOutput extends DumpPipeOutput {
 695         function DumpGZipOutput( $file ) {
 696                 parent::DumpPipeOutput( "gzip", $file );
 697         }
 698 }
 699
 700 /**
 701  * Sends dump output via the bgzip2 compressor.
 702  * @ingroup Dump
 703  */
 704 class DumpBZip2Output extends DumpPipeOutput {
 705         function DumpBZip2Output( $file ) {
 706                 parent::DumpPipeOutput( "bzip2", $file );
 707         }
 708 }
 709
 710 /**
 711  * Sends dump output via the p7zip compressor.
 712  * @ingroup Dump
 713  */
 714 class Dump7ZipOutput extends DumpPipeOutput {
 715         function Dump7ZipOutput( $file ) {
 716                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 717                 // Suppress annoying useless crap from p7zip
 718                 // Unfortunately this could suppress real error messages too
 719                 $command .= ' >' . wfGetNull() . ' 2>&1';
 720                 parent::DumpPipeOutput( $command );
 721         }
 722 }
 723
 724
 725
 726 /**
 727  * Dump output filter class.
 728  * This just does output filtering and streaming; XML formatting is done
 729  * higher up, so be careful in what you do.
 730  * @ingroup Dump
 731  */
 732 class DumpFilter {
 733         function DumpFilter( &$sink ) {
 734                 $this->sink =& $sink;
 735         }
 736
 737         function writeOpenStream( $string ) {
 738                 $this->sink->writeOpenStream( $string );
 739         }
 740
 741         function writeCloseStream( $string ) {
 742                 $this->sink->writeCloseStream( $string );
 743         }
 744
 745         function writeOpenPage( $page, $string ) {
 746                 $this->sendingThisPage = $this->pass( $page, $string );
 747                 if( $this->sendingThisPage ) {
 748                         $this->sink->writeOpenPage( $page, $string );
 749                 }
 750         }
 751
 752         function writeClosePage( $string ) {
 753                 if( $this->sendingThisPage ) {
 754                         $this->sink->writeClosePage( $string );
 755                         $this->sendingThisPage = false;
 756                 }
 757         }
 758
 759         function writeRevision( $rev, $string ) {
 760                 if( $this->sendingThisPage ) {
 761                         $this->sink->writeRevision( $rev, $string );
 762                 }
 763         }
 764
 765         function writeLogItem( $rev, $string ) {
 766                 $this->sink->writeRevision( $rev, $string );
 767         }
 768
 769         /**
 770          * Override for page-based filter types.
 771          * @return bool
 772          */
 773         function pass( $page ) {
 774                 return true;
 775         }
 776 }
 777
 778 /**
 779  * Simple dump output filter to exclude all talk pages.
 780  * @ingroup Dump
 781  */
 782 class DumpNotalkFilter extends DumpFilter {
 783         function pass( $page ) {
 784                 return !MWNamespace::isTalk( $page->page_namespace );
 785         }
 786 }
 787
 788 /**
 789  * Dump output filter to include or exclude pages in a given set of namespaces.
 790  * @ingroup Dump
 791  */
 792 class DumpNamespaceFilter extends DumpFilter {
 793         var $invert = false;
 794         var $namespaces = array();
 795
 796         function DumpNamespaceFilter( &$sink, $param ) {
 797                 parent::DumpFilter( $sink );
 798
 799                 $constants = array(
 800                         "NS_MAIN"           => NS_MAIN,
 801                         "NS_TALK"           => NS_TALK,
 802                         "NS_USER"           => NS_USER,
 803                         "NS_USER_TALK"      => NS_USER_TALK,
 804                         "NS_PROJECT"        => NS_PROJECT,
 805                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
 806                         "NS_FILE"           => NS_FILE,
 807                         "NS_FILE_TALK"      => NS_FILE_TALK,
 808                         "NS_IMAGE"          => NS_IMAGE,  // NS_IMAGE is an alias for NS_FILE
 809                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
 810                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
 811                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
 812                         "NS_TEMPLATE"       => NS_TEMPLATE,
 813                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
 814                         "NS_HELP"           => NS_HELP,
 815                         "NS_HELP_TALK"      => NS_HELP_TALK,
 816                         "NS_CATEGORY"       => NS_CATEGORY,
 817                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
 818
 819                 if( $param{0} == '!' ) {
 820                         $this->invert = true;
 821                         $param = substr( $param, 1 );
 822                 }
 823
 824                 foreach( explode( ',', $param ) as $key ) {
 825                         $key = trim( $key );
 826                         if( isset( $constants[$key] ) ) {
 827                                 $ns = $constants[$key];
 828                                 $this->namespaces[$ns] = true;
 829                         } elseif( is_numeric( $key ) ) {
 830                                 $ns = intval( $key );
 831                                 $this->namespaces[$ns] = true;
 832                         } else {
 833                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
 834                         }
 835                 }
 836         }
 837
 838         function pass( $page ) {
 839                 $match = isset( $this->namespaces[$page->page_namespace] );
 840                 return $this->invert xor $match;
 841         }
 842 }
 843
 844
 845 /**
 846  * Dump output filter to include only the last revision in each page sequence.
 847  * @ingroup Dump
 848  */
 849 class DumpLatestFilter extends DumpFilter {
 850         var $page, $pageString, $rev, $revString;
 851
 852         function writeOpenPage( $page, $string ) {
 853                 $this->page = $page;
 854                 $this->pageString = $string;
 855         }
 856
 857         function writeClosePage( $string ) {
 858                 if( $this->rev ) {
 859                         $this->sink->writeOpenPage( $this->page, $this->pageString );
 860                         $this->sink->writeRevision( $this->rev, $this->revString );
 861                         $this->sink->writeClosePage( $string );
 862                 }
 863                 $this->rev = null;
 864                 $this->revString = null;
 865                 $this->page = null;
 866                 $this->pageString = null;
 867         }
 868
 869         function writeRevision( $rev, $string ) {
 870                 if( $rev->rev_id == $this->page->page_latest ) {
 871                         $this->rev = $rev;
 872                         $this->revString = $string;
 873                 }
 874         }
 875 }
 876
 877 /**
 878  * Base class for output stream; prints to stdout or buffer or whereever.
 879  * @ingroup Dump
 880  */
 881 class DumpMultiWriter {
 882         function DumpMultiWriter( $sinks ) {
 883                 $this->sinks = $sinks;
 884                 $this->count = count( $sinks );
 885         }
 886
 887         function writeOpenStream( $string ) {
 888                 for( $i = 0; $i < $this->count; $i++ ) {
 889                         $this->sinks[$i]->writeOpenStream( $string );
 890                 }
 891         }
 892
 893         function writeCloseStream( $string ) {
 894                 for( $i = 0; $i < $this->count; $i++ ) {
 895                         $this->sinks[$i]->writeCloseStream( $string );
 896                 }
 897         }
 898
 899         function writeOpenPage( $page, $string ) {
 900                 for( $i = 0; $i < $this->count; $i++ ) {
 901                         $this->sinks[$i]->writeOpenPage( $page, $string );
 902                 }
 903         }
 904
 905         function writeClosePage( $string ) {
 906                 for( $i = 0; $i < $this->count; $i++ ) {
 907                         $this->sinks[$i]->writeClosePage( $string );
 908                 }
 909         }
 910
 911         function writeRevision( $rev, $string ) {
 912                 for( $i = 0; $i < $this->count; $i++ ) {
 913                         $this->sinks[$i]->writeRevision( $rev, $string );
 914                 }
 915         }
 916 }
 917
 918 function xmlsafe( $string ) {
 919         wfProfileIn( __FUNCTION__ );
 920
 921         /**
 922          * The page may contain old data which has not been properly normalized.
 923          * Invalid UTF-8 sequences or forbidden control characters will make our
 924          * XML output invalid, so be sure to strip them out.
 925          */
 926         $string = UtfNormal::cleanUp( $string );
 927
 928         $string = htmlspecialchars( $string );
 929         wfProfileOut( __FUNCTION__ );
 930         return $string;
 931 }