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                 $nothidden = '('.$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, $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                         $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                         # 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         }
 329
 330         protected function outputLogStream( $resultset ) {
 331                 while( $row = $resultset->fetchObject() ) {
 332                         $output = $this->writer->writeLogItem( $row );
 333                         $this->sink->writeLogItem( $row, $output );
 334                 }
 335         }
 336 }
 337
 338 /**
 339  * @ingroup Dump
 340  */
 341 class XmlDumpWriter {
 342
 343         /**
 344          * Returns the export schema version.
 345          * @return string
 346          */
 347         function schemaVersion() {
 348                 return "0.3"; // FIXME: upgrade to 0.4 when updated XSD is ready, for the revision deletion bits
 349         }
 350
 351         /**
 352          * Opens the XML output stream's root <mediawiki> element.
 353          * This does not include an xml directive, so is safe to include
 354          * as a subelement in a larger XML stream. Namespace and XML Schema
 355          * references are included.
 356          *
 357          * Output will be encoded in UTF-8.
 358          *
 359          * @return string
 360          */
 361         function openStream() {
 362                 global $wgContLanguageCode;
 363                 $ver = $this->schemaVersion();
 364                 return Xml::element( 'mediawiki', array(
 365                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 366                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 367                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 368                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 369                         'version'            => $ver,
 370                         'xml:lang'           => $wgContLanguageCode ),
 371                         null ) .
 372                         "\n" .
 373                         $this->siteInfo();
 374         }
 375
 376         function siteInfo() {
 377                 $info = array(
 378                         $this->sitename(),
 379                         $this->homelink(),
 380                         $this->generator(),
 381                         $this->caseSetting(),
 382                         $this->namespaces() );
 383                 return "  <siteinfo>\n    " .
 384                         implode( "\n    ", $info ) .
 385                         "\n  </siteinfo>\n";
 386         }
 387
 388         function sitename() {
 389                 global $wgSitename;
 390                 return Xml::element( 'sitename', array(), $wgSitename );
 391         }
 392
 393         function generator() {
 394                 global $wgVersion;
 395                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 396         }
 397
 398         function homelink() {
 399                 return Xml::element( 'base', array(), Title::newMainPage()->getFullUrl() );
 400         }
 401
 402         function caseSetting() {
 403                 global $wgCapitalLinks;
 404                 // "case-insensitive" option is reserved for future
 405                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 406                 return Xml::element( 'case', array(), $sensitivity );
 407         }
 408
 409         function namespaces() {
 410                 global $wgContLang;
 411                 $spaces = "<namespaces>\n";
 412                 foreach( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 413                         $spaces .= '      ' . Xml::element( 'namespace', array( 'key' => $ns ), $title ) . "\n";
 414                 }
 415                 $spaces .= "    </namespaces>";
 416                 return $spaces;
 417         }
 418
 419         /**
 420          * Closes the output stream with the closing root element.
 421          * Call when finished dumping things.
 422          */
 423         function closeStream() {
 424                 return "</mediawiki>\n";
 425         }
 426
 427
 428         /**
 429          * Opens a <page> section on the output stream, with data
 430          * from the given database row.
 431          *
 432          * @param $row object
 433          * @return string
 434          * @access private
 435          */
 436         function openPage( $row ) {
 437                 $out = "  <page>\n";
 438                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 439                 $out .= '    ' . Xml::elementClean( 'title', array(), $title->getPrefixedText() ) . "\n";
 440                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 441                 if( '' != $row->page_restrictions ) {
 442                         $out .= '    ' . Xml::element( 'restrictions', array(),
 443                                 strval( $row->page_restrictions ) ) . "\n";
 444                 }
 445                 return $out;
 446         }
 447
 448         /**
 449          * Closes a <page> section on the output stream.
 450          *
 451          * @access private
 452          */
 453         function closePage() {
 454                 return "  </page>\n";
 455         }
 456
 457         /**
 458          * Dumps a <revision> section on the output stream, with
 459          * data filled in from the given database row.
 460          *
 461          * @param $row object
 462          * @return string
 463          * @access private
 464          */
 465         function writeRevision( $row ) {
 466                 $fname = 'WikiExporter::dumpRev';
 467                 wfProfileIn( $fname );
 468
 469                 $out  = "    <revision>\n";
 470                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 471
 472                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 473
 474                 if( $row->rev_deleted & Revision::DELETED_USER ) {
 475                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 476                 } else {
 477                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 478                 }
 479
 480                 if( $row->rev_minor_edit ) {
 481                         $out .=  "      <minor/>\n";
 482                 }
 483                 if( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 484                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 485                 } elseif( $row->rev_comment != '' ) {
 486                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 487                 }
 488
 489                 if( $row->rev_deleted & Revision::DELETED_TEXT ) {
 490                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 491                 } elseif( isset( $row->old_text ) ) {
 492                         // Raw text from the database may have invalid chars
 493                         $text = strval( Revision::getRevisionText( $row ) );
 494                         $out .= "      " . Xml::elementClean( 'text',
 495                                 array( 'xml:space' => 'preserve' ),
 496                                 strval( $text ) ) . "\n";
 497                 } else {
 498                         // Stub output
 499                         $out .= "      " . Xml::element( 'text',
 500                                 array( 'id' => $row->rev_text_id ),
 501                                 "" ) . "\n";
 502                 }
 503
 504                 $out .= "    </revision>\n";
 505
 506                 wfProfileOut( $fname );
 507                 return $out;
 508         }
 509
 510         /**
 511          * Dumps a <logitem> section on the output stream, with
 512          * data filled in from the given database row.
 513          *
 514          * @param $row object
 515          * @return string
 516          * @access private
 517          */
 518         function writeLogItem( $row ) {
 519                 $fname = 'WikiExporter::writeLogItem';
 520                 wfProfileIn( $fname );
 521
 522                 $out  = "    <logitem>\n";
 523                 $out .= "      " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 524
 525                 $out .= $this->writeTimestamp( $row->log_timestamp );
 526
 527                 if( $row->log_deleted & LogPage::DELETED_USER ) {
 528                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 529                 } else {
 530                         $out .= $this->writeContributor( $row->log_user, $row->user_name );
 531                 }
 532
 533                 if( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 534                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 535                 } elseif( $row->log_comment != '' ) {
 536                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 537                 }
 538
 539                 $out .= "      " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 540                 $out .= "      " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 541
 542                 if( $row->log_deleted & LogPage::DELETED_ACTION ) {
 543                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 544                 } else {
 545                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 546                         $out .= "      " . Xml::elementClean( 'logtitle', null, $title->getPrefixedText() ) . "\n";
 547                         $out .= "      " . Xml::elementClean( 'params',
 548                                 array( 'xml:space' => 'preserve' ),
 549                                 strval( $row->log_params ) ) . "\n";
 550                 }
 551
 552                 $out .= "    </logitem>\n";
 553
 554                 wfProfileOut( $fname );
 555                 return $out;
 556         }
 557
 558         function writeTimestamp( $timestamp ) {
 559                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 560                 return "      " . Xml::element( 'timestamp', null, $ts ) . "\n";
 561         }
 562
 563         function writeContributor( $id, $text ) {
 564                 $out = "      <contributor>\n";
 565                 if( $id ) {
 566                         $out .= "        " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 567                         $out .= "        " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 568                 } else {
 569                         $out .= "        " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 570                 }
 571                 $out .= "      </contributor>\n";
 572                 return $out;
 573         }
 574
 575         /**
 576          * Warning! This data is potentially inconsistent. :(
 577          */
 578         function writeUploads( $row ) {
 579                 if( $row->page_namespace == NS_IMAGE ) {
 580                         $img = wfFindFile( $row->page_title );
 581                         if( $img ) {
 582                                 $out = '';
 583                                 foreach( array_reverse( $img->getHistory() ) as $ver ) {
 584                                         $out .= $this->writeUpload( $ver );
 585                                 }
 586                                 $out .= $this->writeUpload( $img );
 587                                 return $out;
 588                         }
 589                 }
 590                 return '';
 591         }
 592
 593         function writeUpload( $file ) {
 594                 return "    <upload>\n" .
 595                         $this->writeTimestamp( $file->getTimestamp() ) .
 596                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 597                         "      " . Xml::elementClean( 'comment', null, $file->getDescription() ) . "\n" .
 598                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 599                         "      " . Xml::element( 'src', null, $file->getFullUrl() ) . "\n" .
 600                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 601                         "    </upload>\n";
 602         }
 603
 604 }
 605
 606
 607 /**
 608  * Base class for output stream; prints to stdout or buffer or whereever.
 609  * @ingroup Dump
 610  */
 611 class DumpOutput {
 612         function writeOpenStream( $string ) {
 613                 $this->write( $string );
 614         }
 615
 616         function writeCloseStream( $string ) {
 617                 $this->write( $string );
 618         }
 619
 620         function writeOpenPage( $page, $string ) {
 621                 $this->write( $string );
 622         }
 623
 624         function writeClosePage( $string ) {
 625                 $this->write( $string );
 626         }
 627
 628         function writeRevision( $rev, $string ) {
 629                 $this->write( $string );
 630         }
 631
 632         function writeLogItem( $rev, $string ) {
 633                 $this->write( $string );
 634         }
 635
 636         /**
 637          * Override to write to a different stream type.
 638          * @return bool
 639          */
 640         function write( $string ) {
 641                 print $string;
 642         }
 643 }
 644
 645 /**
 646  * Stream outputter to send data to a file.
 647  * @ingroup Dump
 648  */
 649 class DumpFileOutput extends DumpOutput {
 650         var $handle;
 651
 652         function DumpFileOutput( $file ) {
 653                 $this->handle = fopen( $file, "wt" );
 654         }
 655
 656         function write( $string ) {
 657                 fputs( $this->handle, $string );
 658         }
 659 }
 660
 661 /**
 662  * Stream outputter to send data to a file via some filter program.
 663  * Even if compression is available in a library, using a separate
 664  * program can allow us to make use of a multi-processor system.
 665  * @ingroup Dump
 666  */
 667 class DumpPipeOutput extends DumpFileOutput {
 668         function DumpPipeOutput( $command, $file = null ) {
 669                 if( !is_null( $file ) ) {
 670                         $command .=  " > " . wfEscapeShellArg( $file );
 671                 }
 672                 $this->handle = popen( $command, "w" );
 673         }
 674 }
 675
 676 /**
 677  * Sends dump output via the gzip compressor.
 678  * @ingroup Dump
 679  */
 680 class DumpGZipOutput extends DumpPipeOutput {
 681         function DumpGZipOutput( $file ) {
 682                 parent::DumpPipeOutput( "gzip", $file );
 683         }
 684 }
 685
 686 /**
 687  * Sends dump output via the bgzip2 compressor.
 688  * @ingroup Dump
 689  */
 690 class DumpBZip2Output extends DumpPipeOutput {
 691         function DumpBZip2Output( $file ) {
 692                 parent::DumpPipeOutput( "bzip2", $file );
 693         }
 694 }
 695
 696 /**
 697  * Sends dump output via the p7zip compressor.
 698  * @ingroup Dump
 699  */
 700 class Dump7ZipOutput extends DumpPipeOutput {
 701         function Dump7ZipOutput( $file ) {
 702                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 703                 // Suppress annoying useless crap from p7zip
 704                 // Unfortunately this could suppress real error messages too
 705                 $command .= ' >' . wfGetNull() . ' 2>&1';
 706                 parent::DumpPipeOutput( $command );
 707         }
 708 }
 709
 710
 711
 712 /**
 713  * Dump output filter class.
 714  * This just does output filtering and streaming; XML formatting is done
 715  * higher up, so be careful in what you do.
 716  * @ingroup Dump
 717  */
 718 class DumpFilter {
 719         function DumpFilter( &$sink ) {
 720                 $this->sink =& $sink;
 721         }
 722
 723         function writeOpenStream( $string ) {
 724                 $this->sink->writeOpenStream( $string );
 725         }
 726
 727         function writeCloseStream( $string ) {
 728                 $this->sink->writeCloseStream( $string );
 729         }
 730
 731         function writeOpenPage( $page, $string ) {
 732                 $this->sendingThisPage = $this->pass( $page, $string );
 733                 if( $this->sendingThisPage ) {
 734                         $this->sink->writeOpenPage( $page, $string );
 735                 }
 736         }
 737
 738         function writeClosePage( $string ) {
 739                 if( $this->sendingThisPage ) {
 740                         $this->sink->writeClosePage( $string );
 741                         $this->sendingThisPage = false;
 742                 }
 743         }
 744
 745         function writeRevision( $rev, $string ) {
 746                 if( $this->sendingThisPage ) {
 747                         $this->sink->writeRevision( $rev, $string );
 748                 }
 749         }
 750
 751         function writeLogItem( $rev, $string ) {
 752                 $this->sink->writeRevision( $rev, $string );
 753         }
 754
 755         /**
 756          * Override for page-based filter types.
 757          * @return bool
 758          */
 759         function pass( $page ) {
 760                 return true;
 761         }
 762 }
 763
 764 /**
 765  * Simple dump output filter to exclude all talk pages.
 766  * @ingroup Dump
 767  */
 768 class DumpNotalkFilter extends DumpFilter {
 769         function pass( $page ) {
 770                 return !MWNamespace::isTalk( $page->page_namespace );
 771         }
 772 }
 773
 774 /**
 775  * Dump output filter to include or exclude pages in a given set of namespaces.
 776  * @ingroup Dump
 777  */
 778 class DumpNamespaceFilter extends DumpFilter {
 779         var $invert = false;
 780         var $namespaces = array();
 781
 782         function DumpNamespaceFilter( &$sink, $param ) {
 783                 parent::DumpFilter( $sink );
 784
 785                 $constants = array(
 786                         "NS_MAIN"           => NS_MAIN,
 787                         "NS_TALK"           => NS_TALK,
 788                         "NS_USER"           => NS_USER,
 789                         "NS_USER_TALK"      => NS_USER_TALK,
 790                         "NS_PROJECT"        => NS_PROJECT,
 791                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
 792                         "NS_FILE"           => NS_FILE,
 793                         "NS_FILE_TALK"      => NS_FILE_TALK,
 794                         "NS_IMAGE"          => NS_IMAGE,  // NS_IMAGE is an alias for NS_FILE
 795                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
 796                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
 797                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
 798                         "NS_TEMPLATE"       => NS_TEMPLATE,
 799                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
 800                         "NS_HELP"           => NS_HELP,
 801                         "NS_HELP_TALK"      => NS_HELP_TALK,
 802                         "NS_CATEGORY"       => NS_CATEGORY,
 803                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
 804
 805                 if( $param{0} == '!' ) {
 806                         $this->invert = true;
 807                         $param = substr( $param, 1 );
 808                 }
 809
 810                 foreach( explode( ',', $param ) as $key ) {
 811                         $key = trim( $key );
 812                         if( isset( $constants[$key] ) ) {
 813                                 $ns = $constants[$key];
 814                                 $this->namespaces[$ns] = true;
 815                         } elseif( is_numeric( $key ) ) {
 816                                 $ns = intval( $key );
 817                                 $this->namespaces[$ns] = true;
 818                         } else {
 819                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
 820                         }
 821                 }
 822         }
 823
 824         function pass( $page ) {
 825                 $match = isset( $this->namespaces[$page->page_namespace] );
 826                 return $this->invert xor $match;
 827         }
 828 }
 829
 830
 831 /**
 832  * Dump output filter to include only the last revision in each page sequence.
 833  * @ingroup Dump
 834  */
 835 class DumpLatestFilter extends DumpFilter {
 836         var $page, $pageString, $rev, $revString;
 837
 838         function writeOpenPage( $page, $string ) {
 839                 $this->page = $page;
 840                 $this->pageString = $string;
 841         }
 842
 843         function writeClosePage( $string ) {
 844                 if( $this->rev ) {
 845                         $this->sink->writeOpenPage( $this->page, $this->pageString );
 846                         $this->sink->writeRevision( $this->rev, $this->revString );
 847                         $this->sink->writeClosePage( $string );
 848                 }
 849                 $this->rev = null;
 850                 $this->revString = null;
 851                 $this->page = null;
 852                 $this->pageString = null;
 853         }
 854
 855         function writeRevision( $rev, $string ) {
 856                 if( $rev->rev_id == $this->page->page_latest ) {
 857                         $this->rev = $rev;
 858                         $this->revString = $string;
 859                 }
 860         }
 861 }
 862
 863 /**
 864  * Base class for output stream; prints to stdout or buffer or whereever.
 865  * @ingroup Dump
 866  */
 867 class DumpMultiWriter {
 868         function DumpMultiWriter( $sinks ) {
 869                 $this->sinks = $sinks;
 870                 $this->count = count( $sinks );
 871         }
 872
 873         function writeOpenStream( $string ) {
 874                 for( $i = 0; $i < $this->count; $i++ ) {
 875                         $this->sinks[$i]->writeOpenStream( $string );
 876                 }
 877         }
 878
 879         function writeCloseStream( $string ) {
 880                 for( $i = 0; $i < $this->count; $i++ ) {
 881                         $this->sinks[$i]->writeCloseStream( $string );
 882                 }
 883         }
 884
 885         function writeOpenPage( $page, $string ) {
 886                 for( $i = 0; $i < $this->count; $i++ ) {
 887                         $this->sinks[$i]->writeOpenPage( $page, $string );
 888                 }
 889         }
 890
 891         function writeClosePage( $string ) {
 892                 for( $i = 0; $i < $this->count; $i++ ) {
 893                         $this->sinks[$i]->writeClosePage( $string );
 894                 }
 895         }
 896
 897         function writeRevision( $rev, $string ) {
 898                 for( $i = 0; $i < $this->count; $i++ ) {
 899                         $this->sinks[$i]->writeRevision( $rev, $string );
 900                 }
 901         }
 902 }
 903
 904 function xmlsafe( $string ) {
 905         $fname = 'xmlsafe';
 906         wfProfileIn( $fname );
 907
 908         /**
 909          * The page may contain old data which has not been properly normalized.
 910          * Invalid UTF-8 sequences or forbidden control characters will make our
 911          * XML output invalid, so be sure to strip them out.
 912          */
 913         $string = UtfNormal::cleanUp( $string );
 914
 915         $string = htmlspecialchars( $string );
 916         wfProfileOut( $fname );
 917         return $string;
 918 }