includes/Export.php

   1 <?php
   2 # Copyright (C) 2003, 2005, 2006 Brion Vibber <brion@pobox.com>
   3 # http://www.mediawiki.org/
   4 #
   5 # This program is free software; you can redistribute it and/or modify
   6 # it under the terms of the GNU General Public License as published by
   7 # the Free Software Foundation; either version 2 of the License, or
   8 # (at your option) any later version.
   9 #
  10 # This program is distributed in the hope that it will be useful,
  11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13 # GNU General Public License for more details.
  14 #
  15 # You should have received a copy of the GNU General Public License along
  16 # with this program; if not, write to the Free Software Foundation, Inc.,
  17 # 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18 # http://www.gnu.org/copyleft/gpl.html
  19
  20 /**
  21  * @defgroup Dump Dump
  22  */
  23
  24 /**
  25  * @ingroup SpecialPage Dump
  26  */
  27 class WikiExporter {
  28         var $list_authors = false ; # Return distinct author list (when not returning full history)
  29         var $author_list = "" ;
  30
  31         var $dumpUploads = false;
  32
  33         const FULL = 0;
  34         const CURRENT = 1;
  35         const LOGS = 2;
  36
  37         const BUFFER = 0;
  38         const STREAM = 1;
  39
  40         const TEXT = 0;
  41         const STUB = 1;
  42
  43         /**
  44          * If using WikiExporter::STREAM to stream a large amount of data,
  45          * provide a database connection which is not managed by
  46          * LoadBalancer to read from: some history blob types will
  47          * make additional queries to pull source data while the
  48          * main query is still running.
  49          *
  50          * @param $db Database
  51          * @param $history Mixed: one of WikiExporter::FULL or WikiExporter::CURRENT,
  52          *                 or an associative array:
  53          *                   offset: non-inclusive offset at which to start the query
  54          *                   limit: maximum number of rows to return
  55          *                   dir: "asc" or "desc" timestamp order
  56          * @param $buffer Int: one of WikiExporter::BUFFER or WikiExporter::STREAM
  57          */
  58         function __construct( &$db, $history = WikiExporter::CURRENT,
  59                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  60                 $this->db =& $db;
  61                 $this->history = $history;
  62                 $this->buffer  = $buffer;
  63                 $this->writer  = new XmlDumpWriter();
  64                 $this->sink    = new DumpOutput();
  65                 $this->text    = $text;
  66         }
  67
  68         /**
  69          * Set the DumpOutput or DumpFilter object which will receive
  70          * various row objects and XML output for filtering. Filters
  71          * can be chained or used as callbacks.
  72          *
  73          * @param $sink mixed
  74          */
  75         public function setOutputSink( &$sink ) {
  76                 $this->sink =& $sink;
  77         }
  78
  79         public function openStream() {
  80                 $output = $this->writer->openStream();
  81                 $this->sink->writeOpenStream( $output );
  82         }
  83
  84         public function closeStream() {
  85                 $output = $this->writer->closeStream();
  86                 $this->sink->writeCloseStream( $output );
  87         }
  88
  89         /**
  90          * Dumps a series of page and revision records for all pages
  91          * in the database, either including complete history or only
  92          * the most recent version.
  93          */
  94         public function allPages() {
  95                 return $this->dumpFrom( '' );
  96         }
  97
  98         /**
  99          * Dumps a series of page and revision records for those pages
 100          * in the database falling within the page_id range given.
 101          * @param $start Int: inclusive lower limit (this id is included)
 102          * @param $end   Int: Exclusive upper limit (this id is not included)
 103          *                   If 0, no upper limit.
 104          */
 105         public function pagesByRange( $start, $end ) {
 106                 $condition = 'page_id >= ' . intval( $start );
 107                 if( $end ) {
 108                         $condition .= ' AND page_id < ' . intval( $end );
 109                 }
 110                 return $this->dumpFrom( $condition );
 111         }
 112
 113         /**
 114          * @param $title Title
 115          */
 116         public function pageByTitle( $title ) {
 117                 return $this->dumpFrom(
 118                         'page_namespace=' . $title->getNamespace() .
 119                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 120         }
 121
 122         public function pageByName( $name ) {
 123                 $title = Title::newFromText( $name );
 124                 if( is_null( $title ) ) {
 125                         return new WikiError( "Can't export invalid title" );
 126                 } else {
 127                         return $this->pageByTitle( $title );
 128                 }
 129         }
 130
 131         public function pagesByName( $names ) {
 132                 foreach( $names as $name ) {
 133                         $this->pageByName( $name );
 134                 }
 135         }
 136
 137         public function allLogs() {
 138                 return $this->dumpFrom( '' );
 139         }
 140
 141         public function logsByRange( $start, $end ) {
 142                 $condition = 'log_id >= ' . intval( $start );
 143                 if( $end ) {
 144                         $condition .= ' AND log_id < ' . intval( $end );
 145                 }
 146                 return $this->dumpFrom( $condition );
 147         }
 148
 149         # Generates the distinct list of authors of an article
 150         # Not called by default (depends on $this->list_authors)
 151         # Can be set by Special:Export when not exporting whole history
 152         protected function do_list_authors( $page , $revision , $cond ) {
 153                 $fname = "do_list_authors" ;
 154                 wfProfileIn( $fname );
 155                 $this->author_list = "<contributors>";
 156                 //rev_deleted
 157                 $nothidden = '(rev_deleted & '.Revision::DELETED_USER.') = 0';
 158
 159                 $sql = "SELECT DISTINCT rev_user_text,rev_user FROM {$page},{$revision}
 160                 WHERE page_id=rev_page AND $nothidden AND " . $cond ;
 161                 $result = $this->db->query( $sql, $fname );
 162                 $resultset = $this->db->resultObject( $result );
 163                 while( $row = $resultset->fetchObject() ) {
 164                         $this->author_list .= "<contributor>" .
 165                                 "<username>" .
 166                                 htmlentities( $row->rev_user_text )  .
 167                                 "</username>" .
 168                                 "<id>" .
 169                                 $row->rev_user .
 170                                 "</id>" .
 171                                 "</contributor>";
 172                 }
 173                 wfProfileOut( $fname );
 174                 $this->author_list .= "</contributors>";
 175         }
 176
 177         protected function dumpFrom( $cond = '' ) {
 178                 $fname = 'WikiExporter::dumpFrom';
 179                 wfProfileIn( $fname );
 180
 181                 # For logs dumps...
 182                 if( $this->history & self::LOGS ) {
 183                         $where = array( 'user_id = log_user' );
 184                         # Hide private logs
 185                         $where[] = LogEventsList::getExcludeClause( $this->db );
 186                         if( $cond ) $where[] = $cond;
 187                         $result = $this->db->select( array('logging','user'),
 188                                 '*',
 189                                 $where,
 190                                 $fname,
 191                                 array( 'ORDER BY' => 'log_id', 'USE INDEX' => array('logging' => 'PRIMARY') )
 192                         );
 193                         $wrapper = $this->db->resultObject( $result );
 194                         $this->outputLogStream( $wrapper );
 195                 # For page dumps...
 196                 } else {
 197                         list($page,$revision,$text) = $this->db->tableNamesN('page','revision','text');
 198
 199                         $order = 'ORDER BY page_id';
 200                         $limit = '';
 201
 202                         if( $this->history == WikiExporter::FULL ) {
 203                                 $join = 'page_id=rev_page';
 204                         } elseif( $this->history == WikiExporter::CURRENT ) {
 205                                 if ( $this->list_authors && $cond != '' )  { // List authors, if so desired
 206                                         $this->do_list_authors ( $page , $revision , $cond );
 207                                 }
 208                                 $join = 'page_id=rev_page AND page_latest=rev_id';
 209                         } elseif ( is_array( $this->history ) ) {
 210                                 $join = 'page_id=rev_page';
 211                                 if ( $this->history['dir'] == 'asc' ) {
 212                                         $op = '>';
 213                                         $order .= ', rev_timestamp';
 214                                 } else {
 215                                         $op = '<';
 216                                         $order .= ', rev_timestamp DESC';
 217                                 }
 218                                 if ( !empty( $this->history['offset'] ) ) {
 219                                         $join .= " AND rev_timestamp $op " . $this->db->addQuotes(
 220                                                 $this->db->timestamp( $this->history['offset'] ) );
 221                                 }
 222                                 if ( !empty( $this->history['limit'] ) ) {
 223                                         $limitNum = intval( $this->history['limit'] );
 224                                         if ( $limitNum > 0 ) {
 225                                                 $limit = "LIMIT $limitNum";
 226                                         }
 227                                 }
 228                         } else {
 229                                 wfProfileOut( $fname );
 230                                 return new WikiError( "$fname given invalid history dump type." );
 231                         }
 232                         $where = ( $cond == '' ) ? '' : "$cond AND";
 233
 234                         if( $this->buffer == WikiExporter::STREAM ) {
 235                                 $prev = $this->db->bufferResults( false );
 236                         }
 237                         if( $cond == '' ) {
 238                                 // Optimization hack for full-database dump
 239                                 $revindex = $pageindex = $this->db->useIndexClause("PRIMARY");
 240                                 $straight = ' /*! STRAIGHT_JOIN */ ';
 241                         } else {
 242                                 $pageindex = '';
 243                                 $revindex = '';
 244                                 $straight = '';
 245                         }
 246                         if( $this->text == WikiExporter::STUB ) {
 247                                 $sql = "SELECT $straight * FROM
 248                                         $page $pageindex,
 249                                         $revision $revindex
 250                                         WHERE $where $join
 251                                         $order $limit";
 252                         } else {
 253                                 $sql = "SELECT $straight * FROM
 254                                         $page $pageindex,
 255                                         $revision $revindex,
 256                                         $text
 257                                         WHERE $where $join AND rev_text_id=old_id
 258                                         $order $limit";
 259                         }
 260                         $result = $this->db->query( $sql, $fname );
 261                         $wrapper = $this->db->resultObject( $result );
 262                         $this->outputPageStream( $wrapper );
 263
 264                         if ( $this->list_authors ) {
 265                                 $this->outputPageStream( $wrapper );
 266                         }
 267
 268                         if( $this->buffer == WikiExporter::STREAM ) {
 269                                 $this->db->bufferResults( $prev );
 270                         }
 271                 }
 272                 wfProfileOut( $fname );
 273         }
 274
 275         /**
 276          * Runs through a query result set dumping page and revision records.
 277          * The result set should be sorted/grouped by page to avoid duplicate
 278          * page records in the output.
 279          *
 280          * The result set will be freed once complete. Should be safe for
 281          * streaming (non-buffered) queries, as long as it was made on a
 282          * separate database connection not managed by LoadBalancer; some
 283          * blob storage types will make queries to pull source data.
 284          *
 285          * @param $resultset ResultWrapper
 286          */
 287         protected function outputPageStream( $resultset ) {
 288                 $last = null;
 289                 while( $row = $resultset->fetchObject() ) {
 290                         if( is_null( $last ) ||
 291                                 $last->page_namespace != $row->page_namespace ||
 292                                 $last->page_title     != $row->page_title ) {
 293                                 if( isset( $last ) ) {
 294                                         $output = '';
 295                                         if( $this->dumpUploads ) {
 296                                                 $output .= $this->writer->writeUploads( $last );
 297                                         }
 298                                         $output .= $this->writer->closePage();
 299                                         $this->sink->writeClosePage( $output );
 300                                 }
 301                                 $output = $this->writer->openPage( $row );
 302                                 $this->sink->writeOpenPage( $row, $output );
 303                                 $last = $row;
 304                         }
 305                         $output = $this->writer->writeRevision( $row );
 306                         $this->sink->writeRevision( $row, $output );
 307                 }
 308                 if( isset( $last ) ) {
 309                         $output = '';
 310                         if( $this->dumpUploads ) {
 311                                 $output .= $this->writer->writeUploads( $last );
 312                         }
 313                         $output .= $this->author_list;
 314                         $output .= $this->writer->closePage();
 315                         $this->sink->writeClosePage( $output );
 316                 }
 317                 $resultset->free();
 318         }
 319
 320         protected function outputLogStream( $resultset ) {
 321                 while( $row = $resultset->fetchObject() ) {
 322                         $output = $this->writer->writeLogItem( $row );
 323                         $this->sink->writeLogItem( $row, $output );
 324                 }
 325                 $resultset->free();
 326         }
 327 }
 328
 329 /**
 330  * @ingroup Dump
 331  */
 332 class XmlDumpWriter {
 333
 334         /**
 335          * Returns the export schema version.
 336          * @return string
 337          */
 338         function schemaVersion() {
 339                 return "0.3"; // FIXME: upgrade to 0.4 when updated XSD is ready, for the revision deletion bits
 340         }
 341
 342         /**
 343          * Opens the XML output stream's root <mediawiki> element.
 344          * This does not include an xml directive, so is safe to include
 345          * as a subelement in a larger XML stream. Namespace and XML Schema
 346          * references are included.
 347          *
 348          * Output will be encoded in UTF-8.
 349          *
 350          * @return string
 351          */
 352         function openStream() {
 353                 global $wgContLanguageCode;
 354                 $ver = $this->schemaVersion();
 355                 return wfElement( 'mediawiki', array(
 356                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 357                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 358                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 359                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 360                         'version'            => $ver,
 361                         'xml:lang'           => $wgContLanguageCode ),
 362                         null ) .
 363                         "\n" .
 364                         $this->siteInfo();
 365         }
 366
 367         function siteInfo() {
 368                 $info = array(
 369                         $this->sitename(),
 370                         $this->homelink(),
 371                         $this->generator(),
 372                         $this->caseSetting(),
 373                         $this->namespaces() );
 374                 return "  <siteinfo>\n    " .
 375                         implode( "\n    ", $info ) .
 376                         "\n  </siteinfo>\n";
 377         }
 378
 379         function sitename() {
 380                 global $wgSitename;
 381                 return wfElement( 'sitename', array(), $wgSitename );
 382         }
 383
 384         function generator() {
 385                 global $wgVersion;
 386                 return wfElement( 'generator', array(), "MediaWiki $wgVersion" );
 387         }
 388
 389         function homelink() {
 390                 return wfElement( 'base', array(), Title::newMainPage()->getFullUrl() );
 391         }
 392
 393         function caseSetting() {
 394                 global $wgCapitalLinks;
 395                 // "case-insensitive" option is reserved for future
 396                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 397                 return wfElement( 'case', array(), $sensitivity );
 398         }
 399
 400         function namespaces() {
 401                 global $wgContLang;
 402                 $spaces = "  <namespaces>\n";
 403                 foreach( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 404                         $spaces .= '      ' . wfElement( 'namespace', array( 'key' => $ns ), $title ) . "\n";
 405                 }
 406                 $spaces .= "    </namespaces>";
 407                 return $spaces;
 408         }
 409
 410         /**
 411          * Closes the output stream with the closing root element.
 412          * Call when finished dumping things.
 413          */
 414         function closeStream() {
 415                 return "</mediawiki>\n";
 416         }
 417
 418
 419         /**
 420          * Opens a <page> section on the output stream, with data
 421          * from the given database row.
 422          *
 423          * @param $row object
 424          * @return string
 425          * @access private
 426          */
 427         function openPage( $row ) {
 428                 global $wgRestrictionTypes;
 429                 $out = "  <page>\n";
 430                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 431                 $out .= '    ' . wfElementClean( 'title', array(), $title->getPrefixedText() ) . "\n";
 432                 $out .= '    ' . wfElement( 'id', array(), strval( $row->page_id ) ) . "\n";
 433                 # Get page restrictions
 434                 $restrictions = array();
 435                 foreach( $wgRestrictionTypes as $action ) {
 436                         $restrictions[$action] = implode( '', $title->getRestrictions( $action ) );
 437                 }
 438                 $restrictions = Article::flattenRestrictions( $restrictions );
 439                 if( '' != $restrictions ) {
 440                         $out .= '    ' . wfElement( 'restrictions', array(),
 441                                 strval( $restrictions ) ) . "\n";
 442                 }
 443                 return $out;
 444         }
 445
 446         /**
 447          * Closes a <page> section on the output stream.
 448          *
 449          * @access private
 450          */
 451         function closePage() {
 452                 return "  </page>\n";
 453         }
 454
 455         /**
 456          * Dumps a <revision> section on the output stream, with
 457          * data filled in from the given database row.
 458          *
 459          * @param $row object
 460          * @return string
 461          * @access private
 462          */
 463         function writeRevision( $row ) {
 464                 $fname = 'WikiExporter::dumpRev';
 465                 wfProfileIn( $fname );
 466
 467                 $out  = "    <revision>\n";
 468                 $out .= "      " . wfElement( 'id', null, strval( $row->rev_id ) ) . "\n";
 469
 470                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 471
 472                 if( $row->rev_deleted & Revision::DELETED_USER ) {
 473                         $out .= "      " . wfElement( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 474                 } else {
 475                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 476                 }
 477
 478                 if( $row->rev_minor_edit ) {
 479                         $out .=  "      <minor/>\n";
 480                 }
 481                 if( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 482                         $out .= "      " . wfElement( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 483                 } elseif( $row->rev_comment != '' ) {
 484                         $out .= "      " . wfElementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 485                 }
 486
 487                 if( $row->rev_deleted & Revision::DELETED_TEXT ) {
 488                         $out .= "      " . wfElement( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 489                 } elseif( isset( $row->old_text ) ) {
 490                         // Raw text from the database may have invalid chars
 491                         $text = strval( Revision::getRevisionText( $row ) );
 492                         $out .= "      " . wfElementClean( 'text',
 493                                 array( 'xml:space' => 'preserve' ),
 494                                 strval( $text ) ) . "\n";
 495                 } else {
 496                         // Stub output
 497                         $out .= "      " . wfElement( 'text',
 498                                 array( 'id' => $row->rev_text_id ),
 499                                 "" ) . "\n";
 500                 }
 501
 502                 $out .= "    </revision>\n";
 503
 504                 wfProfileOut( $fname );
 505                 return $out;
 506         }
 507
 508         /**
 509          * Dumps a <logitem> section on the output stream, with
 510          * data filled in from the given database row.
 511          *
 512          * @param $row object
 513          * @return string
 514          * @access private
 515          */
 516         function writeLogItem( $row ) {
 517                 $fname = 'WikiExporter::writeLogItem';
 518                 wfProfileIn( $fname );
 519
 520                 $out  = "    <logitem>\n";
 521                 $out .= "      " . wfElement( 'id', null, strval( $row->log_id ) ) . "\n";
 522
 523                 $out .= $this->writeTimestamp( $row->log_timestamp );
 524
 525                 if( $row->log_deleted & LogPage::DELETED_USER ) {
 526                         $out .= "      " . wfElement( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 527                 } else {
 528                         $out .= $this->writeContributor( $row->log_user, $row->user_name );
 529                 }
 530
 531                 if( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 532                         $out .= "      " . wfElement( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 533                 } elseif( $row->log_comment != '' ) {
 534                         $out .= "      " . wfElementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 535                 }
 536
 537                 $out .= "      " . wfElement( 'type', null, strval( $row->log_type ) ) . "\n";
 538                 $out .= "      " . wfElement( 'action', null, strval( $row->log_action ) ) . "\n";
 539
 540                 if( $row->log_deleted & LogPage::DELETED_ACTION ) {
 541                         $out .= "      " . wfElement( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 542                 } else {
 543                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 544                         $out .= "      " . wfElementClean( 'logtitle', null, $title->getPrefixedText() ) . "\n";
 545                         $out .= "      " . wfElementClean( 'params',
 546                                 array( 'xml:space' => 'preserve' ),
 547                                 strval( $row->log_params ) ) . "\n";
 548                 }
 549
 550                 $out .= "    </logitem>\n";
 551
 552                 wfProfileOut( $fname );
 553                 return $out;
 554         }
 555
 556         function writeTimestamp( $timestamp ) {
 557                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 558                 return "      " . wfElement( 'timestamp', null, $ts ) . "\n";
 559         }
 560
 561         function writeContributor( $id, $text ) {
 562                 $out = "      <contributor>\n";
 563                 if( $id ) {
 564                         $out .= "        " . wfElementClean( 'username', null, strval( $text ) ) . "\n";
 565                         $out .= "        " . wfElement( 'id', null, strval( $id ) ) . "\n";
 566                 } else {
 567                         $out .= "        " . wfElementClean( 'ip', null, strval( $text ) ) . "\n";
 568                 }
 569                 $out .= "      </contributor>\n";
 570                 return $out;
 571         }
 572
 573         /**
 574          * Warning! This data is potentially inconsistent. :(
 575          */
 576         function writeUploads( $row ) {
 577                 if( $row->page_namespace == NS_IMAGE ) {
 578                         $img = wfFindFile( $row->page_title );
 579                         if( $img ) {
 580                                 $out = '';
 581                                 foreach( array_reverse( $img->getHistory() ) as $ver ) {
 582                                         $out .= $this->writeUpload( $ver );
 583                                 }
 584                                 $out .= $this->writeUpload( $img );
 585                                 return $out;
 586                         }
 587                 }
 588                 return '';
 589         }
 590
 591         function writeUpload( $file ) {
 592                 return "    <upload>\n" .
 593                         $this->writeTimestamp( $file->getTimestamp() ) .
 594                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 595                         "      " . wfElementClean( 'comment', null, $file->getDescription() ) . "\n" .
 596                         "      " . wfElement( 'filename', null, $file->getName() ) . "\n" .
 597                         "      " . wfElement( 'src', null, $file->getFullUrl() ) . "\n" .
 598                         "      " . wfElement( 'size', null, $file->getSize() ) . "\n" .
 599                         "    </upload>\n";
 600         }
 601
 602 }
 603
 604
 605 /**
 606  * Base class for output stream; prints to stdout or buffer or whereever.
 607  * @ingroup Dump
 608  */
 609 class DumpOutput {
 610         function writeOpenStream( $string ) {
 611                 $this->write( $string );
 612         }
 613
 614         function writeCloseStream( $string ) {
 615                 $this->write( $string );
 616         }
 617
 618         function writeOpenPage( $page, $string ) {
 619                 $this->write( $string );
 620         }
 621
 622         function writeClosePage( $string ) {
 623                 $this->write( $string );
 624         }
 625
 626         function writeRevision( $rev, $string ) {
 627                 $this->write( $string );
 628         }
 629
 630         function writeLogItem( $rev, $string ) {
 631                 $this->write( $string );
 632         }
 633
 634         /**
 635          * Override to write to a different stream type.
 636          * @return bool
 637          */
 638         function write( $string ) {
 639                 print $string;
 640         }
 641 }
 642
 643 /**
 644  * Stream outputter to send data to a file.
 645  * @ingroup Dump
 646  */
 647 class DumpFileOutput extends DumpOutput {
 648         var $handle;
 649
 650         function DumpFileOutput( $file ) {
 651                 $this->handle = fopen( $file, "wt" );
 652         }
 653
 654         function write( $string ) {
 655                 fputs( $this->handle, $string );
 656         }
 657 }
 658
 659 /**
 660  * Stream outputter to send data to a file via some filter program.
 661  * Even if compression is available in a library, using a separate
 662  * program can allow us to make use of a multi-processor system.
 663  * @ingroup Dump
 664  */
 665 class DumpPipeOutput extends DumpFileOutput {
 666         function DumpPipeOutput( $command, $file = null ) {
 667                 if( !is_null( $file ) ) {
 668                         $command .=  " > " . wfEscapeShellArg( $file );
 669                 }
 670                 $this->handle = popen( $command, "w" );
 671         }
 672 }
 673
 674 /**
 675  * Sends dump output via the gzip compressor.
 676  * @ingroup Dump
 677  */
 678 class DumpGZipOutput extends DumpPipeOutput {
 679         function DumpGZipOutput( $file ) {
 680                 parent::DumpPipeOutput( "gzip", $file );
 681         }
 682 }
 683
 684 /**
 685  * Sends dump output via the bgzip2 compressor.
 686  * @ingroup Dump
 687  */
 688 class DumpBZip2Output extends DumpPipeOutput {
 689         function DumpBZip2Output( $file ) {
 690                 parent::DumpPipeOutput( "bzip2", $file );
 691         }
 692 }
 693
 694 /**
 695  * Sends dump output via the p7zip compressor.
 696  * @ingroup Dump
 697  */
 698 class Dump7ZipOutput extends DumpPipeOutput {
 699         function Dump7ZipOutput( $file ) {
 700                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 701                 // Suppress annoying useless crap from p7zip
 702                 // Unfortunately this could suppress real error messages too
 703                 $command .= ' >' . wfGetNull() . ' 2>&1';
 704                 parent::DumpPipeOutput( $command );
 705         }
 706 }
 707
 708
 709
 710 /**
 711  * Dump output filter class.
 712  * This just does output filtering and streaming; XML formatting is done
 713  * higher up, so be careful in what you do.
 714  * @ingroup Dump
 715  */
 716 class DumpFilter {
 717         function DumpFilter( &$sink ) {
 718                 $this->sink =& $sink;
 719         }
 720
 721         function writeOpenStream( $string ) {
 722                 $this->sink->writeOpenStream( $string );
 723         }
 724
 725         function writeCloseStream( $string ) {
 726                 $this->sink->writeCloseStream( $string );
 727         }
 728
 729         function writeOpenPage( $page, $string ) {
 730                 $this->sendingThisPage = $this->pass( $page, $string );
 731                 if( $this->sendingThisPage ) {
 732                         $this->sink->writeOpenPage( $page, $string );
 733                 }
 734         }
 735
 736         function writeClosePage( $string ) {
 737                 if( $this->sendingThisPage ) {
 738                         $this->sink->writeClosePage( $string );
 739                         $this->sendingThisPage = false;
 740                 }
 741         }
 742
 743         function writeRevision( $rev, $string ) {
 744                 if( $this->sendingThisPage ) {
 745                         $this->sink->writeRevision( $rev, $string );
 746                 }
 747         }
 748
 749         function writeLogItem( $rev, $string ) {
 750                 $this->sink->writeRevision( $rev, $string );
 751         }
 752
 753         /**
 754          * Override for page-based filter types.
 755          * @return bool
 756          */
 757         function pass( $page ) {
 758                 return true;
 759         }
 760 }
 761
 762 /**
 763  * Simple dump output filter to exclude all talk pages.
 764  * @ingroup Dump
 765  */
 766 class DumpNotalkFilter extends DumpFilter {
 767         function pass( $page ) {
 768                 return !MWNamespace::isTalk( $page->page_namespace );
 769         }
 770 }
 771
 772 /**
 773  * Dump output filter to include or exclude pages in a given set of namespaces.
 774  * @ingroup Dump
 775  */
 776 class DumpNamespaceFilter extends DumpFilter {
 777         var $invert = false;
 778         var $namespaces = array();
 779
 780         function DumpNamespaceFilter( &$sink, $param ) {
 781                 parent::DumpFilter( $sink );
 782
 783                 $constants = array(
 784                         "NS_MAIN"           => NS_MAIN,
 785                         "NS_TALK"           => NS_TALK,
 786                         "NS_USER"           => NS_USER,
 787                         "NS_USER_TALK"      => NS_USER_TALK,
 788                         "NS_PROJECT"        => NS_PROJECT,
 789                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
 790                         "NS_IMAGE"          => NS_IMAGE,
 791                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
 792                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
 793                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
 794                         "NS_TEMPLATE"       => NS_TEMPLATE,
 795                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
 796                         "NS_HELP"           => NS_HELP,
 797                         "NS_HELP_TALK"      => NS_HELP_TALK,
 798                         "NS_CATEGORY"       => NS_CATEGORY,
 799                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
 800
 801                 if( $param{0} == '!' ) {
 802                         $this->invert = true;
 803                         $param = substr( $param, 1 );
 804                 }
 805
 806                 foreach( explode( ',', $param ) as $key ) {
 807                         $key = trim( $key );
 808                         if( isset( $constants[$key] ) ) {
 809                                 $ns = $constants[$key];
 810                                 $this->namespaces[$ns] = true;
 811                         } elseif( is_numeric( $key ) ) {
 812                                 $ns = intval( $key );
 813                                 $this->namespaces[$ns] = true;
 814                         } else {
 815                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
 816                         }
 817                 }
 818         }
 819
 820         function pass( $page ) {
 821                 $match = isset( $this->namespaces[$page->page_namespace] );
 822                 return $this->invert xor $match;
 823         }
 824 }
 825
 826
 827 /**
 828  * Dump output filter to include only the last revision in each page sequence.
 829  * @ingroup Dump
 830  */
 831 class DumpLatestFilter extends DumpFilter {
 832         var $page, $pageString, $rev, $revString;
 833
 834         function writeOpenPage( $page, $string ) {
 835                 $this->page = $page;
 836                 $this->pageString = $string;
 837         }
 838
 839         function writeClosePage( $string ) {
 840                 if( $this->rev ) {
 841                         $this->sink->writeOpenPage( $this->page, $this->pageString );
 842                         $this->sink->writeRevision( $this->rev, $this->revString );
 843                         $this->sink->writeClosePage( $string );
 844                 }
 845                 $this->rev = null;
 846                 $this->revString = null;
 847                 $this->page = null;
 848                 $this->pageString = null;
 849         }
 850
 851         function writeRevision( $rev, $string ) {
 852                 if( $rev->rev_id == $this->page->page_latest ) {
 853                         $this->rev = $rev;
 854                         $this->revString = $string;
 855                 }
 856         }
 857 }
 858
 859 /**
 860  * Base class for output stream; prints to stdout or buffer or whereever.
 861  * @ingroup Dump
 862  */
 863 class DumpMultiWriter {
 864         function DumpMultiWriter( $sinks ) {
 865                 $this->sinks = $sinks;
 866                 $this->count = count( $sinks );
 867         }
 868
 869         function writeOpenStream( $string ) {
 870                 for( $i = 0; $i < $this->count; $i++ ) {
 871                         $this->sinks[$i]->writeOpenStream( $string );
 872                 }
 873         }
 874
 875         function writeCloseStream( $string ) {
 876                 for( $i = 0; $i < $this->count; $i++ ) {
 877                         $this->sinks[$i]->writeCloseStream( $string );
 878                 }
 879         }
 880
 881         function writeOpenPage( $page, $string ) {
 882                 for( $i = 0; $i < $this->count; $i++ ) {
 883                         $this->sinks[$i]->writeOpenPage( $page, $string );
 884                 }
 885         }
 886
 887         function writeClosePage( $string ) {
 888                 for( $i = 0; $i < $this->count; $i++ ) {
 889                         $this->sinks[$i]->writeClosePage( $string );
 890                 }
 891         }
 892
 893         function writeRevision( $rev, $string ) {
 894                 for( $i = 0; $i < $this->count; $i++ ) {
 895                         $this->sinks[$i]->writeRevision( $rev, $string );
 896                 }
 897         }
 898 }
 899
 900 function xmlsafe( $string ) {
 901         $fname = 'xmlsafe';
 902         wfProfileIn( $fname );
 903
 904         /**
 905          * The page may contain old data which has not been properly normalized.
 906          * Invalid UTF-8 sequences or forbidden control characters will make our
 907          * XML output invalid, so be sure to strip them out.
 908          */
 909         $string = UtfNormal::cleanUp( $string );
 910
 911         $string = htmlspecialchars( $string );
 912         wfProfileOut( $fname );
 913         return $string;
 914 }