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