includes/Export.php

   1 <?php
   2 /**
   3  * Base classes for dumps and export
   4  *
   5  * Copyright © 2003, 2005, 2006 Brion Vibber <brion@pobox.com>
   6  * http://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  */
  25
  26 /**
  27  * @defgroup Dump Dump
  28  */
  29
  30 /**
  31  * @ingroup SpecialPage Dump
  32  */
  33 class WikiExporter {
  34         var $list_authors = false ; # Return distinct author list (when not returning full history)
  35         var $author_list = "" ;
  36
  37         var $dumpUploads = false;
  38         var $dumpUploadFileContents = false;
  39
  40         const FULL = 1;
  41         const CURRENT = 2;
  42         const STABLE = 4; // extension defined
  43         const LOGS = 8;
  44
  45         const BUFFER = 0;
  46         const STREAM = 1;
  47
  48         const TEXT = 0;
  49         const STUB = 1;
  50
  51         /**
  52          * If using WikiExporter::STREAM to stream a large amount of data,
  53          * provide a database connection which is not managed by
  54          * LoadBalancer to read from: some history blob types will
  55          * make additional queries to pull source data while the
  56          * main query is still running.
  57          *
  58          * @param $db Database
  59          * @param $history Mixed: one of WikiExporter::FULL or WikiExporter::CURRENT,
  60          *                 or an associative array:
  61          *                   offset: non-inclusive offset at which to start the query
  62          *                   limit: maximum number of rows to return
  63          *                   dir: "asc" or "desc" timestamp order
  64          * @param $buffer Int: one of WikiExporter::BUFFER or WikiExporter::STREAM
  65          * @param $text Int: one of WikiExporter::TEXT or WikiExporter::STUB
  66          */
  67         function __construct( &$db, $history = WikiExporter::CURRENT,
  68                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  69                 $this->db =& $db;
  70                 $this->history = $history;
  71                 $this->buffer  = $buffer;
  72                 $this->writer  = new XmlDumpWriter();
  73                 $this->sink    = new DumpOutput();
  74                 $this->text    = $text;
  75         }
  76
  77         /**
  78          * Set the DumpOutput or DumpFilter object which will receive
  79          * various row objects and XML output for filtering. Filters
  80          * can be chained or used as callbacks.
  81          *
  82          * @param $sink mixed
  83          */
  84         public function setOutputSink( &$sink ) {
  85                 $this->sink =& $sink;
  86         }
  87
  88         public function openStream() {
  89                 $output = $this->writer->openStream();
  90                 $this->sink->writeOpenStream( $output );
  91         }
  92
  93         public function closeStream() {
  94                 $output = $this->writer->closeStream();
  95                 $this->sink->writeCloseStream( $output );
  96         }
  97
  98         /**
  99          * Dumps a series of page and revision records for all pages
 100          * in the database, either including complete history or only
 101          * the most recent version.
 102          */
 103         public function allPages() {
 104                 return $this->dumpFrom( '' );
 105         }
 106
 107         /**
 108          * Dumps a series of page and revision records for those pages
 109          * in the database falling within the page_id range given.
 110          * @param $start Int: inclusive lower limit (this id is included)
 111          * @param $end   Int: Exclusive upper limit (this id is not included)
 112          *                   If 0, no upper limit.
 113          */
 114         public function pagesByRange( $start, $end ) {
 115                 $condition = 'page_id >= ' . intval( $start );
 116                 if ( $end ) {
 117                         $condition .= ' AND page_id < ' . intval( $end );
 118                 }
 119                 return $this->dumpFrom( $condition );
 120         }
 121
 122         /**
 123          * @param $title Title
 124          */
 125         public function pageByTitle( $title ) {
 126                 return $this->dumpFrom(
 127                         'page_namespace=' . $title->getNamespace() .
 128                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 129         }
 130
 131         public function pageByName( $name ) {
 132                 $title = Title::newFromText( $name );
 133                 if ( is_null( $title ) ) {
 134                         throw new MWException( "Can't export invalid title" );
 135                 } else {
 136                         return $this->pageByTitle( $title );
 137                 }
 138         }
 139
 140         public function pagesByName( $names ) {
 141                 foreach ( $names as $name ) {
 142                         $this->pageByName( $name );
 143                 }
 144         }
 145
 146         public function allLogs() {
 147                 return $this->dumpFrom( '' );
 148         }
 149
 150         public function logsByRange( $start, $end ) {
 151                 $condition = 'log_id >= ' . intval( $start );
 152                 if ( $end ) {
 153                         $condition .= ' AND log_id < ' . intval( $end );
 154                 }
 155                 return $this->dumpFrom( $condition );
 156         }
 157
 158         # Generates the distinct list of authors of an article
 159         # Not called by default (depends on $this->list_authors)
 160         # Can be set by Special:Export when not exporting whole history
 161         protected function do_list_authors( $cond ) {
 162                 wfProfileIn( __METHOD__ );
 163                 $this->author_list = "<contributors>";
 164                 // rev_deleted
 165
 166                 $res = $this->db->select(
 167                         array( 'page', 'revision' ),
 168                         array( 'DISTINCT rev_user_text', 'rev_user' ),
 169                         array(
 170                                 $this->db->bitAnd( 'rev_deleted', Revision::DELETED_USER ) . ' = 0',
 171                                 $cond,
 172                                 'page_id = rev_id',
 173                         ),
 174                         __METHOD__
 175                 );
 176
 177                 foreach ( $res as $row ) {
 178                         $this->author_list .= "<contributor>" .
 179                                 "<username>" .
 180                                 htmlentities( $row->rev_user_text )  .
 181                                 "</username>" .
 182                                 "<id>" .
 183                                 $row->rev_user .
 184                                 "</id>" .
 185                                 "</contributor>";
 186                 }
 187                 $this->author_list .= "</contributors>";
 188                 wfProfileOut( __METHOD__ );
 189         }
 190
 191         protected function dumpFrom( $cond = '' ) {
 192                 wfProfileIn( __METHOD__ );
 193                 # For logging dumps...
 194                 if ( $this->history & self::LOGS ) {
 195                         if ( $this->buffer == WikiExporter::STREAM ) {
 196                                 $prev = $this->db->bufferResults( false );
 197                         }
 198                         $where = array( 'user_id = log_user' );
 199                         # Hide private logs
 200                         $hideLogs = LogEventsList::getExcludeClause( $this->db );
 201                         if ( $hideLogs ) $where[] = $hideLogs;
 202                         # Add on any caller specified conditions
 203                         if ( $cond ) $where[] = $cond;
 204                         # Get logging table name for logging.* clause
 205                         $logging = $this->db->tableName( 'logging' );
 206                         $result = $this->db->select( array( 'logging', 'user' ),
 207                                 array( "{$logging}.*", 'user_name' ), // grab the user name
 208                                 $where,
 209                                 __METHOD__,
 210                                 array( 'ORDER BY' => 'log_id', 'USE INDEX' => array( 'logging' => 'PRIMARY' ) )
 211                         );
 212                         $wrapper = $this->db->resultObject( $result );
 213                         $this->outputLogStream( $wrapper );
 214                         if ( $this->buffer == WikiExporter::STREAM ) {
 215                                 $this->db->bufferResults( $prev );
 216                         }
 217                 # For page dumps...
 218                 } else {
 219                         $tables = array( 'page', 'revision' );
 220                         $opts = array( 'ORDER BY' => 'page_id ASC' );
 221                         $opts['USE INDEX'] = array();
 222                         $join = array();
 223                         if ( is_array( $this->history ) ) {
 224                                 # Time offset/limit for all pages/history...
 225                                 $revJoin = 'page_id=rev_page';
 226                                 # Set time order
 227                                 if ( $this->history['dir'] == 'asc' ) {
 228                                         $op = '>';
 229                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 230                                 } else {
 231                                         $op = '<';
 232                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 233                                 }
 234                                 # Set offset
 235                                 if ( !empty( $this->history['offset'] ) ) {
 236                                         $revJoin .= " AND rev_timestamp $op " .
 237                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 238                                 }
 239                                 $join['revision'] = array( 'INNER JOIN', $revJoin );
 240                                 # Set query limit
 241                                 if ( !empty( $this->history['limit'] ) ) {
 242                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 243                                 }
 244                         } elseif ( $this->history & WikiExporter::FULL ) {
 245                                 # Full history dumps...
 246                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 247                         } elseif ( $this->history & WikiExporter::CURRENT ) {
 248                                 # Latest revision dumps...
 249                                 if ( $this->list_authors && $cond != '' )  { // List authors, if so desired
 250                                         $this->do_list_authors( $cond );
 251                                 }
 252                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 253                         } elseif ( $this->history & WikiExporter::STABLE ) {
 254                                 # "Stable" revision dumps...
 255                                 # Default JOIN, to be overridden...
 256                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 257                                 # One, and only one hook should set this, and return false
 258                                 if ( wfRunHooks( 'WikiExporter::dumpStableQuery', array( &$tables, &$opts, &$join ) ) ) {
 259                                         wfProfileOut( __METHOD__ );
 260                                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 261                                 }
 262                         } else {
 263                                 # Uknown history specification parameter?
 264                                 wfProfileOut( __METHOD__ );
 265                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 266                         }
 267                         # Query optimization hacks
 268                         if ( $cond == '' ) {
 269                                 $opts[] = 'STRAIGHT_JOIN';
 270                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 271                         }
 272                         # Build text join options
 273                         if ( $this->text != WikiExporter::STUB ) { // 1-pass
 274                                 $tables[] = 'text';
 275                                 $join['text'] = array( 'INNER JOIN', 'rev_text_id=old_id' );
 276                         }
 277
 278                         if ( $this->buffer == WikiExporter::STREAM ) {
 279                                 $prev = $this->db->bufferResults( false );
 280                         }
 281
 282                         wfRunHooks( 'ModifyExportQuery',
 283                                                 array( $this->db, &$tables, &$cond, &$opts, &$join ) );
 284
 285                         # Do the query!
 286                         $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 287                         $wrapper = $this->db->resultObject( $result );
 288                         # Output dump results
 289                         $this->outputPageStream( $wrapper );
 290                         if ( $this->list_authors ) {
 291                                 $this->outputPageStream( $wrapper );
 292                         }
 293
 294                         if ( $this->buffer == WikiExporter::STREAM ) {
 295                                 $this->db->bufferResults( $prev );
 296                         }
 297                 }
 298                 wfProfileOut( __METHOD__ );
 299         }
 300
 301         /**
 302          * Runs through a query result set dumping page and revision records.
 303          * The result set should be sorted/grouped by page to avoid duplicate
 304          * page records in the output.
 305          *
 306          * The result set will be freed once complete. Should be safe for
 307          * streaming (non-buffered) queries, as long as it was made on a
 308          * separate database connection not managed by LoadBalancer; some
 309          * blob storage types will make queries to pull source data.
 310          *
 311          * @param $resultset ResultWrapper
 312          */
 313         protected function outputPageStream( $resultset ) {
 314                 $last = null;
 315                 foreach ( $resultset as $row ) {
 316                         if ( is_null( $last ) ||
 317                                 $last->page_namespace != $row->page_namespace ||
 318                                 $last->page_title     != $row->page_title ) {
 319                                 if ( isset( $last ) ) {
 320                                         $output = '';
 321                                         if ( $this->dumpUploads ) {
 322                                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 323                                         }
 324                                         $output .= $this->writer->closePage();
 325                                         $this->sink->writeClosePage( $output );
 326                                 }
 327                                 $output = $this->writer->openPage( $row );
 328                                 $this->sink->writeOpenPage( $row, $output );
 329                                 $last = $row;
 330                         }
 331                         $output = $this->writer->writeRevision( $row );
 332                         $this->sink->writeRevision( $row, $output );
 333                 }
 334                 if ( isset( $last ) ) {
 335                         $output = '';
 336                         if ( $this->dumpUploads ) {
 337                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 338                         }
 339                         $output .= $this->author_list;
 340                         $output .= $this->writer->closePage();
 341                         $this->sink->writeClosePage( $output );
 342                 }
 343         }
 344
 345         protected function outputLogStream( $resultset ) {
 346                 foreach ( $resultset as $row ) {
 347                         $output = $this->writer->writeLogItem( $row );
 348                         $this->sink->writeLogItem( $row, $output );
 349                 }
 350         }
 351 }
 352
 353 /**
 354  * @ingroup Dump
 355  */
 356 class XmlDumpWriter {
 357
 358         /**
 359          * Returns the export schema version.
 360          * @return string
 361          */
 362         function schemaVersion() {
 363                 return "0.5";
 364         }
 365
 366         /**
 367          * Opens the XML output stream's root <mediawiki> element.
 368          * This does not include an xml directive, so is safe to include
 369          * as a subelement in a larger XML stream. Namespace and XML Schema
 370          * references are included.
 371          *
 372          * Output will be encoded in UTF-8.
 373          *
 374          * @return string
 375          */
 376         function openStream() {
 377                 global $wgLanguageCode;
 378                 $ver = $this->schemaVersion();
 379                 return Xml::element( 'mediawiki', array(
 380                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 381                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 382                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 383                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 384                         'version'            => $ver,
 385                         'xml:lang'           => $wgLanguageCode ),
 386                         null ) .
 387                         "\n" .
 388                         $this->siteInfo();
 389         }
 390
 391         function siteInfo() {
 392                 $info = array(
 393                         $this->sitename(),
 394                         $this->homelink(),
 395                         $this->generator(),
 396                         $this->caseSetting(),
 397                         $this->namespaces() );
 398                 return "  <siteinfo>\n    " .
 399                         implode( "\n    ", $info ) .
 400                         "\n  </siteinfo>\n";
 401         }
 402
 403         function sitename() {
 404                 global $wgSitename;
 405                 return Xml::element( 'sitename', array(), $wgSitename );
 406         }
 407
 408         function generator() {
 409                 global $wgVersion;
 410                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 411         }
 412
 413         function homelink() {
 414                 return Xml::element( 'base', array(), Title::newMainPage()->getFullUrl() );
 415         }
 416
 417         function caseSetting() {
 418                 global $wgCapitalLinks;
 419                 // "case-insensitive" option is reserved for future
 420                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 421                 return Xml::element( 'case', array(), $sensitivity );
 422         }
 423
 424         function namespaces() {
 425                 global $wgContLang;
 426                 $spaces = "<namespaces>\n";
 427                 foreach ( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 428                         $spaces .= '      ' .
 429                                 Xml::element( 'namespace',
 430                                         array(  'key' => $ns,
 431                                                         'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 432                                         ), $title ) . "\n";
 433                 }
 434                 $spaces .= "    </namespaces>";
 435                 return $spaces;
 436         }
 437
 438         /**
 439          * Closes the output stream with the closing root element.
 440          * Call when finished dumping things.
 441          *
 442          * @return string
 443          */
 444         function closeStream() {
 445                 return "</mediawiki>\n";
 446         }
 447
 448         /**
 449          * Opens a <page> section on the output stream, with data
 450          * from the given database row.
 451          *
 452          * @param $row object
 453          * @return string
 454          * @access private
 455          */
 456         function openPage( $row ) {
 457                 $out = "  <page>\n";
 458                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 459                 $out .= '    ' . Xml::elementClean( 'title', array(), $title->getPrefixedText() ) . "\n";
 460                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 461                 if ( $row->page_is_redirect ) {
 462                         $out .= '    ' . Xml::element( 'redirect', array() ) . "\n";
 463                 }
 464                 if ( $row->page_restrictions != '' ) {
 465                         $out .= '    ' . Xml::element( 'restrictions', array(),
 466                                 strval( $row->page_restrictions ) ) . "\n";
 467                 }
 468
 469                 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 470
 471                 return $out;
 472         }
 473
 474         /**
 475          * Closes a <page> section on the output stream.
 476          *
 477          * @access private
 478          */
 479         function closePage() {
 480                 return "  </page>\n";
 481         }
 482
 483         /**
 484          * Dumps a <revision> section on the output stream, with
 485          * data filled in from the given database row.
 486          *
 487          * @param $row object
 488          * @return string
 489          * @access private
 490          */
 491         function writeRevision( $row ) {
 492                 wfProfileIn( __METHOD__ );
 493
 494                 $out  = "    <revision>\n";
 495                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 496
 497                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 498
 499                 if ( $row->rev_deleted & Revision::DELETED_USER ) {
 500                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 501                 } else {
 502                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 503                 }
 504
 505                 if ( $row->rev_minor_edit ) {
 506                         $out .=  "      <minor/>\n";
 507                 }
 508                 if ( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 509                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 510                 } elseif ( $row->rev_comment != '' ) {
 511                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 512                 }
 513
 514                 $text = '';
 515                 if ( $row->rev_deleted & Revision::DELETED_TEXT ) {
 516                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 517                 } elseif ( isset( $row->old_text ) ) {
 518                         // Raw text from the database may have invalid chars
 519                         $text = strval( Revision::getRevisionText( $row ) );
 520                         $out .= "      " . Xml::elementClean( 'text',
 521                                 array( 'xml:space' => 'preserve', 'bytes' => $row->rev_len ),
 522                                 strval( $text ) ) . "\n";
 523                 } else {
 524                         // Stub output
 525                         $out .= "      " . Xml::element( 'text',
 526                                 array( 'id' => $row->rev_text_id, 'bytes' => $row->rev_len ),
 527                                 "" ) . "\n";
 528                 }
 529
 530                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 531
 532                 $out .= "    </revision>\n";
 533
 534                 wfProfileOut( __METHOD__ );
 535                 return $out;
 536         }
 537
 538         /**
 539          * Dumps a <logitem> section on the output stream, with
 540          * data filled in from the given database row.
 541          *
 542          * @param $row object
 543          * @return string
 544          * @access private
 545          */
 546         function writeLogItem( $row ) {
 547                 wfProfileIn( __METHOD__ );
 548
 549                 $out  = "    <logitem>\n";
 550                 $out .= "      " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 551
 552                 $out .= $this->writeTimestamp( $row->log_timestamp );
 553
 554                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 555                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 556                 } else {
 557                         $out .= $this->writeContributor( $row->log_user, $row->user_name );
 558                 }
 559
 560                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 561                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 562                 } elseif ( $row->log_comment != '' ) {
 563                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 564                 }
 565
 566                 $out .= "      " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 567                 $out .= "      " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 568
 569                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 570                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 571                 } else {
 572                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 573                         $out .= "      " . Xml::elementClean( 'logtitle', null, $title->getPrefixedText() ) . "\n";
 574                         $out .= "      " . Xml::elementClean( 'params',
 575                                 array( 'xml:space' => 'preserve' ),
 576                                 strval( $row->log_params ) ) . "\n";
 577                 }
 578
 579                 $out .= "    </logitem>\n";
 580
 581                 wfProfileOut( __METHOD__ );
 582                 return $out;
 583         }
 584
 585         function writeTimestamp( $timestamp ) {
 586                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 587                 return "      " . Xml::element( 'timestamp', null, $ts ) . "\n";
 588         }
 589
 590         function writeContributor( $id, $text ) {
 591                 $out = "      <contributor>\n";
 592                 if ( $id ) {
 593                         $out .= "        " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 594                         $out .= "        " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 595                 } else {
 596                         $out .= "        " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 597                 }
 598                 $out .= "      </contributor>\n";
 599                 return $out;
 600         }
 601
 602         /**
 603          * Warning! This data is potentially inconsistent. :(
 604          */
 605         function writeUploads( $row, $dumpContents = false ) {
 606                 if ( $row->page_namespace == NS_IMAGE ) {
 607                         $img = wfLocalFile( $row->page_title );
 608                         if ( $img && $img->exists() ) {
 609                                 $out = '';
 610                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 611                                         $out .= $this->writeUpload( $ver, $dumpContents );
 612                                 }
 613                                 $out .= $this->writeUpload( $img, $dumpContents );
 614                                 return $out;
 615                         }
 616                 }
 617                 return '';
 618         }
 619
 620         /**
 621          * @param $file File
 622          * @param $dumpContents bool
 623          * @return string
 624          */
 625         function writeUpload( $file, $dumpContents = false ) {
 626                 if ( $file->isOld() ) {
 627                         $archiveName = "      " .
 628                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 629                 } else {
 630                         $archiveName = '';
 631                 }
 632                 if ( $dumpContents ) {
 633                         # Dump file as base64
 634                         # Uses only XML-safe characters, so does not need escaping
 635                         $contents = '      <contents encoding="base64">' .
 636                                 chunk_split( base64_encode( file_get_contents( $file->getPath() ) ) ) .
 637                                 "      </contents>\n";
 638                 } else {
 639                         $contents = '';
 640                 }
 641                 return "    <upload>\n" .
 642                         $this->writeTimestamp( $file->getTimestamp() ) .
 643                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 644                         "      " . Xml::elementClean( 'comment', null, $file->getDescription() ) . "\n" .
 645                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 646                         $archiveName .
 647                         "      " . Xml::element( 'src', null, $file->getFullUrl() ) . "\n" .
 648                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 649                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 650                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 651                         $contents .
 652                         "    </upload>\n";
 653         }
 654
 655 }
 656
 657
 658 /**
 659  * Base class for output stream; prints to stdout or buffer or whereever.
 660  * @ingroup Dump
 661  */
 662 class DumpOutput {
 663         function writeOpenStream( $string ) {
 664                 $this->write( $string );
 665         }
 666
 667         function writeCloseStream( $string ) {
 668                 $this->write( $string );
 669         }
 670
 671         function writeOpenPage( $page, $string ) {
 672                 $this->write( $string );
 673         }
 674
 675         function writeClosePage( $string ) {
 676                 $this->write( $string );
 677         }
 678
 679         function writeRevision( $rev, $string ) {
 680                 $this->write( $string );
 681         }
 682
 683         function writeLogItem( $rev, $string ) {
 684                 $this->write( $string );
 685         }
 686
 687         /**
 688          * Override to write to a different stream type.
 689          * @return bool
 690          */
 691         function write( $string ) {
 692                 print $string;
 693         }
 694 }
 695
 696 /**
 697  * Stream outputter to send data to a file.
 698  * @ingroup Dump
 699  */
 700 class DumpFileOutput extends DumpOutput {
 701         var $handle;
 702
 703         function __construct( $file ) {
 704                 $this->handle = fopen( $file, "wt" );
 705         }
 706
 707         function write( $string ) {
 708                 fputs( $this->handle, $string );
 709         }
 710 }
 711
 712 /**
 713  * Stream outputter to send data to a file via some filter program.
 714  * Even if compression is available in a library, using a separate
 715  * program can allow us to make use of a multi-processor system.
 716  * @ingroup Dump
 717  */
 718 class DumpPipeOutput extends DumpFileOutput {
 719         function __construct( $command, $file = null ) {
 720                 if ( !is_null( $file ) ) {
 721                         $command .=  " > " . wfEscapeShellArg( $file );
 722                 }
 723                 $this->handle = popen( $command, "w" );
 724         }
 725 }
 726
 727 /**
 728  * Sends dump output via the gzip compressor.
 729  * @ingroup Dump
 730  */
 731 class DumpGZipOutput extends DumpPipeOutput {
 732         function __construct( $file ) {
 733                 parent::__construct( "gzip", $file );
 734         }
 735 }
 736
 737 /**
 738  * Sends dump output via the bgzip2 compressor.
 739  * @ingroup Dump
 740  */
 741 class DumpBZip2Output extends DumpPipeOutput {
 742         function __construct( $file ) {
 743                 parent::__construct( "bzip2", $file );
 744         }
 745 }
 746
 747 /**
 748  * Sends dump output via the p7zip compressor.
 749  * @ingroup Dump
 750  */
 751 class Dump7ZipOutput extends DumpPipeOutput {
 752         function __construct( $file ) {
 753                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 754                 // Suppress annoying useless crap from p7zip
 755                 // Unfortunately this could suppress real error messages too
 756                 $command .= ' >' . wfGetNull() . ' 2>&1';
 757                 parent::__construct( $command );
 758         }
 759 }
 760
 761
 762
 763 /**
 764  * Dump output filter class.
 765  * This just does output filtering and streaming; XML formatting is done
 766  * higher up, so be careful in what you do.
 767  * @ingroup Dump
 768  */
 769 class DumpFilter {
 770         function __construct( &$sink ) {
 771                 $this->sink =& $sink;
 772         }
 773
 774         function writeOpenStream( $string ) {
 775                 $this->sink->writeOpenStream( $string );
 776         }
 777
 778         function writeCloseStream( $string ) {
 779                 $this->sink->writeCloseStream( $string );
 780         }
 781
 782         function writeOpenPage( $page, $string ) {
 783                 $this->sendingThisPage = $this->pass( $page, $string );
 784                 if ( $this->sendingThisPage ) {
 785                         $this->sink->writeOpenPage( $page, $string );
 786                 }
 787         }
 788
 789         function writeClosePage( $string ) {
 790                 if ( $this->sendingThisPage ) {
 791                         $this->sink->writeClosePage( $string );
 792                         $this->sendingThisPage = false;
 793                 }
 794         }
 795
 796         function writeRevision( $rev, $string ) {
 797                 if ( $this->sendingThisPage ) {
 798                         $this->sink->writeRevision( $rev, $string );
 799                 }
 800         }
 801
 802         function writeLogItem( $rev, $string ) {
 803                 $this->sink->writeRevision( $rev, $string );
 804         }
 805
 806         /**
 807          * Override for page-based filter types.
 808          * @return bool
 809          */
 810         function pass( $page ) {
 811                 return true;
 812         }
 813 }
 814
 815 /**
 816  * Simple dump output filter to exclude all talk pages.
 817  * @ingroup Dump
 818  */
 819 class DumpNotalkFilter extends DumpFilter {
 820         function pass( $page ) {
 821                 return !MWNamespace::isTalk( $page->page_namespace );
 822         }
 823 }
 824
 825 /**
 826  * Dump output filter to include or exclude pages in a given set of namespaces.
 827  * @ingroup Dump
 828  */
 829 class DumpNamespaceFilter extends DumpFilter {
 830         var $invert = false;
 831         var $namespaces = array();
 832
 833         function __construct( &$sink, $param ) {
 834                 parent::__construct( $sink );
 835
 836                 $constants = array(
 837                         "NS_MAIN"           => NS_MAIN,
 838                         "NS_TALK"           => NS_TALK,
 839                         "NS_USER"           => NS_USER,
 840                         "NS_USER_TALK"      => NS_USER_TALK,
 841                         "NS_PROJECT"        => NS_PROJECT,
 842                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
 843                         "NS_FILE"           => NS_FILE,
 844                         "NS_FILE_TALK"      => NS_FILE_TALK,
 845                         "NS_IMAGE"          => NS_IMAGE,  // NS_IMAGE is an alias for NS_FILE
 846                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
 847                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
 848                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
 849                         "NS_TEMPLATE"       => NS_TEMPLATE,
 850                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
 851                         "NS_HELP"           => NS_HELP,
 852                         "NS_HELP_TALK"      => NS_HELP_TALK,
 853                         "NS_CATEGORY"       => NS_CATEGORY,
 854                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
 855
 856                 if ( $param { 0 } == '!' ) {
 857                         $this->invert = true;
 858                         $param = substr( $param, 1 );
 859                 }
 860
 861                 foreach ( explode( ',', $param ) as $key ) {
 862                         $key = trim( $key );
 863                         if ( isset( $constants[$key] ) ) {
 864                                 $ns = $constants[$key];
 865                                 $this->namespaces[$ns] = true;
 866                         } elseif ( is_numeric( $key ) ) {
 867                                 $ns = intval( $key );
 868                                 $this->namespaces[$ns] = true;
 869                         } else {
 870                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
 871                         }
 872                 }
 873         }
 874
 875         function pass( $page ) {
 876                 $match = isset( $this->namespaces[$page->page_namespace] );
 877                 return $this->invert xor $match;
 878         }
 879 }
 880
 881
 882 /**
 883  * Dump output filter to include only the last revision in each page sequence.
 884  * @ingroup Dump
 885  */
 886 class DumpLatestFilter extends DumpFilter {
 887         var $page, $pageString, $rev, $revString;
 888
 889         function writeOpenPage( $page, $string ) {
 890                 $this->page = $page;
 891                 $this->pageString = $string;
 892         }
 893
 894         function writeClosePage( $string ) {
 895                 if ( $this->rev ) {
 896                         $this->sink->writeOpenPage( $this->page, $this->pageString );
 897                         $this->sink->writeRevision( $this->rev, $this->revString );
 898                         $this->sink->writeClosePage( $string );
 899                 }
 900                 $this->rev = null;
 901                 $this->revString = null;
 902                 $this->page = null;
 903                 $this->pageString = null;
 904         }
 905
 906         function writeRevision( $rev, $string ) {
 907                 if ( $rev->rev_id == $this->page->page_latest ) {
 908                         $this->rev = $rev;
 909                         $this->revString = $string;
 910                 }
 911         }
 912 }
 913
 914 /**
 915  * Base class for output stream; prints to stdout or buffer or whereever.
 916  * @ingroup Dump
 917  */
 918 class DumpMultiWriter {
 919         function __construct( $sinks ) {
 920                 $this->sinks = $sinks;
 921                 $this->count = count( $sinks );
 922         }
 923
 924         function writeOpenStream( $string ) {
 925                 for ( $i = 0; $i < $this->count; $i++ ) {
 926                         $this->sinks[$i]->writeOpenStream( $string );
 927                 }
 928         }
 929
 930         function writeCloseStream( $string ) {
 931                 for ( $i = 0; $i < $this->count; $i++ ) {
 932                         $this->sinks[$i]->writeCloseStream( $string );
 933                 }
 934         }
 935
 936         function writeOpenPage( $page, $string ) {
 937                 for ( $i = 0; $i < $this->count; $i++ ) {
 938                         $this->sinks[$i]->writeOpenPage( $page, $string );
 939                 }
 940         }
 941
 942         function writeClosePage( $string ) {
 943                 for ( $i = 0; $i < $this->count; $i++ ) {
 944                         $this->sinks[$i]->writeClosePage( $string );
 945                 }
 946         }
 947
 948         function writeRevision( $rev, $string ) {
 949                 for ( $i = 0; $i < $this->count; $i++ ) {
 950                         $this->sinks[$i]->writeRevision( $rev, $string );
 951                 }
 952         }
 953 }
 954
 955 function xmlsafe( $string ) {
 956         wfProfileIn( __FUNCTION__ );
 957
 958         /**
 959          * The page may contain old data which has not been properly normalized.
 960          * Invalid UTF-8 sequences or forbidden control characters will make our
 961          * XML output invalid, so be sure to strip them out.
 962          */
 963         $string = UtfNormal::cleanUp( $string );
 964
 965         $string = htmlspecialchars( $string );
 966         wfProfileOut( __FUNCTION__ );
 967         return $string;
 968 }