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 $multiPart = false;
  39         var $files = array();
  40
  41         const FULL = 1;
  42         const CURRENT = 2;
  43         const STABLE = 4; // extension defined
  44         const LOGS = 8;
  45
  46         const BUFFER = 0;
  47         const STREAM = 1;
  48
  49         const TEXT = 0;
  50         const STUB = 1;
  51
  52         /**
  53          * If using WikiExporter::STREAM to stream a large amount of data,
  54          * provide a database connection which is not managed by
  55          * LoadBalancer to read from: some history blob types will
  56          * make additional queries to pull source data while the
  57          * main query is still running.
  58          *
  59          * @param $db Database
  60          * @param $history Mixed: one of WikiExporter::FULL or WikiExporter::CURRENT,
  61          *                 or an associative array:
  62          *                   offset: non-inclusive offset at which to start the query
  63          *                   limit: maximum number of rows to return
  64          *                   dir: "asc" or "desc" timestamp order
  65          * @param $buffer Int: one of WikiExporter::BUFFER or WikiExporter::STREAM
  66          * @param $text Int: one of WikiExporter::TEXT or WikiExporter::STUB
  67          */
  68         function __construct( &$db, $history = WikiExporter::CURRENT,
  69                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  70                 $this->db =& $db;
  71                 $this->history = $history;
  72                 $this->buffer  = $buffer;
  73                 $this->writer  = new XmlDumpWriter();
  74                 $this->sink    = new DumpOutput();
  75                 $this->text    = $text;
  76         }
  77
  78         /**
  79          * Set the DumpOutput or DumpFilter object which will receive
  80          * various row objects and XML output for filtering. Filters
  81          * can be chained or used as callbacks.
  82          *
  83          * @param $sink mixed
  84          */
  85         public function setOutputSink( &$sink ) {
  86                 $this->sink =& $sink;
  87         }
  88
  89         public function openStream() {
  90                 $output = $this->writer->openStream();
  91
  92                 if ( $this->multiPart ) {
  93                         $this->openMultipart();
  94                         $this->sink->write( $output );
  95                 } else {
  96                         $this->sink->writeOpenStream( $output );
  97                 }
  98         }
  99
 100         public function closeStream() {
 101                 $output = $this->writer->closeStream();
 102                 if ( $this->multiPart ) {
 103                         $this->sink->write( $output );
 104                         $this->closeMultipart();
 105                 } else {
 106                         $this->sink->writeCloseStream( $output );
 107                 }
 108         }
 109
 110         /**
 111          * Dumps a series of page and revision records for all pages
 112          * in the database, either including complete history or only
 113          * the most recent version.
 114          */
 115         public function allPages() {
 116                 return $this->dumpFrom( '' );
 117         }
 118
 119         /**
 120          * Dumps a series of page and revision records for those pages
 121          * in the database falling within the page_id range given.
 122          * @param $start Int: inclusive lower limit (this id is included)
 123          * @param $end   Int: Exclusive upper limit (this id is not included)
 124          *                   If 0, no upper limit.
 125          */
 126         public function pagesByRange( $start, $end ) {
 127                 $condition = 'page_id >= ' . intval( $start );
 128                 if ( $end ) {
 129                         $condition .= ' AND page_id < ' . intval( $end );
 130                 }
 131                 return $this->dumpFrom( $condition );
 132         }
 133
 134         /**
 135          * @param $title Title
 136          */
 137         public function pageByTitle( $title ) {
 138                 return $this->dumpFrom(
 139                         'page_namespace=' . $title->getNamespace() .
 140                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 141         }
 142
 143         public function pageByName( $name ) {
 144                 $title = Title::newFromText( $name );
 145                 if ( is_null( $title ) ) {
 146                         throw new MWException( "Can't export invalid title" );
 147                 } else {
 148                         return $this->pageByTitle( $title );
 149                 }
 150         }
 151
 152         public function pagesByName( $names ) {
 153                 foreach ( $names as $name ) {
 154                         $this->pageByName( $name );
 155                 }
 156         }
 157
 158         public function allLogs() {
 159                 return $this->dumpFrom( '' );
 160         }
 161
 162         public function logsByRange( $start, $end ) {
 163                 $condition = 'log_id >= ' . intval( $start );
 164                 if ( $end ) {
 165                         $condition .= ' AND log_id < ' . intval( $end );
 166                 }
 167                 return $this->dumpFrom( $condition );
 168         }
 169
 170         # Generates the distinct list of authors of an article
 171         # Not called by default (depends on $this->list_authors)
 172         # Can be set by Special:Export when not exporting whole history
 173         protected function do_list_authors( $page , $revision , $cond ) {
 174                 wfProfileIn( __METHOD__ );
 175                 $this->author_list = "<contributors>";
 176                 // rev_deleted
 177                 $nothidden = '(' . $this->db->bitAnd( 'rev_deleted', Revision::DELETED_USER ) . ') = 0';
 178
 179                 $sql = "SELECT DISTINCT rev_user_text,rev_user FROM {$page},{$revision}
 180                 WHERE page_id=rev_page AND $nothidden AND " . $cond ;
 181                 $result = $this->db->query( $sql, __METHOD__ );
 182                 $resultset = $this->db->resultObject( $result );
 183                 foreach ( $resultset as $row ) {
 184                         $this->author_list .= "<contributor>" .
 185                                 "<username>" .
 186                                 htmlentities( $row->rev_user_text )  .
 187                                 "</username>" .
 188                                 "<id>" .
 189                                 $row->rev_user .
 190                                 "</id>" .
 191                                 "</contributor>";
 192                 }
 193                 $this->author_list .= "</contributors>";
 194                 wfProfileOut( __METHOD__ );
 195         }
 196
 197         protected function dumpFrom( $cond = '' ) {
 198                 wfProfileIn( __METHOD__ );
 199                 # For logging dumps...
 200                 if ( $this->history & self::LOGS ) {
 201                         if ( $this->buffer == WikiExporter::STREAM ) {
 202                                 $prev = $this->db->bufferResults( false );
 203                         }
 204                         $where = array( 'user_id = log_user' );
 205                         # Hide private logs
 206                         $hideLogs = LogEventsList::getExcludeClause( $this->db );
 207                         if ( $hideLogs ) $where[] = $hideLogs;
 208                         # Add on any caller specified conditions
 209                         if ( $cond ) $where[] = $cond;
 210                         # Get logging table name for logging.* clause
 211                         $logging = $this->db->tableName( 'logging' );
 212                         $result = $this->db->select( array( 'logging', 'user' ),
 213                                 array( "{$logging}.*", 'user_name' ), // grab the user name
 214                                 $where,
 215                                 __METHOD__,
 216                                 array( 'ORDER BY' => 'log_id', 'USE INDEX' => array( 'logging' => 'PRIMARY' ) )
 217                         );
 218                         $wrapper = $this->db->resultObject( $result );
 219                         $this->outputLogStream( $wrapper );
 220                         if ( $this->buffer == WikiExporter::STREAM ) {
 221                                 $this->db->bufferResults( $prev );
 222                         }
 223                 # For page dumps...
 224                 } else {
 225                         $tables = array( 'page', 'revision' );
 226                         $opts = array( 'ORDER BY' => 'page_id ASC' );
 227                         $opts['USE INDEX'] = array();
 228                         $join = array();
 229                         if ( is_array( $this->history ) ) {
 230                                 # Time offset/limit for all pages/history...
 231                                 $revJoin = 'page_id=rev_page';
 232                                 # Set time order
 233                                 if ( $this->history['dir'] == 'asc' ) {
 234                                         $op = '>';
 235                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 236                                 } else {
 237                                         $op = '<';
 238                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 239                                 }
 240                                 # Set offset
 241                                 if ( !empty( $this->history['offset'] ) ) {
 242                                         $revJoin .= " AND rev_timestamp $op " .
 243                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 244                                 }
 245                                 $join['revision'] = array( 'INNER JOIN', $revJoin );
 246                                 # Set query limit
 247                                 if ( !empty( $this->history['limit'] ) ) {
 248                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 249                                 }
 250                         } elseif ( $this->history & WikiExporter::FULL ) {
 251                                 # Full history dumps...
 252                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 253                         } elseif ( $this->history & WikiExporter::CURRENT ) {
 254                                 # Latest revision dumps...
 255                                 if ( $this->list_authors && $cond != '' )  { // List authors, if so desired
 256                                         list( $page, $revision ) = $this->db->tableNamesN( 'page', 'revision' );
 257                                         $this->do_list_authors( $page, $revision, $cond );
 258                                 }
 259                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 260                         } elseif ( $this->history & WikiExporter::STABLE ) {
 261                                 # "Stable" revision dumps...
 262                                 # Default JOIN, to be overridden...
 263                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 264                                 # One, and only one hook should set this, and return false
 265                                 if ( wfRunHooks( 'WikiExporter::dumpStableQuery', array( &$tables, &$opts, &$join ) ) ) {
 266                                         wfProfileOut( __METHOD__ );
 267                                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 268                                 }
 269                         } else {
 270                                 # Uknown history specification parameter?
 271                                 wfProfileOut( __METHOD__ );
 272                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 273                         }
 274                         # Query optimization hacks
 275                         if ( $cond == '' ) {
 276                                 $opts[] = 'STRAIGHT_JOIN';
 277                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 278                         }
 279                         # Build text join options
 280                         if ( $this->text != WikiExporter::STUB ) { // 1-pass
 281                                 $tables[] = 'text';
 282                                 $join['text'] = array( 'INNER JOIN', 'rev_text_id=old_id' );
 283                         }
 284
 285                         if ( $this->buffer == WikiExporter::STREAM ) {
 286                                 $prev = $this->db->bufferResults( false );
 287                         }
 288
 289                         wfRunHooks( 'ModifyExportQuery',
 290                                                 array( $this->db, &$tables, &$cond, &$opts, &$join ) );
 291
 292                         # Do the query!
 293                         $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 294                         $wrapper = $this->db->resultObject( $result );
 295                         # Output dump results
 296                         $this->outputPageStream( $wrapper );
 297                         if ( $this->list_authors ) {
 298                                 $this->outputPageStream( $wrapper );
 299                         }
 300
 301                         if ( $this->buffer == WikiExporter::STREAM ) {
 302                                 $this->db->bufferResults( $prev );
 303                         }
 304                 }
 305                 wfProfileOut( __METHOD__ );
 306         }
 307
 308         /**
 309          * Runs through a query result set dumping page and revision records.
 310          * The result set should be sorted/grouped by page to avoid duplicate
 311          * page records in the output.
 312          *
 313          * The result set will be freed once complete. Should be safe for
 314          * streaming (non-buffered) queries, as long as it was made on a
 315          * separate database connection not managed by LoadBalancer; some
 316          * blob storage types will make queries to pull source data.
 317          *
 318          * @param $resultset ResultWrapper
 319          */
 320         protected function outputPageStream( $resultset ) {
 321                 $last = null;
 322                 foreach ( $resultset as $row ) {
 323                         if ( is_null( $last ) ||
 324                                 $last->page_namespace != $row->page_namespace ||
 325                                 $last->page_title     != $row->page_title ) {
 326                                 if ( isset( $last ) ) {
 327                                         $output = '';
 328                                         if ( $this->dumpUploads ) {
 329                                                 $output .= $this->writer->writeUploads( $last );
 330                                                 $this->attachUploads( $last );
 331                                         }
 332                                         $output .= $this->writer->closePage();
 333                                         $this->sink->writeClosePage( $output );
 334                                 }
 335                                 $output = $this->writer->openPage( $row );
 336                                 $this->sink->writeOpenPage( $row, $output );
 337                                 $last = $row;
 338                         }
 339                         $output = $this->writer->writeRevision( $row );
 340                         $this->sink->writeRevision( $row, $output );
 341                 }
 342                 if ( isset( $last ) ) {
 343                         $output = '';
 344                         if ( $this->dumpUploads ) {
 345                                 $output .= $this->writer->writeUploads( $last );
 346                                 $this->attachUploads( $last );
 347                         }
 348                         $output .= $this->author_list;
 349                         $output .= $this->writer->closePage();
 350                         $this->sink->writeClosePage( $output );
 351                 }
 352         }
 353
 354         protected function outputLogStream( $resultset ) {
 355                 foreach ( $resultset as $row ) {
 356                         $output = $this->writer->writeLogItem( $row );
 357                         $this->sink->writeLogItem( $row, $output );
 358                 }
 359         }
 360
 361         protected function attachUploads( $row ) {
 362                 $title = Title::newFromRow( $row );
 363                 $file = wfLocalFile( $title );
 364                 $this->files[] = $file;
 365                 $this->files = array_merge( $this->files, $file->getHistory() );
 366         }
 367
 368         protected function openMultipart() {
 369                 # Multipart boundary purposely invalid XML
 370                 $this->boundary = '<' . dechex( mt_rand() ) . dechex( mt_rand() ) . '<';
 371                 $this->sink->writeOpenStream(
 372                         "Content-Type: multipart/related; boundary={$this->boundary};" .
 373                                 " type=text/xml\n\n" .
 374                         "--{$this->boundary}\nContent-Type: text/xml\n\n"
 375                 );
 376         }
 377
 378         protected function closeMultipart() {
 379                 $output = '';
 380
 381                 foreach ( $this->files as $file ) {
 382                         $output .= "\n--{$this->boundary}\n" .
 383                                 'Content-Type: ' . $file->getMimeType() . "\n" .
 384                                 'Content-ID: ' . $file->getRel() . "\n" .
 385                                 'Content-Length: ' . $file->getSize() . "\n" .
 386                                 'X-Sha1Base36: ' . $file->getSha1() . "\n\n";
 387                         $this->sink->write( $output );
 388                         $this->sink->write( file_get_contents( $file->getPath() ) );
 389                 }
 390                 $this->sink->writeCloseStream( "\n--{$this->boundary}\n" );
 391         }
 392 }
 393
 394 /**
 395  * @ingroup Dump
 396  */
 397 class XmlDumpWriter {
 398
 399         /**
 400          * Returns the export schema version.
 401          * @return string
 402          */
 403         function schemaVersion() {
 404                 return "0.5";
 405         }
 406
 407         /**
 408          * Opens the XML output stream's root <mediawiki> element.
 409          * This does not include an xml directive, so is safe to include
 410          * as a subelement in a larger XML stream. Namespace and XML Schema
 411          * references are included.
 412          *
 413          * Output will be encoded in UTF-8.
 414          *
 415          * @return string
 416          */
 417         function openStream() {
 418                 global $wgLanguageCode;
 419                 $ver = $this->schemaVersion();
 420                 return Xml::element( 'mediawiki', array(
 421                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 422                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 423                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 424                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 425                         'version'            => $ver,
 426                         'xml:lang'           => $wgLanguageCode ),
 427                         null ) .
 428                         "\n" .
 429                         $this->siteInfo();
 430         }
 431
 432         function siteInfo() {
 433                 $info = array(
 434                         $this->sitename(),
 435                         $this->homelink(),
 436                         $this->generator(),
 437                         $this->caseSetting(),
 438                         $this->namespaces() );
 439                 return "  <siteinfo>\n    " .
 440                         implode( "\n    ", $info ) .
 441                         "\n  </siteinfo>\n";
 442         }
 443
 444         function sitename() {
 445                 global $wgSitename;
 446                 return Xml::element( 'sitename', array(), $wgSitename );
 447         }
 448
 449         function generator() {
 450                 global $wgVersion;
 451                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 452         }
 453
 454         function homelink() {
 455                 return Xml::element( 'base', array(), Title::newMainPage()->getFullUrl() );
 456         }
 457
 458         function caseSetting() {
 459                 global $wgCapitalLinks;
 460                 // "case-insensitive" option is reserved for future
 461                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 462                 return Xml::element( 'case', array(), $sensitivity );
 463         }
 464
 465         function namespaces() {
 466                 global $wgContLang;
 467                 $spaces = "<namespaces>\n";
 468                 foreach ( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 469                         $spaces .= '      ' .
 470                                 Xml::element( 'namespace',
 471                                         array(  'key' => $ns,
 472                                                         'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 473                                         ), $title ) . "\n";
 474                 }
 475                 $spaces .= "    </namespaces>";
 476                 return $spaces;
 477         }
 478
 479         /**
 480          * Closes the output stream with the closing root element.
 481          * Call when finished dumping things.
 482          */
 483         function closeStream() {
 484                 return "</mediawiki>\n";
 485         }
 486
 487
 488         /**
 489          * Opens a <page> section on the output stream, with data
 490          * from the given database row.
 491          *
 492          * @param $row object
 493          * @return string
 494          * @access private
 495          */
 496         function openPage( $row ) {
 497                 $out = "  <page>\n";
 498                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 499                 $out .= '    ' . Xml::elementClean( 'title', array(), $title->getPrefixedText() ) . "\n";
 500                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 501                 if ( $row->page_is_redirect ) {
 502                         $out .= '    ' . Xml::element( 'redirect', array() ) . "\n";
 503                 }
 504                 if ( $row->page_restrictions != '' ) {
 505                         $out .= '    ' . Xml::element( 'restrictions', array(),
 506                                 strval( $row->page_restrictions ) ) . "\n";
 507                 }
 508
 509                 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 510
 511                 return $out;
 512         }
 513
 514         /**
 515          * Closes a <page> section on the output stream.
 516          *
 517          * @access private
 518          */
 519         function closePage() {
 520                 return "  </page>\n";
 521         }
 522
 523         /**
 524          * Dumps a <revision> section on the output stream, with
 525          * data filled in from the given database row.
 526          *
 527          * @param $row object
 528          * @return string
 529          * @access private
 530          */
 531         function writeRevision( $row ) {
 532                 wfProfileIn( __METHOD__ );
 533
 534                 $out  = "    <revision>\n";
 535                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 536
 537                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 538
 539                 if ( $row->rev_deleted & Revision::DELETED_USER ) {
 540                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 541                 } else {
 542                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 543                 }
 544
 545                 if ( $row->rev_minor_edit ) {
 546                         $out .=  "      <minor/>\n";
 547                 }
 548                 if ( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 549                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 550                 } elseif ( $row->rev_comment != '' ) {
 551                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 552                 }
 553
 554                 $text = '';
 555                 if ( $row->rev_deleted & Revision::DELETED_TEXT ) {
 556                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 557                 } elseif ( isset( $row->old_text ) ) {
 558                         // Raw text from the database may have invalid chars
 559                         $text = strval( Revision::getRevisionText( $row ) );
 560                         $out .= "      " . Xml::elementClean( 'text',
 561                                 array( 'xml:space' => 'preserve', 'bytes' => $row->rev_len ),
 562                                 strval( $text ) ) . "\n";
 563                 } else {
 564                         // Stub output
 565                         $out .= "      " . Xml::element( 'text',
 566                                 array( 'id' => $row->rev_text_id, 'bytes' => $row->rev_len ),
 567                                 "" ) . "\n";
 568                 }
 569
 570                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 571
 572                 $out .= "    </revision>\n";
 573
 574                 wfProfileOut( __METHOD__ );
 575                 return $out;
 576         }
 577
 578         /**
 579          * Dumps a <logitem> section on the output stream, with
 580          * data filled in from the given database row.
 581          *
 582          * @param $row object
 583          * @return string
 584          * @access private
 585          */
 586         function writeLogItem( $row ) {
 587                 wfProfileIn( __METHOD__ );
 588
 589                 $out  = "    <logitem>\n";
 590                 $out .= "      " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 591
 592                 $out .= $this->writeTimestamp( $row->log_timestamp );
 593
 594                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 595                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 596                 } else {
 597                         $out .= $this->writeContributor( $row->log_user, $row->user_name );
 598                 }
 599
 600                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 601                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 602                 } elseif ( $row->log_comment != '' ) {
 603                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 604                 }
 605
 606                 $out .= "      " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 607                 $out .= "      " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 608
 609                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 610                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 611                 } else {
 612                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 613                         $out .= "      " . Xml::elementClean( 'logtitle', null, $title->getPrefixedText() ) . "\n";
 614                         $out .= "      " . Xml::elementClean( 'params',
 615                                 array( 'xml:space' => 'preserve' ),
 616                                 strval( $row->log_params ) ) . "\n";
 617                 }
 618
 619                 $out .= "    </logitem>\n";
 620
 621                 wfProfileOut( __METHOD__ );
 622                 return $out;
 623         }
 624
 625         function writeTimestamp( $timestamp ) {
 626                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 627                 return "      " . Xml::element( 'timestamp', null, $ts ) . "\n";
 628         }
 629
 630         function writeContributor( $id, $text ) {
 631                 $out = "      <contributor>\n";
 632                 if ( $id ) {
 633                         $out .= "        " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 634                         $out .= "        " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 635                 } else {
 636                         $out .= "        " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 637                 }
 638                 $out .= "      </contributor>\n";
 639                 return $out;
 640         }
 641
 642         /**
 643          * Warning! This data is potentially inconsistent. :(
 644          */
 645         function writeUploads( $row ) {
 646                 if ( $row->page_namespace == NS_IMAGE ) {
 647                         $img = wfFindFile( $row->page_title );
 648                         if ( $img ) {
 649                                 $out = '';
 650                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 651                                         $out .= $this->writeUpload( $ver );
 652                                 }
 653                                 $out .= $this->writeUpload( $img );
 654                                 return $out;
 655                         }
 656                 }
 657                 return '';
 658         }
 659
 660         function writeUpload( $file ) {
 661                 return "    <upload>\n" .
 662                         $this->writeTimestamp( $file->getTimestamp() ) .
 663                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 664                         "      " . Xml::elementClean( 'comment', null, $file->getDescription() ) . "\n" .
 665                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 666                         "      " . Xml::element( 'src', null, $file->getFullUrl() ) . "\n" .
 667                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 668                         "    </upload>\n";
 669         }
 670
 671 }
 672
 673
 674 /**
 675  * Base class for output stream; prints to stdout or buffer or whereever.
 676  * @ingroup Dump
 677  */
 678 class DumpOutput {
 679         function writeOpenStream( $string ) {
 680                 $this->write( $string );
 681         }
 682
 683         function writeCloseStream( $string ) {
 684                 $this->write( $string );
 685         }
 686
 687         function writeOpenPage( $page, $string ) {
 688                 $this->write( $string );
 689         }
 690
 691         function writeClosePage( $string ) {
 692                 $this->write( $string );
 693         }
 694
 695         function writeRevision( $rev, $string ) {
 696                 $this->write( $string );
 697         }
 698
 699         function writeLogItem( $rev, $string ) {
 700                 $this->write( $string );
 701         }
 702
 703         /**
 704          * Override to write to a different stream type.
 705          * @return bool
 706          */
 707         function write( $string ) {
 708                 print $string;
 709         }
 710 }
 711
 712 /**
 713  * Stream outputter to send data to a file.
 714  * @ingroup Dump
 715  */
 716 class DumpFileOutput extends DumpOutput {
 717         var $handle;
 718
 719         function __construct( $file ) {
 720                 $this->handle = fopen( $file, "wt" );
 721         }
 722
 723         function write( $string ) {
 724                 fputs( $this->handle, $string );
 725         }
 726 }
 727
 728 /**
 729  * Stream outputter to send data to a file via some filter program.
 730  * Even if compression is available in a library, using a separate
 731  * program can allow us to make use of a multi-processor system.
 732  * @ingroup Dump
 733  */
 734 class DumpPipeOutput extends DumpFileOutput {
 735         function __construct( $command, $file = null ) {
 736                 if ( !is_null( $file ) ) {
 737                         $command .=  " > " . wfEscapeShellArg( $file );
 738                 }
 739                 $this->handle = popen( $command, "w" );
 740         }
 741 }
 742
 743 /**
 744  * Sends dump output via the gzip compressor.
 745  * @ingroup Dump
 746  */
 747 class DumpGZipOutput extends DumpPipeOutput {
 748         function __construct( $file ) {
 749                 parent::__construct( "gzip", $file );
 750         }
 751 }
 752
 753 /**
 754  * Sends dump output via the bgzip2 compressor.
 755  * @ingroup Dump
 756  */
 757 class DumpBZip2Output extends DumpPipeOutput {
 758         function __construct( $file ) {
 759                 parent::__construct( "bzip2", $file );
 760         }
 761 }
 762
 763 /**
 764  * Sends dump output via the p7zip compressor.
 765  * @ingroup Dump
 766  */
 767 class Dump7ZipOutput extends DumpPipeOutput {
 768         function __construct( $file ) {
 769                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 770                 // Suppress annoying useless crap from p7zip
 771                 // Unfortunately this could suppress real error messages too
 772                 $command .= ' >' . wfGetNull() . ' 2>&1';
 773                 parent::__construct( $command );
 774         }
 775 }
 776
 777
 778
 779 /**
 780  * Dump output filter class.
 781  * This just does output filtering and streaming; XML formatting is done
 782  * higher up, so be careful in what you do.
 783  * @ingroup Dump
 784  */
 785 class DumpFilter {
 786         function __construct( &$sink ) {
 787                 $this->sink =& $sink;
 788         }
 789
 790         function writeOpenStream( $string ) {
 791                 $this->sink->writeOpenStream( $string );
 792         }
 793
 794         function writeCloseStream( $string ) {
 795                 $this->sink->writeCloseStream( $string );
 796         }
 797
 798         function writeOpenPage( $page, $string ) {
 799                 $this->sendingThisPage = $this->pass( $page, $string );
 800                 if ( $this->sendingThisPage ) {
 801                         $this->sink->writeOpenPage( $page, $string );
 802                 }
 803         }
 804
 805         function writeClosePage( $string ) {
 806                 if ( $this->sendingThisPage ) {
 807                         $this->sink->writeClosePage( $string );
 808                         $this->sendingThisPage = false;
 809                 }
 810         }
 811
 812         function writeRevision( $rev, $string ) {
 813                 if ( $this->sendingThisPage ) {
 814                         $this->sink->writeRevision( $rev, $string );
 815                 }
 816         }
 817
 818         function writeLogItem( $rev, $string ) {
 819                 $this->sink->writeRevision( $rev, $string );
 820         }
 821
 822         /**
 823          * Override for page-based filter types.
 824          * @return bool
 825          */
 826         function pass( $page ) {
 827                 return true;
 828         }
 829 }
 830
 831 /**
 832  * Simple dump output filter to exclude all talk pages.
 833  * @ingroup Dump
 834  */
 835 class DumpNotalkFilter extends DumpFilter {
 836         function pass( $page ) {
 837                 return !MWNamespace::isTalk( $page->page_namespace );
 838         }
 839 }
 840
 841 /**
 842  * Dump output filter to include or exclude pages in a given set of namespaces.
 843  * @ingroup Dump
 844  */
 845 class DumpNamespaceFilter extends DumpFilter {
 846         var $invert = false;
 847         var $namespaces = array();
 848
 849         function __construct( &$sink, $param ) {
 850                 parent::__construct( $sink );
 851
 852                 $constants = array(
 853                         "NS_MAIN"           => NS_MAIN,
 854                         "NS_TALK"           => NS_TALK,
 855                         "NS_USER"           => NS_USER,
 856                         "NS_USER_TALK"      => NS_USER_TALK,
 857                         "NS_PROJECT"        => NS_PROJECT,
 858                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
 859                         "NS_FILE"           => NS_FILE,
 860                         "NS_FILE_TALK"      => NS_FILE_TALK,
 861                         "NS_IMAGE"          => NS_IMAGE,  // NS_IMAGE is an alias for NS_FILE
 862                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
 863                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
 864                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
 865                         "NS_TEMPLATE"       => NS_TEMPLATE,
 866                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
 867                         "NS_HELP"           => NS_HELP,
 868                         "NS_HELP_TALK"      => NS_HELP_TALK,
 869                         "NS_CATEGORY"       => NS_CATEGORY,
 870                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
 871
 872                 if ( $param { 0 } == '!' ) {
 873                         $this->invert = true;
 874                         $param = substr( $param, 1 );
 875                 }
 876
 877                 foreach ( explode( ',', $param ) as $key ) {
 878                         $key = trim( $key );
 879                         if ( isset( $constants[$key] ) ) {
 880                                 $ns = $constants[$key];
 881                                 $this->namespaces[$ns] = true;
 882                         } elseif ( is_numeric( $key ) ) {
 883                                 $ns = intval( $key );
 884                                 $this->namespaces[$ns] = true;
 885                         } else {
 886                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
 887                         }
 888                 }
 889         }
 890
 891         function pass( $page ) {
 892                 $match = isset( $this->namespaces[$page->page_namespace] );
 893                 return $this->invert xor $match;
 894         }
 895 }
 896
 897
 898 /**
 899  * Dump output filter to include only the last revision in each page sequence.
 900  * @ingroup Dump
 901  */
 902 class DumpLatestFilter extends DumpFilter {
 903         var $page, $pageString, $rev, $revString;
 904
 905         function writeOpenPage( $page, $string ) {
 906                 $this->page = $page;
 907                 $this->pageString = $string;
 908         }
 909
 910         function writeClosePage( $string ) {
 911                 if ( $this->rev ) {
 912                         $this->sink->writeOpenPage( $this->page, $this->pageString );
 913                         $this->sink->writeRevision( $this->rev, $this->revString );
 914                         $this->sink->writeClosePage( $string );
 915                 }
 916                 $this->rev = null;
 917                 $this->revString = null;
 918                 $this->page = null;
 919                 $this->pageString = null;
 920         }
 921
 922         function writeRevision( $rev, $string ) {
 923                 if ( $rev->rev_id == $this->page->page_latest ) {
 924                         $this->rev = $rev;
 925                         $this->revString = $string;
 926                 }
 927         }
 928 }
 929
 930 /**
 931  * Base class for output stream; prints to stdout or buffer or whereever.
 932  * @ingroup Dump
 933  */
 934 class DumpMultiWriter {
 935         function __construct( $sinks ) {
 936                 $this->sinks = $sinks;
 937                 $this->count = count( $sinks );
 938         }
 939
 940         function writeOpenStream( $string ) {
 941                 for ( $i = 0; $i < $this->count; $i++ ) {
 942                         $this->sinks[$i]->writeOpenStream( $string );
 943                 }
 944         }
 945
 946         function writeCloseStream( $string ) {
 947                 for ( $i = 0; $i < $this->count; $i++ ) {
 948                         $this->sinks[$i]->writeCloseStream( $string );
 949                 }
 950         }
 951
 952         function writeOpenPage( $page, $string ) {
 953                 for ( $i = 0; $i < $this->count; $i++ ) {
 954                         $this->sinks[$i]->writeOpenPage( $page, $string );
 955                 }
 956         }
 957
 958         function writeClosePage( $string ) {
 959                 for ( $i = 0; $i < $this->count; $i++ ) {
 960                         $this->sinks[$i]->writeClosePage( $string );
 961                 }
 962         }
 963
 964         function writeRevision( $rev, $string ) {
 965                 for ( $i = 0; $i < $this->count; $i++ ) {
 966                         $this->sinks[$i]->writeRevision( $rev, $string );
 967                 }
 968         }
 969 }
 970
 971 function xmlsafe( $string ) {
 972         wfProfileIn( __FUNCTION__ );
 973
 974         /**
 975          * The page may contain old data which has not been properly normalized.
 976          * Invalid UTF-8 sequences or forbidden control characters will make our
 977          * XML output invalid, so be sure to strip them out.
 978          */
 979         $string = UtfNormal::cleanUp( $string );
 980
 981         $string = htmlspecialchars( $string );
 982         wfProfileOut( __FUNCTION__ );
 983         return $string;
 984 }