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         const RANGE = 16;
  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 DatabaseBase
  60          * @param $history Mixed: one of WikiExporter::FULL, WikiExporter::CURRENT,
  61          *                 WikiExporter::RANGE or WikiExporter::STABLE,
  62          *                 or an associative array:
  63          *                   offset: non-inclusive offset at which to start the query
  64          *                   limit: maximum number of rows to return
  65          *                   dir: "asc" or "desc" timestamp order
  66          * @param $buffer Int: one of WikiExporter::BUFFER or WikiExporter::STREAM
  67          * @param $text Int: one of WikiExporter::TEXT or WikiExporter::STUB
  68          */
  69         function __construct( &$db, $history = WikiExporter::CURRENT,
  70                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  71                 $this->db =& $db;
  72                 $this->history = $history;
  73                 $this->buffer  = $buffer;
  74                 $this->writer  = new XmlDumpWriter();
  75                 $this->sink    = new DumpOutput();
  76                 $this->text    = $text;
  77         }
  78
  79         /**
  80          * Set the DumpOutput or DumpFilter object which will receive
  81          * various row objects and XML output for filtering. Filters
  82          * can be chained or used as callbacks.
  83          *
  84          * @param $sink mixed
  85          */
  86         public function setOutputSink( &$sink ) {
  87                 $this->sink =& $sink;
  88         }
  89
  90         public function openStream() {
  91                 $output = $this->writer->openStream();
  92                 $this->sink->writeOpenStream( $output );
  93         }
  94
  95         public function closeStream() {
  96                 $output = $this->writer->closeStream();
  97                 $this->sink->writeCloseStream( $output );
  98         }
  99
 100         /**
 101          * Dumps a series of page and revision records for all pages
 102          * in the database, either including complete history or only
 103          * the most recent version.
 104          */
 105         public function allPages() {
 106                 $this->dumpFrom( '' );
 107         }
 108
 109         /**
 110          * Dumps a series of page and revision records for those pages
 111          * in the database falling within the page_id range given.
 112          * @param $start Int: inclusive lower limit (this id is included)
 113          * @param $end   Int: Exclusive upper limit (this id is not included)
 114          *                   If 0, no upper limit.
 115          */
 116         public function pagesByRange( $start, $end ) {
 117                 $condition = 'page_id >= ' . intval( $start );
 118                 if ( $end ) {
 119                         $condition .= ' AND page_id < ' . intval( $end );
 120                 }
 121                 $this->dumpFrom( $condition );
 122         }
 123
 124         /**
 125          * Dumps a series of page and revision records for those pages
 126          * in the database with revisions falling within the rev_id range given.
 127          * @param $start Int: inclusive lower limit (this id is included)
 128          * @param $end   Int: Exclusive upper limit (this id is not included)
 129          *                   If 0, no upper limit.
 130          */
 131         public function revsByRange( $start, $end ) {
 132                 $condition = 'rev_id >= ' . intval( $start );
 133                 if ( $end ) {
 134                         $condition .= ' AND rev_id < ' . intval( $end );
 135                 }
 136                 $this->dumpFrom( $condition );
 137         }
 138
 139         /**
 140          * @param $title Title
 141          */
 142         public function pageByTitle( $title ) {
 143                 $this->dumpFrom(
 144                         'page_namespace=' . $title->getNamespace() .
 145                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 146         }
 147
 148         public function pageByName( $name ) {
 149                 $title = Title::newFromText( $name );
 150                 if ( is_null( $title ) ) {
 151                         throw new MWException( "Can't export invalid title" );
 152                 } else {
 153                         $this->pageByTitle( $title );
 154                 }
 155         }
 156
 157         public function pagesByName( $names ) {
 158                 foreach ( $names as $name ) {
 159                         $this->pageByName( $name );
 160                 }
 161         }
 162
 163         public function allLogs() {
 164                 $this->dumpFrom( '' );
 165         }
 166
 167         public function logsByRange( $start, $end ) {
 168                 $condition = 'log_id >= ' . intval( $start );
 169                 if ( $end ) {
 170                         $condition .= ' AND log_id < ' . intval( $end );
 171                 }
 172                 $this->dumpFrom( $condition );
 173         }
 174
 175         # Generates the distinct list of authors of an article
 176         # Not called by default (depends on $this->list_authors)
 177         # Can be set by Special:Export when not exporting whole history
 178         protected function do_list_authors( $cond ) {
 179                 wfProfileIn( __METHOD__ );
 180                 $this->author_list = "<contributors>";
 181                 // rev_deleted
 182
 183                 $res = $this->db->select(
 184                         array( 'page', 'revision' ),
 185                         array( 'DISTINCT rev_user_text', 'rev_user' ),
 186                         array(
 187                                 $this->db->bitAnd( 'rev_deleted', Revision::DELETED_USER ) . ' = 0',
 188                                 $cond,
 189                                 'page_id = rev_id',
 190                         ),
 191                         __METHOD__
 192                 );
 193
 194                 foreach ( $res as $row ) {
 195                         $this->author_list .= "<contributor>" .
 196                                 "<username>" .
 197                                 htmlentities( $row->rev_user_text )  .
 198                                 "</username>" .
 199                                 "<id>" .
 200                                 $row->rev_user .
 201                                 "</id>" .
 202                                 "</contributor>";
 203                 }
 204                 $this->author_list .= "</contributors>";
 205                 wfProfileOut( __METHOD__ );
 206         }
 207
 208         protected function dumpFrom( $cond = '' ) {
 209                 wfProfileIn( __METHOD__ );
 210                 # For logging dumps...
 211                 if ( $this->history & self::LOGS ) {
 212                         $where = array( 'user_id = log_user' );
 213                         # Hide private logs
 214                         $hideLogs = LogEventsList::getExcludeClause( $this->db );
 215                         if ( $hideLogs ) $where[] = $hideLogs;
 216                         # Add on any caller specified conditions
 217                         if ( $cond ) $where[] = $cond;
 218                         # Get logging table name for logging.* clause
 219                         $logging = $this->db->tableName( 'logging' );
 220
 221                         if ( $this->buffer == WikiExporter::STREAM ) {
 222                                 $prev = $this->db->bufferResults( false );
 223                         }
 224                         $wrapper = null; // Assuring $wrapper is not undefined, if exception occurs early
 225                         try {
 226                                 $result = $this->db->select( array( 'logging', 'user' ),
 227                                         array( "{$logging}.*", 'user_name' ), // grab the user name
 228                                         $where,
 229                                         __METHOD__,
 230                                         array( 'ORDER BY' => 'log_id', 'USE INDEX' => array( 'logging' => 'PRIMARY' ) )
 231                                 );
 232                                 $wrapper = $this->db->resultObject( $result );
 233                                 $this->outputLogStream( $wrapper );
 234                                 if ( $this->buffer == WikiExporter::STREAM ) {
 235                                         $this->db->bufferResults( $prev );
 236                                 }
 237                         } catch ( Exception $e ) {
 238                                 // Throwing the exception does not reliably free the resultset, and
 239                                 // would also leave the connection in unbuffered mode.
 240
 241                                 // Freeing result
 242                                 try {
 243                                         if ( $wrapper ) {
 244                                                 $wrapper->free();
 245                                         }
 246                                 } catch ( Exception $e2 ) {
 247                                         // Already in panic mode -> ignoring $e2 as $e has
 248                                         // higher priority
 249                                 }
 250
 251                                 // Putting database back in previous buffer mode
 252                                 try {
 253                                         if ( $this->buffer == WikiExporter::STREAM ) {
 254                                                 $this->db->bufferResults( $prev );
 255                                         }
 256                                 } catch ( Exception $e2 ) {
 257                                         // Already in panic mode -> ignoring $e2 as $e has
 258                                         // higher priority
 259                                 }
 260
 261                                 // Inform caller about problem
 262                                 throw $e;
 263                         }
 264                 # For page dumps...
 265                 } else {
 266                         $tables = array( 'page', 'revision' );
 267                         $opts = array( 'ORDER BY' => 'page_id ASC' );
 268                         $opts['USE INDEX'] = array();
 269                         $join = array();
 270                         if ( is_array( $this->history ) ) {
 271                                 # Time offset/limit for all pages/history...
 272                                 $revJoin = 'page_id=rev_page';
 273                                 # Set time order
 274                                 if ( $this->history['dir'] == 'asc' ) {
 275                                         $op = '>';
 276                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 277                                 } else {
 278                                         $op = '<';
 279                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 280                                 }
 281                                 # Set offset
 282                                 if ( !empty( $this->history['offset'] ) ) {
 283                                         $revJoin .= " AND rev_timestamp $op " .
 284                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 285                                 }
 286                                 $join['revision'] = array( 'INNER JOIN', $revJoin );
 287                                 # Set query limit
 288                                 if ( !empty( $this->history['limit'] ) ) {
 289                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 290                                 }
 291                         } elseif ( $this->history & WikiExporter::FULL ) {
 292                                 # Full history dumps...
 293                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 294                         } elseif ( $this->history & WikiExporter::CURRENT ) {
 295                                 # Latest revision dumps...
 296                                 if ( $this->list_authors && $cond != '' )  { // List authors, if so desired
 297                                         $this->do_list_authors( $cond );
 298                                 }
 299                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 300                         } elseif ( $this->history & WikiExporter::STABLE ) {
 301                                 # "Stable" revision dumps...
 302                                 # Default JOIN, to be overridden...
 303                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 304                                 # One, and only one hook should set this, and return false
 305                                 if ( wfRunHooks( 'WikiExporter::dumpStableQuery', array( &$tables, &$opts, &$join ) ) ) {
 306                                         wfProfileOut( __METHOD__ );
 307                                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 308                                 }
 309                         } elseif ( $this->history & WikiExporter::RANGE ) {
 310                                 # Dump of revisions within a specified range
 311                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 312                                 $opts['ORDER BY'] = 'rev_page ASC, rev_id ASC';
 313                         } else {
 314                                 # Uknown history specification parameter?
 315                                 wfProfileOut( __METHOD__ );
 316                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 317                         }
 318                         # Query optimization hacks
 319                         if ( $cond == '' ) {
 320                                 $opts[] = 'STRAIGHT_JOIN';
 321                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 322                         }
 323                         # Build text join options
 324                         if ( $this->text != WikiExporter::STUB ) { // 1-pass
 325                                 $tables[] = 'text';
 326                                 $join['text'] = array( 'INNER JOIN', 'rev_text_id=old_id' );
 327                         }
 328
 329                         if ( $this->buffer == WikiExporter::STREAM ) {
 330                                 $prev = $this->db->bufferResults( false );
 331                         }
 332
 333                         $wrapper = null; // Assuring $wrapper is not undefined, if exception occurs early
 334                         try {
 335                                 wfRunHooks( 'ModifyExportQuery',
 336                                                 array( $this->db, &$tables, &$cond, &$opts, &$join ) );
 337
 338                                 # Do the query!
 339                                 $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 340                                 $wrapper = $this->db->resultObject( $result );
 341                                 # Output dump results
 342                                 $this->outputPageStream( $wrapper );
 343
 344                                 if ( $this->buffer == WikiExporter::STREAM ) {
 345                                         $this->db->bufferResults( $prev );
 346                                 }
 347                         } catch ( Exception $e ) {
 348                                 // Throwing the exception does not reliably free the resultset, and
 349                                 // would also leave the connection in unbuffered mode.
 350
 351                                 // Freeing result
 352                                 try {
 353                                         if ( $wrapper ) {
 354                                                 $wrapper->free();
 355                                         }
 356                                 } catch ( Exception $e2 ) {
 357                                         // Already in panic mode -> ignoring $e2 as $e has
 358                                         // higher priority
 359                                 }
 360
 361                                 // Putting database back in previous buffer mode
 362                                 try {
 363                                         if ( $this->buffer == WikiExporter::STREAM ) {
 364                                                 $this->db->bufferResults( $prev );
 365                                         }
 366                                 } catch ( Exception $e2 ) {
 367                                         // Already in panic mode -> ignoring $e2 as $e has
 368                                         // higher priority
 369                                 }
 370
 371                                 // Inform caller about problem
 372                                 throw $e;
 373                         }
 374                 }
 375                 wfProfileOut( __METHOD__ );
 376         }
 377
 378         /**
 379          * Runs through a query result set dumping page and revision records.
 380          * The result set should be sorted/grouped by page to avoid duplicate
 381          * page records in the output.
 382          *
 383          * Should be safe for
 384          * streaming (non-buffered) queries, as long as it was made on a
 385          * separate database connection not managed by LoadBalancer; some
 386          * blob storage types will make queries to pull source data.
 387          *
 388          * @param $resultset ResultWrapper
 389          */
 390         protected function outputPageStream( $resultset ) {
 391                 $last = null;
 392                 foreach ( $resultset as $row ) {
 393                         if ( is_null( $last ) ||
 394                                 $last->page_namespace != $row->page_namespace ||
 395                                 $last->page_title     != $row->page_title ) {
 396                                 if ( isset( $last ) ) {
 397                                         $output = '';
 398                                         if ( $this->dumpUploads ) {
 399                                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 400                                         }
 401                                         $output .= $this->writer->closePage();
 402                                         $this->sink->writeClosePage( $output );
 403                                 }
 404                                 $output = $this->writer->openPage( $row );
 405                                 $this->sink->writeOpenPage( $row, $output );
 406                                 $last = $row;
 407                         }
 408                         $output = $this->writer->writeRevision( $row );
 409                         $this->sink->writeRevision( $row, $output );
 410                 }
 411                 if ( isset( $last ) ) {
 412                         $output = '';
 413                         if ( $this->dumpUploads ) {
 414                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 415                         }
 416                         $output .= $this->author_list;
 417                         $output .= $this->writer->closePage();
 418                         $this->sink->writeClosePage( $output );
 419                 }
 420         }
 421
 422         protected function outputLogStream( $resultset ) {
 423                 foreach ( $resultset as $row ) {
 424                         $output = $this->writer->writeLogItem( $row );
 425                         $this->sink->writeLogItem( $row, $output );
 426                 }
 427         }
 428 }
 429
 430 /**
 431  * @ingroup Dump
 432  */
 433 class XmlDumpWriter {
 434         /**
 435          * Returns the export schema version.
 436          * @return string
 437          */
 438         function schemaVersion() {
 439                 return "0.6";
 440         }
 441
 442         /**
 443          * Opens the XML output stream's root <mediawiki> element.
 444          * This does not include an xml directive, so is safe to include
 445          * as a subelement in a larger XML stream. Namespace and XML Schema
 446          * references are included.
 447          *
 448          * Output will be encoded in UTF-8.
 449          *
 450          * @return string
 451          */
 452         function openStream() {
 453                 global $wgLanguageCode;
 454                 $ver = $this->schemaVersion();
 455                 return Xml::element( 'mediawiki', array(
 456                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 457                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 458                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 459                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 460                         'version'            => $ver,
 461                         'xml:lang'           => $wgLanguageCode ),
 462                         null ) .
 463                         "\n" .
 464                         $this->siteInfo();
 465         }
 466
 467         function siteInfo() {
 468                 $info = array(
 469                         $this->sitename(),
 470                         $this->homelink(),
 471                         $this->generator(),
 472                         $this->caseSetting(),
 473                         $this->namespaces() );
 474                 return "  <siteinfo>\n    " .
 475                         implode( "\n    ", $info ) .
 476                         "\n  </siteinfo>\n";
 477         }
 478
 479         function sitename() {
 480                 global $wgSitename;
 481                 return Xml::element( 'sitename', array(), $wgSitename );
 482         }
 483
 484         function generator() {
 485                 global $wgVersion;
 486                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 487         }
 488
 489         function homelink() {
 490                 return Xml::element( 'base', array(), Title::newMainPage()->getCanonicalUrl() );
 491         }
 492
 493         function caseSetting() {
 494                 global $wgCapitalLinks;
 495                 // "case-insensitive" option is reserved for future
 496                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 497                 return Xml::element( 'case', array(), $sensitivity );
 498         }
 499
 500         function namespaces() {
 501                 global $wgContLang;
 502                 $spaces = "<namespaces>\n";
 503                 foreach ( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 504                         $spaces .= '      ' .
 505                                 Xml::element( 'namespace',
 506                                         array(  'key' => $ns,
 507                                                         'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 508                                         ), $title ) . "\n";
 509                 }
 510                 $spaces .= "    </namespaces>";
 511                 return $spaces;
 512         }
 513
 514         /**
 515          * Closes the output stream with the closing root element.
 516          * Call when finished dumping things.
 517          *
 518          * @return string
 519          */
 520         function closeStream() {
 521                 return "</mediawiki>\n";
 522         }
 523
 524         /**
 525          * Opens a <page> section on the output stream, with data
 526          * from the given database row.
 527          *
 528          * @param $row object
 529          * @return string
 530          * @access private
 531          */
 532         function openPage( $row ) {
 533                 $out = "  <page>\n";
 534                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 535                 $out .= '    ' . Xml::elementClean( 'title', array(), self::canonicalTitle( $title ) ) . "\n";
 536                 $out .= '    ' . Xml::element( 'ns', array(), strval( $row->page_namespace) ) . "\n";
 537                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 538                 if ( $row->page_is_redirect ) {
 539                         $page = WikiPage::factory( $title );
 540                         $redirect = $page->getRedirectTarget();
 541                         if ( $redirect instanceOf Title && $redirect->isValidRedirectTarget() ) {
 542                                 $out .= '    ' . Xml::element( 'redirect', array( 'title' => self::canonicalTitle( $redirect ) ) ) . "\n";
 543                         }
 544                 }
 545
 546                 if ( $row->page_restrictions != '' ) {
 547                         $out .= '    ' . Xml::element( 'restrictions', array(),
 548                                 strval( $row->page_restrictions ) ) . "\n";
 549                 }
 550
 551                 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 552
 553                 return $out;
 554         }
 555
 556         /**
 557          * Closes a <page> section on the output stream.
 558          *
 559          * @access private
 560          * @return string
 561          */
 562         function closePage() {
 563                 return "  </page>\n";
 564         }
 565
 566         /**
 567          * Dumps a <revision> section on the output stream, with
 568          * data filled in from the given database row.
 569          *
 570          * @param $row object
 571          * @return string
 572          * @access private
 573          */
 574         function writeRevision( $row ) {
 575                 wfProfileIn( __METHOD__ );
 576
 577                 $out  = "    <revision>\n";
 578                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 579
 580                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 581
 582                 if ( $row->rev_deleted & Revision::DELETED_USER ) {
 583                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 584                 } else {
 585                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 586                 }
 587
 588                 if ( $row->rev_minor_edit ) {
 589                         $out .=  "      <minor/>\n";
 590                 }
 591                 if ( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 592                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 593                 } elseif ( $row->rev_comment != '' ) {
 594                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 595                 }
 596
 597                 if ( isset( $row->rev_content_model ) && !is_null( $row->rev_content_model )  ) {
 598                         $name = ContentHandler::getContentModelName( $row->rev_content_model );
 599                         $out .= "      " . Xml::element('model', array( 'name' => $name ), strval( $row->rev_content_model ) ) . "\n";
 600                 }
 601
 602                 if ( isset( $row->rev_content_format ) && !is_null( $row->rev_content_format ) ) {
 603                         $mime = ContentHandler::getContentFormatMimeType( $row->rev_content_format );
 604                         $out .= "      " . Xml::element('format', array( 'mime' => $mime ), strval( $row->rev_content_format ) ) . "\n";
 605                 }
 606
 607                 $text = '';
 608                 if ( $row->rev_deleted & Revision::DELETED_TEXT ) {
 609                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 610                 } elseif ( isset( $row->old_text ) ) {
 611                         // Raw text from the database may have invalid chars
 612                         $text = strval( Revision::getRevisionText( $row ) );
 613                         $out .= "      " . Xml::elementClean( 'text',
 614                                 array( 'xml:space' => 'preserve', 'bytes' => intval( $row->rev_len ) ),
 615                                 strval( $text ) ) . "\n";
 616                 } else {
 617                         // Stub output
 618                         $out .= "      " . Xml::element( 'text',
 619                                 array( 'id' => $row->rev_text_id, 'bytes' => intval( $row->rev_len ) ),
 620                                 "" ) . "\n";
 621                 }
 622
 623                 if ( $row->rev_sha1 && !( $row->rev_deleted & Revision::DELETED_TEXT ) ) {
 624                         $out .= "      " . Xml::element('sha1', null, strval( $row->rev_sha1 ) ) . "\n";
 625                 } else {
 626                         $out .= "      <sha1/>\n";
 627                 }
 628
 629                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 630
 631                 $out .= "    </revision>\n";
 632
 633                 wfProfileOut( __METHOD__ );
 634                 return $out;
 635         }
 636
 637         /**
 638          * Dumps a <logitem> section on the output stream, with
 639          * data filled in from the given database row.
 640          *
 641          * @param $row object
 642          * @return string
 643          * @access private
 644          */
 645         function writeLogItem( $row ) {
 646                 wfProfileIn( __METHOD__ );
 647
 648                 $out  = "    <logitem>\n";
 649                 $out .= "      " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 650
 651                 $out .= $this->writeTimestamp( $row->log_timestamp );
 652
 653                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 654                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 655                 } else {
 656                         $out .= $this->writeContributor( $row->log_user, $row->user_name );
 657                 }
 658
 659                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 660                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 661                 } elseif ( $row->log_comment != '' ) {
 662                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 663                 }
 664
 665                 $out .= "      " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 666                 $out .= "      " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 667
 668                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 669                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 670                 } else {
 671                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 672                         $out .= "      " . Xml::elementClean( 'logtitle', null, self::canonicalTitle( $title ) ) . "\n";
 673                         $out .= "      " . Xml::elementClean( 'params',
 674                                 array( 'xml:space' => 'preserve' ),
 675                                 strval( $row->log_params ) ) . "\n";
 676                 }
 677
 678                 $out .= "    </logitem>\n";
 679
 680                 wfProfileOut( __METHOD__ );
 681                 return $out;
 682         }
 683
 684         function writeTimestamp( $timestamp ) {
 685                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 686                 return "      " . Xml::element( 'timestamp', null, $ts ) . "\n";
 687         }
 688
 689         function writeContributor( $id, $text ) {
 690                 $out = "      <contributor>\n";
 691                 if ( $id || !IP::isValid( $text ) ) {
 692                         $out .= "        " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 693                         $out .= "        " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 694                 } else {
 695                         $out .= "        " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 696                 }
 697                 $out .= "      </contributor>\n";
 698                 return $out;
 699         }
 700
 701         /**
 702          * Warning! This data is potentially inconsistent. :(
 703          * @return string
 704          */
 705         function writeUploads( $row, $dumpContents = false ) {
 706                 if ( $row->page_namespace == NS_IMAGE ) {
 707                         $img = wfLocalFile( $row->page_title );
 708                         if ( $img && $img->exists() ) {
 709                                 $out = '';
 710                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 711                                         $out .= $this->writeUpload( $ver, $dumpContents );
 712                                 }
 713                                 $out .= $this->writeUpload( $img, $dumpContents );
 714                                 return $out;
 715                         }
 716                 }
 717                 return '';
 718         }
 719
 720         /**
 721          * @param $file File
 722          * @param $dumpContents bool
 723          * @return string
 724          */
 725         function writeUpload( $file, $dumpContents = false ) {
 726                 if ( $file->isOld() ) {
 727                         $archiveName = "      " .
 728                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 729                 } else {
 730                         $archiveName = '';
 731                 }
 732                 if ( $dumpContents ) {
 733                         # Dump file as base64
 734                         # Uses only XML-safe characters, so does not need escaping
 735                         $contents = '      <contents encoding="base64">' .
 736                                 chunk_split( base64_encode( file_get_contents( $file->getPath() ) ) ) .
 737                                 "      </contents>\n";
 738                 } else {
 739                         $contents = '';
 740                 }
 741                 return "    <upload>\n" .
 742                         $this->writeTimestamp( $file->getTimestamp() ) .
 743                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 744                         "      " . Xml::elementClean( 'comment', null, $file->getDescription() ) . "\n" .
 745                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 746                         $archiveName .
 747                         "      " . Xml::element( 'src', null, $file->getCanonicalUrl() ) . "\n" .
 748                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 749                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 750                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 751                         $contents .
 752                         "    </upload>\n";
 753         }
 754
 755         /**
 756          * Return prefixed text form of title, but using the content language's
 757          * canonical namespace. This skips any special-casing such as gendered
 758          * user namespaces -- which while useful, are not yet listed in the
 759          * XML <siteinfo> data so are unsafe in export.
 760          *
 761          * @param Title $title
 762          * @return string
 763          * @since 1.18
 764          */
 765         public static function canonicalTitle( Title $title ) {
 766                 if ( $title->getInterwiki() ) {
 767                         return $title->getPrefixedText();
 768                 }
 769
 770                 global $wgContLang;
 771                 $prefix = str_replace( '_', ' ', $wgContLang->getNsText( $title->getNamespace() ) );
 772
 773                 if ( $prefix !== '' ) {
 774                         $prefix .= ':';
 775                 }
 776
 777                 return $prefix . $title->getText();
 778         }
 779 }
 780
 781
 782 /**
 783  * Base class for output stream; prints to stdout or buffer or whereever.
 784  * @ingroup Dump
 785  */
 786 class DumpOutput {
 787         function writeOpenStream( $string ) {
 788                 $this->write( $string );
 789         }
 790
 791         function writeCloseStream( $string ) {
 792                 $this->write( $string );
 793         }
 794
 795         function writeOpenPage( $page, $string ) {
 796                 $this->write( $string );
 797         }
 798
 799         function writeClosePage( $string ) {
 800                 $this->write( $string );
 801         }
 802
 803         function writeRevision( $rev, $string ) {
 804                 $this->write( $string );
 805         }
 806
 807         function writeLogItem( $rev, $string ) {
 808                 $this->write( $string );
 809         }
 810
 811         /**
 812          * Override to write to a different stream type.
 813          * @return bool
 814          */
 815         function write( $string ) {
 816                 print $string;
 817         }
 818
 819         /**
 820          * Close the old file, move it to a specified name,
 821          * and reopen new file with the old name. Use this
 822          * for writing out a file in multiple pieces
 823          * at specified checkpoints (e.g. every n hours).
 824          * @param $newname mixed File name. May be a string or an array with one element
 825          */
 826         function closeRenameAndReopen( $newname ) {
 827                 return;
 828         }
 829
 830         /**
 831          * Close the old file, and move it to a specified name.
 832          * Use this for the last piece of a file written out
 833          * at specified checkpoints (e.g. every n hours).
 834          * @param $newname mixed File name. May be a string or an array with one element
 835          * @param $open bool If true, a new file with the old filename will be opened again for writing (default: false)
 836          */
 837         function closeAndRename( $newname, $open = false ) {
 838                 return;
 839         }
 840
 841         /**
 842          * Returns the name of the file or files which are
 843          * being written to, if there are any.
 844          * @return null
 845          */
 846         function getFilenames() {
 847                 return NULL;
 848         }
 849 }
 850
 851 /**
 852  * Stream outputter to send data to a file.
 853  * @ingroup Dump
 854  */
 855 class DumpFileOutput extends DumpOutput {
 856         protected $handle = false, $filename;
 857
 858         function __construct( $file ) {
 859                 $this->handle = fopen( $file, "wt" );
 860                 $this->filename = $file;
 861         }
 862
 863         function writeCloseStream( $string ) {
 864                 parent::writeCloseStream( $string );
 865                 if ( $this->handle ) {
 866                         fclose( $this->handle );
 867                         $this->handle = false;
 868                 }
 869         }
 870
 871         function write( $string ) {
 872                 fputs( $this->handle, $string );
 873         }
 874
 875         function closeRenameAndReopen( $newname ) {
 876                 $this->closeAndRename( $newname, true );
 877         }
 878
 879         function renameOrException( $newname ) {
 880                         if (! rename( $this->filename, $newname ) ) {
 881                                 throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" );
 882                         }
 883         }
 884
 885         function checkRenameArgCount( $newname ) {
 886                 if ( is_array( $newname ) ) {
 887                         if ( count( $newname ) > 1 ) {
 888                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
 889                         } else {
 890                                 $newname = $newname[0];
 891                         }
 892                 }
 893                 return $newname;
 894         }
 895
 896         function closeAndRename( $newname, $open = false ) {
 897                 $newname = $this->checkRenameArgCount( $newname );
 898                 if ( $newname ) {
 899                         if ( $this->handle ) {
 900                                 fclose( $this->handle );
 901                                 $this->handle = false;
 902                         }
 903                         $this->renameOrException( $newname );
 904                         if ( $open ) {
 905                                 $this->handle = fopen( $this->filename, "wt" );
 906                         }
 907                 }
 908         }
 909
 910         function getFilenames() {
 911                 return $this->filename;
 912         }
 913 }
 914
 915 /**
 916  * Stream outputter to send data to a file via some filter program.
 917  * Even if compression is available in a library, using a separate
 918  * program can allow us to make use of a multi-processor system.
 919  * @ingroup Dump
 920  */
 921 class DumpPipeOutput extends DumpFileOutput {
 922         protected $command, $filename;
 923         protected $procOpenResource = false;
 924
 925         function __construct( $command, $file = null ) {
 926                 if ( !is_null( $file ) ) {
 927                         $command .=  " > " . wfEscapeShellArg( $file );
 928                 }
 929
 930                 $this->startCommand( $command );
 931                 $this->command = $command;
 932                 $this->filename = $file;
 933         }
 934
 935         function writeCloseStream( $string ) {
 936                 parent::writeCloseStream( $string );
 937                 if ( $this->procOpenResource ) {
 938                         proc_close( $this->procOpenResource );
 939                         $this->procOpenResource = false;
 940                 }
 941         }
 942
 943         function startCommand( $command ) {
 944                 $spec = array(
 945                         0 => array( "pipe", "r" ),
 946                 );
 947                 $pipes = array();
 948                 $this->procOpenResource = proc_open( $command, $spec, $pipes );
 949                 $this->handle = $pipes[0];
 950         }
 951
 952         function closeRenameAndReopen( $newname ) {
 953                 $this->closeAndRename( $newname, true );
 954         }
 955
 956         function closeAndRename( $newname, $open = false ) {
 957                 $newname = $this->checkRenameArgCount( $newname );
 958                 if ( $newname ) {
 959                         if ( $this->handle ) {
 960                                 fclose( $this->handle );
 961                                 $this->handle = false;
 962                         }
 963                         if ( $this->procOpenResource ) {
 964                                 proc_close( $this->procOpenResource );
 965                                 $this->procOpenResource = false;
 966                         }
 967                         $this->renameOrException( $newname );
 968                         if ( $open ) {
 969                                 $command = $this->command;
 970                                 $command .=  " > " . wfEscapeShellArg( $this->filename );
 971                                 $this->startCommand( $command );
 972                         }
 973                 }
 974         }
 975
 976 }
 977
 978 /**
 979  * Sends dump output via the gzip compressor.
 980  * @ingroup Dump
 981  */
 982 class DumpGZipOutput extends DumpPipeOutput {
 983         function __construct( $file ) {
 984                 parent::__construct( "gzip", $file );
 985         }
 986 }
 987
 988 /**
 989  * Sends dump output via the bgzip2 compressor.
 990  * @ingroup Dump
 991  */
 992 class DumpBZip2Output extends DumpPipeOutput {
 993         function __construct( $file ) {
 994                 parent::__construct( "bzip2", $file );
 995         }
 996 }
 997
 998 /**
 999  * Sends dump output via the p7zip compressor.
1000  * @ingroup Dump
1001  */
1002 class Dump7ZipOutput extends DumpPipeOutput {
1003         function __construct( $file ) {
1004                 $command = $this->setup7zCommand( $file );
1005                 parent::__construct( $command );
1006                 $this->filename = $file;
1007         }
1008
1009         function setup7zCommand( $file ) {
1010                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
1011                 // Suppress annoying useless crap from p7zip
1012                 // Unfortunately this could suppress real error messages too
1013                 $command .= ' >' . wfGetNull() . ' 2>&1';
1014                 return( $command );
1015         }
1016
1017         function closeAndRename( $newname, $open = false ) {
1018                 $newname = $this->checkRenameArgCount( $newname );
1019                 if ( $newname ) {
1020                         fclose( $this->handle );
1021                         proc_close( $this->procOpenResource );
1022                         $this->renameOrException( $newname );
1023                         if ( $open ) {
1024                                 $command = $this->setup7zCommand( $this->filename );
1025                                 $this->startCommand( $command );
1026                         }
1027                 }
1028         }
1029 }
1030
1031
1032
1033 /**
1034  * Dump output filter class.
1035  * This just does output filtering and streaming; XML formatting is done
1036  * higher up, so be careful in what you do.
1037  * @ingroup Dump
1038  */
1039 class DumpFilter {
1040         function __construct( &$sink ) {
1041                 $this->sink =& $sink;
1042         }
1043
1044         function writeOpenStream( $string ) {
1045                 $this->sink->writeOpenStream( $string );
1046         }
1047
1048         function writeCloseStream( $string ) {
1049                 $this->sink->writeCloseStream( $string );
1050         }
1051
1052         function writeOpenPage( $page, $string ) {
1053                 $this->sendingThisPage = $this->pass( $page, $string );
1054                 if ( $this->sendingThisPage ) {
1055                         $this->sink->writeOpenPage( $page, $string );
1056                 }
1057         }
1058
1059         function writeClosePage( $string ) {
1060                 if ( $this->sendingThisPage ) {
1061                         $this->sink->writeClosePage( $string );
1062                         $this->sendingThisPage = false;
1063                 }
1064         }
1065
1066         function writeRevision( $rev, $string ) {
1067                 if ( $this->sendingThisPage ) {
1068                         $this->sink->writeRevision( $rev, $string );
1069                 }
1070         }
1071
1072         function writeLogItem( $rev, $string ) {
1073                 $this->sink->writeRevision( $rev, $string );
1074         }
1075
1076         function closeRenameAndReopen( $newname ) {
1077                 $this->sink->closeRenameAndReopen( $newname );
1078         }
1079
1080         function closeAndRename( $newname, $open = false ) {
1081                 $this->sink->closeAndRename( $newname, $open );
1082         }
1083
1084         function getFilenames() {
1085                 return $this->sink->getFilenames();
1086         }
1087
1088         /**
1089          * Override for page-based filter types.
1090          * @return bool
1091          */
1092         function pass( $page ) {
1093                 return true;
1094         }
1095 }
1096
1097 /**
1098  * Simple dump output filter to exclude all talk pages.
1099  * @ingroup Dump
1100  */
1101 class DumpNotalkFilter extends DumpFilter {
1102         function pass( $page ) {
1103                 return !MWNamespace::isTalk( $page->page_namespace );
1104         }
1105 }
1106
1107 /**
1108  * Dump output filter to include or exclude pages in a given set of namespaces.
1109  * @ingroup Dump
1110  */
1111 class DumpNamespaceFilter extends DumpFilter {
1112         var $invert = false;
1113         var $namespaces = array();
1114
1115         function __construct( &$sink, $param ) {
1116                 parent::__construct( $sink );
1117
1118                 $constants = array(
1119                         "NS_MAIN"           => NS_MAIN,
1120                         "NS_TALK"           => NS_TALK,
1121                         "NS_USER"           => NS_USER,
1122                         "NS_USER_TALK"      => NS_USER_TALK,
1123                         "NS_PROJECT"        => NS_PROJECT,
1124                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
1125                         "NS_FILE"           => NS_FILE,
1126                         "NS_FILE_TALK"      => NS_FILE_TALK,
1127                         "NS_IMAGE"          => NS_IMAGE,  // NS_IMAGE is an alias for NS_FILE
1128                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
1129                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
1130                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
1131                         "NS_TEMPLATE"       => NS_TEMPLATE,
1132                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
1133                         "NS_HELP"           => NS_HELP,
1134                         "NS_HELP_TALK"      => NS_HELP_TALK,
1135                         "NS_CATEGORY"       => NS_CATEGORY,
1136                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
1137
1138                 if ( $param { 0 } == '!' ) {
1139                         $this->invert = true;
1140                         $param = substr( $param, 1 );
1141                 }
1142
1143                 foreach ( explode( ',', $param ) as $key ) {
1144                         $key = trim( $key );
1145                         if ( isset( $constants[$key] ) ) {
1146                                 $ns = $constants[$key];
1147                                 $this->namespaces[$ns] = true;
1148                         } elseif ( is_numeric( $key ) ) {
1149                                 $ns = intval( $key );
1150                                 $this->namespaces[$ns] = true;
1151                         } else {
1152                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
1153                         }
1154                 }
1155         }
1156
1157         function pass( $page ) {
1158                 $match = isset( $this->namespaces[$page->page_namespace] );
1159                 return $this->invert xor $match;
1160         }
1161 }
1162
1163
1164 /**
1165  * Dump output filter to include only the last revision in each page sequence.
1166  * @ingroup Dump
1167  */
1168 class DumpLatestFilter extends DumpFilter {
1169         var $page, $pageString, $rev, $revString;
1170
1171         function writeOpenPage( $page, $string ) {
1172                 $this->page = $page;
1173                 $this->pageString = $string;
1174         }
1175
1176         function writeClosePage( $string ) {
1177                 if ( $this->rev ) {
1178                         $this->sink->writeOpenPage( $this->page, $this->pageString );
1179                         $this->sink->writeRevision( $this->rev, $this->revString );
1180                         $this->sink->writeClosePage( $string );
1181                 }
1182                 $this->rev = null;
1183                 $this->revString = null;
1184                 $this->page = null;
1185                 $this->pageString = null;
1186         }
1187
1188         function writeRevision( $rev, $string ) {
1189                 if ( $rev->rev_id == $this->page->page_latest ) {
1190                         $this->rev = $rev;
1191                         $this->revString = $string;
1192                 }
1193         }
1194 }
1195
1196 /**
1197  * Base class for output stream; prints to stdout or buffer or whereever.
1198  * @ingroup Dump
1199  */
1200 class DumpMultiWriter {
1201         function __construct( $sinks ) {
1202                 $this->sinks = $sinks;
1203                 $this->count = count( $sinks );
1204         }
1205
1206         function writeOpenStream( $string ) {
1207                 for ( $i = 0; $i < $this->count; $i++ ) {
1208                         $this->sinks[$i]->writeOpenStream( $string );
1209                 }
1210         }
1211
1212         function writeCloseStream( $string ) {
1213                 for ( $i = 0; $i < $this->count; $i++ ) {
1214                         $this->sinks[$i]->writeCloseStream( $string );
1215                 }
1216         }
1217
1218         function writeOpenPage( $page, $string ) {
1219                 for ( $i = 0; $i < $this->count; $i++ ) {
1220                         $this->sinks[$i]->writeOpenPage( $page, $string );
1221                 }
1222         }
1223
1224         function writeClosePage( $string ) {
1225                 for ( $i = 0; $i < $this->count; $i++ ) {
1226                         $this->sinks[$i]->writeClosePage( $string );
1227                 }
1228         }
1229
1230         function writeRevision( $rev, $string ) {
1231                 for ( $i = 0; $i < $this->count; $i++ ) {
1232                         $this->sinks[$i]->writeRevision( $rev, $string );
1233                 }
1234         }
1235
1236         function closeRenameAndReopen( $newnames ) {
1237                 $this->closeAndRename( $newnames, true );
1238         }
1239
1240         function closeAndRename( $newnames, $open = false ) {
1241                 for ( $i = 0; $i < $this->count; $i++ ) {
1242                         $this->sinks[$i]->closeAndRename( $newnames[$i], $open );
1243                 }
1244         }
1245
1246         function getFilenames() {
1247                 $filenames = array();
1248                 for ( $i = 0; $i < $this->count; $i++ ) {
1249                         $filenames[] =  $this->sinks[$i]->getFilenames();
1250                 }
1251                 return $filenames;
1252         }
1253
1254 }
1255
1256 function xmlsafe( $string ) {
1257         wfProfileIn( __FUNCTION__ );
1258
1259         /**
1260          * The page may contain old data which has not been properly normalized.
1261          * Invalid UTF-8 sequences or forbidden control characters will make our
1262          * XML output invalid, so be sure to strip them out.
1263          */
1264         $string = UtfNormal::cleanUp( $string );
1265
1266         $string = htmlspecialchars( $string );
1267         wfProfileOut( __FUNCTION__ );
1268         return $string;
1269 }