includes/Export.php

   1 <?php
   2 /**
   3  * Base classes for dumps and export
   4  *
   5  * Copyright © 2003, 2005, 2006 Brion Vibber <brion@pobox.com>
   6  * http://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  */
  25
  26 /**
  27  * @defgroup Dump Dump
  28  */
  29
  30 /**
  31  * @ingroup SpecialPage Dump
  32  */
  33 class WikiExporter {
  34         var $list_authors = false ; # Return distinct author list (when not returning full history)
  35         var $author_list = "" ;
  36
  37         var $dumpUploads = false;
  38         var $dumpUploadFileContents = false;
  39
  40         const FULL = 1;
  41         const CURRENT = 2;
  42         const STABLE = 4; // extension defined
  43         const LOGS = 8;
  44
  45         const BUFFER = 0;
  46         const STREAM = 1;
  47
  48         const TEXT = 0;
  49         const STUB = 1;
  50
  51         /**
  52          * If using WikiExporter::STREAM to stream a large amount of data,
  53          * provide a database connection which is not managed by
  54          * LoadBalancer to read from: some history blob types will
  55          * make additional queries to pull source data while the
  56          * main query is still running.
  57          *
  58          * @param $db Database
  59          * @param $history Mixed: one of WikiExporter::FULL or WikiExporter::CURRENT,
  60          *                 or an associative array:
  61          *                   offset: non-inclusive offset at which to start the query
  62          *                   limit: maximum number of rows to return
  63          *                   dir: "asc" or "desc" timestamp order
  64          * @param $buffer Int: one of WikiExporter::BUFFER or WikiExporter::STREAM
  65          * @param $text Int: one of WikiExporter::TEXT or WikiExporter::STUB
  66          */
  67         function __construct( &$db, $history = WikiExporter::CURRENT,
  68                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  69                 $this->db =& $db;
  70                 $this->history = $history;
  71                 $this->buffer  = $buffer;
  72                 $this->writer  = new XmlDumpWriter();
  73                 $this->sink    = new DumpOutput();
  74                 $this->text    = $text;
  75         }
  76
  77         /**
  78          * Set the DumpOutput or DumpFilter object which will receive
  79          * various row objects and XML output for filtering. Filters
  80          * can be chained or used as callbacks.
  81          *
  82          * @param $sink mixed
  83          */
  84         public function setOutputSink( &$sink ) {
  85                 $this->sink =& $sink;
  86         }
  87
  88         public function openStream() {
  89                 $output = $this->writer->openStream();
  90                 $this->sink->writeOpenStream( $output );
  91         }
  92
  93         public function closeStream() {
  94                 $output = $this->writer->closeStream();
  95                 $this->sink->writeCloseStream( $output );
  96         }
  97
  98         /**
  99          * Dumps a series of page and revision records for all pages
 100          * in the database, either including complete history or only
 101          * the most recent version.
 102          */
 103         public function allPages() {
 104                 return $this->dumpFrom( '' );
 105         }
 106
 107         /**
 108          * Dumps a series of page and revision records for those pages
 109          * in the database falling within the page_id range given.
 110          * @param $start Int: inclusive lower limit (this id is included)
 111          * @param $end   Int: Exclusive upper limit (this id is not included)
 112          *                   If 0, no upper limit.
 113          */
 114         public function pagesByRange( $start, $end ) {
 115                 $condition = 'page_id >= ' . intval( $start );
 116                 if ( $end ) {
 117                         $condition .= ' AND page_id < ' . intval( $end );
 118                 }
 119                 return $this->dumpFrom( $condition );
 120         }
 121
 122         /**
 123          * @param $title Title
 124          */
 125         public function pageByTitle( $title ) {
 126                 return $this->dumpFrom(
 127                         'page_namespace=' . $title->getNamespace() .
 128                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 129         }
 130
 131         public function pageByName( $name ) {
 132                 $title = Title::newFromText( $name );
 133                 if ( is_null( $title ) ) {
 134                         throw new MWException( "Can't export invalid title" );
 135                 } else {
 136                         return $this->pageByTitle( $title );
 137                 }
 138         }
 139
 140         public function pagesByName( $names ) {
 141                 foreach ( $names as $name ) {
 142                         $this->pageByName( $name );
 143                 }
 144         }
 145
 146         public function allLogs() {
 147                 return $this->dumpFrom( '' );
 148         }
 149
 150         public function logsByRange( $start, $end ) {
 151                 $condition = 'log_id >= ' . intval( $start );
 152                 if ( $end ) {
 153                         $condition .= ' AND log_id < ' . intval( $end );
 154                 }
 155                 return $this->dumpFrom( $condition );
 156         }
 157
 158         # Generates the distinct list of authors of an article
 159         # Not called by default (depends on $this->list_authors)
 160         # Can be set by Special:Export when not exporting whole history
 161         protected function do_list_authors( $cond ) {
 162                 wfProfileIn( __METHOD__ );
 163                 $this->author_list = "<contributors>";
 164                 // rev_deleted
 165
 166                 $res = $this->db->select(
 167                         array( 'page', 'revision' ),
 168                         array( 'DISTINCT rev_user_text', 'rev_user' ),
 169                         array(
 170                                 $this->db->bitAnd( 'rev_deleted', Revision::DELETED_USER ) . ' = 0',
 171                                 $cond,
 172                                 'page_id = rev_id',
 173                         ),
 174                         __METHOD__
 175                 );
 176
 177                 foreach ( $res as $row ) {
 178                         $this->author_list .= "<contributor>" .
 179                                 "<username>" .
 180                                 htmlentities( $row->rev_user_text )  .
 181                                 "</username>" .
 182                                 "<id>" .
 183                                 $row->rev_user .
 184                                 "</id>" .
 185                                 "</contributor>";
 186                 }
 187                 $this->author_list .= "</contributors>";
 188                 wfProfileOut( __METHOD__ );
 189         }
 190
 191         protected function dumpFrom( $cond = '' ) {
 192                 wfProfileIn( __METHOD__ );
 193                 # For logging dumps...
 194                 if ( $this->history & self::LOGS ) {
 195                         if ( $this->buffer == WikiExporter::STREAM ) {
 196                                 $prev = $this->db->bufferResults( false );
 197                         }
 198                         $where = array( 'user_id = log_user' );
 199                         # Hide private logs
 200                         $hideLogs = LogEventsList::getExcludeClause( $this->db );
 201                         if ( $hideLogs ) $where[] = $hideLogs;
 202                         # Add on any caller specified conditions
 203                         if ( $cond ) $where[] = $cond;
 204                         # Get logging table name for logging.* clause
 205                         $logging = $this->db->tableName( 'logging' );
 206                         $result = $this->db->select( array( 'logging', 'user' ),
 207                                 array( "{$logging}.*", 'user_name' ), // grab the user name
 208                                 $where,
 209                                 __METHOD__,
 210                                 array( 'ORDER BY' => 'log_id', 'USE INDEX' => array( 'logging' => 'PRIMARY' ) )
 211                         );
 212                         $wrapper = $this->db->resultObject( $result );
 213                         $this->outputLogStream( $wrapper );
 214                         if ( $this->buffer == WikiExporter::STREAM ) {
 215                                 $this->db->bufferResults( $prev );
 216                         }
 217                 # For page dumps...
 218                 } else {
 219                         $tables = array( 'page', 'revision' );
 220                         $opts = array( 'ORDER BY' => 'page_id ASC' );
 221                         $opts['USE INDEX'] = array();
 222                         $join = array();
 223                         if ( is_array( $this->history ) ) {
 224                                 # Time offset/limit for all pages/history...
 225                                 $revJoin = 'page_id=rev_page';
 226                                 # Set time order
 227                                 if ( $this->history['dir'] == 'asc' ) {
 228                                         $op = '>';
 229                                         $opts['ORDER BY'] = 'rev_timestamp ASC';
 230                                 } else {
 231                                         $op = '<';
 232                                         $opts['ORDER BY'] = 'rev_timestamp DESC';
 233                                 }
 234                                 # Set offset
 235                                 if ( !empty( $this->history['offset'] ) ) {
 236                                         $revJoin .= " AND rev_timestamp $op " .
 237                                                 $this->db->addQuotes( $this->db->timestamp( $this->history['offset'] ) );
 238                                 }
 239                                 $join['revision'] = array( 'INNER JOIN', $revJoin );
 240                                 # Set query limit
 241                                 if ( !empty( $this->history['limit'] ) ) {
 242                                         $opts['LIMIT'] = intval( $this->history['limit'] );
 243                                 }
 244                         } elseif ( $this->history & WikiExporter::FULL ) {
 245                                 # Full history dumps...
 246                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
 247                         } elseif ( $this->history & WikiExporter::CURRENT ) {
 248                                 # Latest revision dumps...
 249                                 if ( $this->list_authors && $cond != '' )  { // List authors, if so desired
 250                                         $this->do_list_authors( $cond );
 251                                 }
 252                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 253                         } elseif ( $this->history & WikiExporter::STABLE ) {
 254                                 # "Stable" revision dumps...
 255                                 # Default JOIN, to be overridden...
 256                                 $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page AND page_latest=rev_id' );
 257                                 # One, and only one hook should set this, and return false
 258                                 if ( wfRunHooks( 'WikiExporter::dumpStableQuery', array( &$tables, &$opts, &$join ) ) ) {
 259                                         wfProfileOut( __METHOD__ );
 260                                         throw new MWException( __METHOD__ . " given invalid history dump type." );
 261                                 }
 262                         } else {
 263                                 # Uknown history specification parameter?
 264                                 wfProfileOut( __METHOD__ );
 265                                 throw new MWException( __METHOD__ . " given invalid history dump type." );
 266                         }
 267                         # Query optimization hacks
 268                         if ( $cond == '' ) {
 269                                 $opts[] = 'STRAIGHT_JOIN';
 270                                 $opts['USE INDEX']['page'] = 'PRIMARY';
 271                         }
 272                         # Build text join options
 273                         if ( $this->text != WikiExporter::STUB ) { // 1-pass
 274                                 $tables[] = 'text';
 275                                 $join['text'] = array( 'INNER JOIN', 'rev_text_id=old_id' );
 276                         }
 277
 278                         if ( $this->buffer == WikiExporter::STREAM ) {
 279                                 $prev = $this->db->bufferResults( false );
 280                         }
 281
 282                         wfRunHooks( 'ModifyExportQuery',
 283                                                 array( $this->db, &$tables, &$cond, &$opts, &$join ) );
 284
 285                         # Do the query!
 286                         $result = $this->db->select( $tables, '*', $cond, __METHOD__, $opts, $join );
 287                         $wrapper = $this->db->resultObject( $result );
 288                         # Output dump results
 289                         $this->outputPageStream( $wrapper );
 290                         if ( $this->list_authors ) {
 291                                 $this->outputPageStream( $wrapper );
 292                         }
 293
 294                         if ( $this->buffer == WikiExporter::STREAM ) {
 295                                 $this->db->bufferResults( $prev );
 296                         }
 297                 }
 298                 wfProfileOut( __METHOD__ );
 299         }
 300
 301         /**
 302          * Runs through a query result set dumping page and revision records.
 303          * The result set should be sorted/grouped by page to avoid duplicate
 304          * page records in the output.
 305          *
 306          * The result set will be freed once complete. Should be safe for
 307          * streaming (non-buffered) queries, as long as it was made on a
 308          * separate database connection not managed by LoadBalancer; some
 309          * blob storage types will make queries to pull source data.
 310          *
 311          * @param $resultset ResultWrapper
 312          */
 313         protected function outputPageStream( $resultset ) {
 314                 $last = null;
 315                 foreach ( $resultset as $row ) {
 316                         if ( is_null( $last ) ||
 317                                 $last->page_namespace != $row->page_namespace ||
 318                                 $last->page_title     != $row->page_title ) {
 319                                 if ( isset( $last ) ) {
 320                                         $output = '';
 321                                         if ( $this->dumpUploads ) {
 322                                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 323                                         }
 324                                         $output .= $this->writer->closePage();
 325                                         $this->sink->writeClosePage( $output );
 326                                 }
 327                                 $output = $this->writer->openPage( $row );
 328                                 $this->sink->writeOpenPage( $row, $output );
 329                                 $last = $row;
 330                         }
 331                         $output = $this->writer->writeRevision( $row );
 332                         $this->sink->writeRevision( $row, $output );
 333                 }
 334                 if ( isset( $last ) ) {
 335                         $output = '';
 336                         if ( $this->dumpUploads ) {
 337                                 $output .= $this->writer->writeUploads( $last, $this->dumpUploadFileContents );
 338                         }
 339                         $output .= $this->author_list;
 340                         $output .= $this->writer->closePage();
 341                         $this->sink->writeClosePage( $output );
 342                 }
 343         }
 344
 345         protected function outputLogStream( $resultset ) {
 346                 foreach ( $resultset as $row ) {
 347                         $output = $this->writer->writeLogItem( $row );
 348                         $this->sink->writeLogItem( $row, $output );
 349                 }
 350         }
 351 }
 352
 353 /**
 354  * @ingroup Dump
 355  */
 356 class XmlDumpWriter {
 357         /**
 358          * Returns the export schema version.
 359          * @return string
 360          */
 361         function schemaVersion() {
 362                 return "0.5";
 363         }
 364
 365         /**
 366          * Opens the XML output stream's root <mediawiki> element.
 367          * This does not include an xml directive, so is safe to include
 368          * as a subelement in a larger XML stream. Namespace and XML Schema
 369          * references are included.
 370          *
 371          * Output will be encoded in UTF-8.
 372          *
 373          * @return string
 374          */
 375         function openStream() {
 376                 global $wgLanguageCode;
 377                 $ver = $this->schemaVersion();
 378                 return Xml::element( 'mediawiki', array(
 379                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 380                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 381                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 382                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 383                         'version'            => $ver,
 384                         'xml:lang'           => $wgLanguageCode ),
 385                         null ) .
 386                         "\n" .
 387                         $this->siteInfo();
 388         }
 389
 390         function siteInfo() {
 391                 $info = array(
 392                         $this->sitename(),
 393                         $this->homelink(),
 394                         $this->generator(),
 395                         $this->caseSetting(),
 396                         $this->namespaces() );
 397                 return "  <siteinfo>\n    " .
 398                         implode( "\n    ", $info ) .
 399                         "\n  </siteinfo>\n";
 400         }
 401
 402         function sitename() {
 403                 global $wgSitename;
 404                 return Xml::element( 'sitename', array(), $wgSitename );
 405         }
 406
 407         function generator() {
 408                 global $wgVersion;
 409                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 410         }
 411
 412         function homelink() {
 413                 return Xml::element( 'base', array(), Title::newMainPage()->getCanonicalUrl() );
 414         }
 415
 416         function caseSetting() {
 417                 global $wgCapitalLinks;
 418                 // "case-insensitive" option is reserved for future
 419                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 420                 return Xml::element( 'case', array(), $sensitivity );
 421         }
 422
 423         function namespaces() {
 424                 global $wgContLang;
 425                 $spaces = "<namespaces>\n";
 426                 foreach ( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 427                         $spaces .= '      ' .
 428                                 Xml::element( 'namespace',
 429                                         array(  'key' => $ns,
 430                                                         'case' => MWNamespace::isCapitalized( $ns ) ? 'first-letter' : 'case-sensitive',
 431                                         ), $title ) . "\n";
 432                 }
 433                 $spaces .= "    </namespaces>";
 434                 return $spaces;
 435         }
 436
 437         /**
 438          * Closes the output stream with the closing root element.
 439          * Call when finished dumping things.
 440          *
 441          * @return string
 442          */
 443         function closeStream() {
 444                 return "</mediawiki>\n";
 445         }
 446
 447         /**
 448          * Opens a <page> section on the output stream, with data
 449          * from the given database row.
 450          *
 451          * @param $row object
 452          * @return string
 453          * @access private
 454          */
 455         function openPage( $row ) {
 456                 $out = "  <page>\n";
 457                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 458                 $out .= '    ' . Xml::elementClean( 'title', array(), $title->getPrefixedText() ) . "\n";
 459                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 460                 if ( $row->page_is_redirect ) {
 461                         $out .= '    ' . Xml::element( 'redirect', array() ) . "\n";
 462                 }
 463                 if ( $row->page_restrictions != '' ) {
 464                         $out .= '    ' . Xml::element( 'restrictions', array(),
 465                                 strval( $row->page_restrictions ) ) . "\n";
 466                 }
 467
 468                 wfRunHooks( 'XmlDumpWriterOpenPage', array( $this, &$out, $row, $title ) );
 469
 470                 return $out;
 471         }
 472
 473         /**
 474          * Closes a <page> section on the output stream.
 475          *
 476          * @access private
 477          */
 478         function closePage() {
 479                 return "  </page>\n";
 480         }
 481
 482         /**
 483          * Dumps a <revision> section on the output stream, with
 484          * data filled in from the given database row.
 485          *
 486          * @param $row object
 487          * @return string
 488          * @access private
 489          */
 490         function writeRevision( $row ) {
 491                 wfProfileIn( __METHOD__ );
 492
 493                 $out  = "    <revision>\n";
 494                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 495
 496                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 497
 498                 if ( $row->rev_deleted & Revision::DELETED_USER ) {
 499                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 500                 } else {
 501                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 502                 }
 503
 504                 if ( $row->rev_minor_edit ) {
 505                         $out .=  "      <minor/>\n";
 506                 }
 507                 if ( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 508                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 509                 } elseif ( $row->rev_comment != '' ) {
 510                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 511                 }
 512
 513                 $text = '';
 514                 if ( $row->rev_deleted & Revision::DELETED_TEXT ) {
 515                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 516                 } elseif ( isset( $row->old_text ) ) {
 517                         // Raw text from the database may have invalid chars
 518                         $text = strval( Revision::getRevisionText( $row ) );
 519                         $out .= "      " . Xml::elementClean( 'text',
 520                                 array( 'xml:space' => 'preserve', 'bytes' => $row->rev_len ),
 521                                 strval( $text ) ) . "\n";
 522                 } else {
 523                         // Stub output
 524                         $out .= "      " . Xml::element( 'text',
 525                                 array( 'id' => $row->rev_text_id, 'bytes' => $row->rev_len ),
 526                                 "" ) . "\n";
 527                 }
 528
 529                 wfRunHooks( 'XmlDumpWriterWriteRevision', array( &$this, &$out, $row, $text ) );
 530
 531                 $out .= "    </revision>\n";
 532
 533                 wfProfileOut( __METHOD__ );
 534                 return $out;
 535         }
 536
 537         /**
 538          * Dumps a <logitem> section on the output stream, with
 539          * data filled in from the given database row.
 540          *
 541          * @param $row object
 542          * @return string
 543          * @access private
 544          */
 545         function writeLogItem( $row ) {
 546                 wfProfileIn( __METHOD__ );
 547
 548                 $out  = "    <logitem>\n";
 549                 $out .= "      " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 550
 551                 $out .= $this->writeTimestamp( $row->log_timestamp );
 552
 553                 if ( $row->log_deleted & LogPage::DELETED_USER ) {
 554                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 555                 } else {
 556                         $out .= $this->writeContributor( $row->log_user, $row->user_name );
 557                 }
 558
 559                 if ( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 560                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 561                 } elseif ( $row->log_comment != '' ) {
 562                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 563                 }
 564
 565                 $out .= "      " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 566                 $out .= "      " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 567
 568                 if ( $row->log_deleted & LogPage::DELETED_ACTION ) {
 569                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 570                 } else {
 571                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 572                         $out .= "      " . Xml::elementClean( 'logtitle', null, $title->getPrefixedText() ) . "\n";
 573                         $out .= "      " . Xml::elementClean( 'params',
 574                                 array( 'xml:space' => 'preserve' ),
 575                                 strval( $row->log_params ) ) . "\n";
 576                 }
 577
 578                 $out .= "    </logitem>\n";
 579
 580                 wfProfileOut( __METHOD__ );
 581                 return $out;
 582         }
 583
 584         function writeTimestamp( $timestamp ) {
 585                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 586                 return "      " . Xml::element( 'timestamp', null, $ts ) . "\n";
 587         }
 588
 589         function writeContributor( $id, $text ) {
 590                 $out = "      <contributor>\n";
 591                 if ( $id ) {
 592                         $out .= "        " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 593                         $out .= "        " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 594                 } else {
 595                         $out .= "        " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 596                 }
 597                 $out .= "      </contributor>\n";
 598                 return $out;
 599         }
 600
 601         /**
 602          * Warning! This data is potentially inconsistent. :(
 603          */
 604         function writeUploads( $row, $dumpContents = false ) {
 605                 if ( $row->page_namespace == NS_IMAGE ) {
 606                         $img = wfLocalFile( $row->page_title );
 607                         if ( $img && $img->exists() ) {
 608                                 $out = '';
 609                                 foreach ( array_reverse( $img->getHistory() ) as $ver ) {
 610                                         $out .= $this->writeUpload( $ver, $dumpContents );
 611                                 }
 612                                 $out .= $this->writeUpload( $img, $dumpContents );
 613                                 return $out;
 614                         }
 615                 }
 616                 return '';
 617         }
 618
 619         /**
 620          * @param $file File
 621          * @param $dumpContents bool
 622          * @return string
 623          */
 624         function writeUpload( $file, $dumpContents = false ) {
 625                 if ( $file->isOld() ) {
 626                         $archiveName = "      " .
 627                                 Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
 628                 } else {
 629                         $archiveName = '';
 630                 }
 631                 if ( $dumpContents ) {
 632                         # Dump file as base64
 633                         # Uses only XML-safe characters, so does not need escaping
 634                         $contents = '      <contents encoding="base64">' .
 635                                 chunk_split( base64_encode( file_get_contents( $file->getPath() ) ) ) .
 636                                 "      </contents>\n";
 637                 } else {
 638                         $contents = '';
 639                 }
 640                 return "    <upload>\n" .
 641                         $this->writeTimestamp( $file->getTimestamp() ) .
 642                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 643                         "      " . Xml::elementClean( 'comment', null, $file->getDescription() ) . "\n" .
 644                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 645                         $archiveName .
 646                         "      " . Xml::element( 'src', null, $file->getCanonicalUrl() ) . "\n" .
 647                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 648                         "      " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
 649                         "      " . Xml::element( 'rel', null, $file->getRel() ) . "\n" .
 650                         $contents .
 651                         "    </upload>\n";
 652         }
 653
 654 }
 655
 656
 657 /**
 658  * Base class for output stream; prints to stdout or buffer or whereever.
 659  * @ingroup Dump
 660  */
 661 class DumpOutput {
 662         function writeOpenStream( $string ) {
 663                 $this->write( $string );
 664         }
 665
 666         function writeCloseStream( $string ) {
 667                 $this->write( $string );
 668         }
 669
 670         function writeOpenPage( $page, $string ) {
 671                 $this->write( $string );
 672         }
 673
 674         function writeClosePage( $string ) {
 675                 $this->write( $string );
 676         }
 677
 678         function writeRevision( $rev, $string ) {
 679                 $this->write( $string );
 680         }
 681
 682         function writeLogItem( $rev, $string ) {
 683                 $this->write( $string );
 684         }
 685
 686         /**
 687          * Override to write to a different stream type.
 688          * @return bool
 689          */
 690         function write( $string ) {
 691                 print $string;
 692         }
 693
 694         /**
 695          * Close the old file, move it to a specified name,
 696          * and reopen new file with the old name. Use this
 697          * for writing out a file in multiple pieces
 698          * at specified checkpoints (e.g. every n hours).
 699          * @param $newname mixed File name. May be a string or an array with one element
 700          */
 701         function closeRenameAndReopen( $newname ) {
 702                 return;
 703         }
 704
 705         /**
 706          * Close the old file, and move it to a specified name.
 707          * Use this for the last piece of a file written out
 708          * at specified checkpoints (e.g. every n hours).
 709          * @param $newname mixed File name. May be a string or an array with one element
 710          * @param $open bool If true, a new file with the old filename will be opened again for writing (default: false)
 711          */
 712         function closeAndRename( $newname, $open = false ) {
 713                 return;
 714         }
 715
 716         /**
 717          * Returns the name of the file or files which are
 718          * being written to, if there are any.
 719          */
 720         function getFilenames() {
 721                 return NULL;
 722         }
 723 }
 724
 725 /**
 726  * Stream outputter to send data to a file.
 727  * @ingroup Dump
 728  */
 729 class DumpFileOutput extends DumpOutput {
 730         protected $handle, $filename;
 731
 732         function __construct( $file ) {
 733                 $this->handle = fopen( $file, "wt" );
 734                 $this->filename = $file;
 735         }
 736
 737         function write( $string ) {
 738                 fputs( $this->handle, $string );
 739         }
 740
 741         function closeRenameAndReopen( $newname ) {
 742                 $this->closeAndRename( $newname, true );
 743         }
 744
 745         function closeAndRename( $newname, $open = false ) {
 746                 if ( is_array( $newname ) ) {
 747                         if ( count( $newname ) > 1 ) {
 748                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
 749                         } else {
 750                                 $newname = $newname[0];
 751                         }
 752                 }
 753                 if ( $newname ) {
 754                         fclose( $this->handle );
 755                         if (! rename( $this->filename, $newname ) ) {
 756                                 throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" );
 757                         }
 758                         elseif ( $open ) {
 759                                 $this->handle = fopen( $this->filename, "wt" );
 760                         }
 761                 }
 762         }
 763
 764         function getFilenames() {
 765                 return $this->filename;
 766         }
 767 }
 768
 769 /**
 770  * Stream outputter to send data to a file via some filter program.
 771  * Even if compression is available in a library, using a separate
 772  * program can allow us to make use of a multi-processor system.
 773  * @ingroup Dump
 774  */
 775 class DumpPipeOutput extends DumpFileOutput {
 776         protected $command, $filename;
 777
 778         function __construct( $command, $file = null ) {
 779                 if ( !is_null( $file ) ) {
 780                         $command .=  " > " . wfEscapeShellArg( $file );
 781                 }
 782
 783                 $this->startCommand( $command );
 784                 $this->command = $command;
 785                 $this->filename = $file;
 786         }
 787
 788         function startCommand( $command ) {
 789                 $spec = array(
 790                         0 => array( "pipe", "r" ),
 791                 );
 792                 $pipes = array();
 793                 $this->procOpenResource = proc_open( $command, $spec, $pipes );
 794                 $this->handle = $pipes[0];
 795         }
 796
 797         function closeRenameAndReopen( $newname ) {
 798                 $this->closeAndRename( $newname, true );
 799         }
 800
 801         function closeAndRename( $newname, $open = false ) {
 802                 if ( is_array( $newname ) ) {
 803                         if ( count( $newname ) > 1 ) {
 804                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
 805                         } else {
 806                                 $newname = $newname[0];
 807                         }
 808                 }
 809                 if ( $newname ) {
 810                         fclose( $this->handle );
 811                         proc_close( $this->procOpenResource );
 812                         if (! rename( $this->filename, $newname ) ) {
 813                                 throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" );
 814                         }
 815                         elseif ( $open ) {
 816                                 $command = $this->command;
 817                                 $command .=  " > " . wfEscapeShellArg( $this->filename );
 818                                 $this->startCommand( $command );
 819                         }
 820                 }
 821         }
 822
 823 }
 824
 825 /**
 826  * Sends dump output via the gzip compressor.
 827  * @ingroup Dump
 828  */
 829 class DumpGZipOutput extends DumpPipeOutput {
 830         function __construct( $file ) {
 831                 parent::__construct( "gzip", $file );
 832         }
 833 }
 834
 835 /**
 836  * Sends dump output via the bgzip2 compressor.
 837  * @ingroup Dump
 838  */
 839 class DumpBZip2Output extends DumpPipeOutput {
 840         function __construct( $file ) {
 841                 parent::__construct( "bzip2", $file );
 842         }
 843 }
 844
 845 /**
 846  * Sends dump output via the p7zip compressor.
 847  * @ingroup Dump
 848  */
 849 class Dump7ZipOutput extends DumpPipeOutput {
 850         protected $filename;
 851
 852         function __construct( $file ) {
 853                 $command = setup7zCommand( $file );
 854                 parent::__construct( $command );
 855                 $this->filename = $file;
 856         }
 857
 858         function closeRenameAndReopen( $newname ) {
 859                 $this->closeAndRename( $newname, true );
 860         }
 861
 862         function closeAndRename( $newname, $open = false ) {
 863                 if ( is_array( $newname ) ) {
 864                         if ( count( $newname ) > 1 ) {
 865                                 throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
 866                         } else {
 867                                 $newname = $newname[0];
 868                         }
 869                 }
 870                 if ( $newname ) {
 871                         fclose( $this->handle );
 872                         proc_close( $this->procOpenResource );
 873                         if (! rename( $this->filename, $newname ) ) {
 874                                 throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" );
 875                         }
 876                         elseif ( $open ) {
 877                                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 878                                 // Suppress annoying useless crap from p7zip
 879                                 // Unfortunately this could suppress real error messages too
 880                                 $command .= ' >' . wfGetNull() . ' 2>&1';
 881                                 $this->startCommand( $command );
 882                         }
 883                 }
 884         }
 885 }
 886
 887
 888
 889 /**
 890  * Dump output filter class.
 891  * This just does output filtering and streaming; XML formatting is done
 892  * higher up, so be careful in what you do.
 893  * @ingroup Dump
 894  */
 895 class DumpFilter {
 896         function __construct( &$sink ) {
 897                 $this->sink =& $sink;
 898         }
 899
 900         function writeOpenStream( $string ) {
 901                 $this->sink->writeOpenStream( $string );
 902         }
 903
 904         function writeCloseStream( $string ) {
 905                 $this->sink->writeCloseStream( $string );
 906         }
 907
 908         function writeOpenPage( $page, $string ) {
 909                 $this->sendingThisPage = $this->pass( $page, $string );
 910                 if ( $this->sendingThisPage ) {
 911                         $this->sink->writeOpenPage( $page, $string );
 912                 }
 913         }
 914
 915         function writeClosePage( $string ) {
 916                 if ( $this->sendingThisPage ) {
 917                         $this->sink->writeClosePage( $string );
 918                         $this->sendingThisPage = false;
 919                 }
 920         }
 921
 922         function writeRevision( $rev, $string ) {
 923                 if ( $this->sendingThisPage ) {
 924                         $this->sink->writeRevision( $rev, $string );
 925                 }
 926         }
 927
 928         function writeLogItem( $rev, $string ) {
 929                 $this->sink->writeRevision( $rev, $string );
 930         }
 931
 932         function closeRenameAndReopen( $newname ) {
 933                 $this->sink->closeRenameAndReopen( $newname );
 934         }
 935
 936         function closeAndRename( $newname, $open = false ) {
 937                 $this->sink->closeAndRename( $newname, $open );
 938         }
 939
 940         function getFilenames() {
 941                 return $this->sink->getFilenames();
 942         }
 943
 944         /**
 945          * Override for page-based filter types.
 946          * @return bool
 947          */
 948         function pass( $page ) {
 949                 return true;
 950         }
 951 }
 952
 953 /**
 954  * Simple dump output filter to exclude all talk pages.
 955  * @ingroup Dump
 956  */
 957 class DumpNotalkFilter extends DumpFilter {
 958         function pass( $page ) {
 959                 return !MWNamespace::isTalk( $page->page_namespace );
 960         }
 961 }
 962
 963 /**
 964  * Dump output filter to include or exclude pages in a given set of namespaces.
 965  * @ingroup Dump
 966  */
 967 class DumpNamespaceFilter extends DumpFilter {
 968         var $invert = false;
 969         var $namespaces = array();
 970
 971         function __construct( &$sink, $param ) {
 972                 parent::__construct( $sink );
 973
 974                 $constants = array(
 975                         "NS_MAIN"           => NS_MAIN,
 976                         "NS_TALK"           => NS_TALK,
 977                         "NS_USER"           => NS_USER,
 978                         "NS_USER_TALK"      => NS_USER_TALK,
 979                         "NS_PROJECT"        => NS_PROJECT,
 980                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
 981                         "NS_FILE"           => NS_FILE,
 982                         "NS_FILE_TALK"      => NS_FILE_TALK,
 983                         "NS_IMAGE"          => NS_IMAGE,  // NS_IMAGE is an alias for NS_FILE
 984                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
 985                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
 986                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
 987                         "NS_TEMPLATE"       => NS_TEMPLATE,
 988                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
 989                         "NS_HELP"           => NS_HELP,
 990                         "NS_HELP_TALK"      => NS_HELP_TALK,
 991                         "NS_CATEGORY"       => NS_CATEGORY,
 992                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
 993
 994                 if ( $param { 0 } == '!' ) {
 995                         $this->invert = true;
 996                         $param = substr( $param, 1 );
 997                 }
 998
 999                 foreach ( explode( ',', $param ) as $key ) {
1000                         $key = trim( $key );
1001                         if ( isset( $constants[$key] ) ) {
1002                                 $ns = $constants[$key];
1003                                 $this->namespaces[$ns] = true;
1004                         } elseif ( is_numeric( $key ) ) {
1005                                 $ns = intval( $key );
1006                                 $this->namespaces[$ns] = true;
1007                         } else {
1008                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
1009                         }
1010                 }
1011         }
1012
1013         function pass( $page ) {
1014                 $match = isset( $this->namespaces[$page->page_namespace] );
1015                 return $this->invert xor $match;
1016         }
1017 }
1018
1019
1020 /**
1021  * Dump output filter to include only the last revision in each page sequence.
1022  * @ingroup Dump
1023  */
1024 class DumpLatestFilter extends DumpFilter {
1025         var $page, $pageString, $rev, $revString;
1026
1027         function writeOpenPage( $page, $string ) {
1028                 $this->page = $page;
1029                 $this->pageString = $string;
1030         }
1031
1032         function writeClosePage( $string ) {
1033                 if ( $this->rev ) {
1034                         $this->sink->writeOpenPage( $this->page, $this->pageString );
1035                         $this->sink->writeRevision( $this->rev, $this->revString );
1036                         $this->sink->writeClosePage( $string );
1037                 }
1038                 $this->rev = null;
1039                 $this->revString = null;
1040                 $this->page = null;
1041                 $this->pageString = null;
1042         }
1043
1044         function writeRevision( $rev, $string ) {
1045                 if ( $rev->rev_id == $this->page->page_latest ) {
1046                         $this->rev = $rev;
1047                         $this->revString = $string;
1048                 }
1049         }
1050 }
1051
1052 /**
1053  * Base class for output stream; prints to stdout or buffer or whereever.
1054  * @ingroup Dump
1055  */
1056 class DumpMultiWriter {
1057         function __construct( $sinks ) {
1058                 $this->sinks = $sinks;
1059                 $this->count = count( $sinks );
1060         }
1061
1062         function writeOpenStream( $string ) {
1063                 for ( $i = 0; $i < $this->count; $i++ ) {
1064                         $this->sinks[$i]->writeOpenStream( $string );
1065                 }
1066         }
1067
1068         function writeCloseStream( $string ) {
1069                 for ( $i = 0; $i < $this->count; $i++ ) {
1070                         $this->sinks[$i]->writeCloseStream( $string );
1071                 }
1072         }
1073
1074         function writeOpenPage( $page, $string ) {
1075                 for ( $i = 0; $i < $this->count; $i++ ) {
1076                         $this->sinks[$i]->writeOpenPage( $page, $string );
1077                 }
1078         }
1079
1080         function writeClosePage( $string ) {
1081                 for ( $i = 0; $i < $this->count; $i++ ) {
1082                         $this->sinks[$i]->writeClosePage( $string );
1083                 }
1084         }
1085
1086         function writeRevision( $rev, $string ) {
1087                 for ( $i = 0; $i < $this->count; $i++ ) {
1088                         $this->sinks[$i]->writeRevision( $rev, $string );
1089                 }
1090         }
1091
1092         function closeRenameAndReopen( $newnames ) {
1093                 $this->closeAndRename( $newnames, true );
1094         }
1095
1096         function closeAndRename( $newnames, $open = false ) {
1097                 for ( $i = 0; $i < $this->count; $i++ ) {
1098                         $this->sinks[$i]->closeAndRename( $newnames[$i], $open );
1099                 }
1100         }
1101
1102         function getFilenames() {
1103                 $filenames = array();
1104                 for ( $i = 0; $i < $this->count; $i++ ) {
1105                         $filenames[] =  $this->sinks[$i]->getFilenames();
1106                 }
1107                 return $filenames;
1108         }
1109
1110 }
1111
1112 function xmlsafe( $string ) {
1113         wfProfileIn( __FUNCTION__ );
1114
1115         /**
1116          * The page may contain old data which has not been properly normalized.
1117          * Invalid UTF-8 sequences or forbidden control characters will make our
1118          * XML output invalid, so be sure to strip them out.
1119          */
1120         $string = UtfNormal::cleanUp( $string );
1121
1122         $string = htmlspecialchars( $string );
1123         wfProfileOut( __FUNCTION__ );
1124         return $string;
1125 }