includes/Export.php

   1 <?php
   2 # Copyright (C) 2003, 2005, 2006 Brion Vibber <brion@pobox.com>
   3 # http://www.mediawiki.org/
   4 #
   5 # This program is free software; you can redistribute it and/or modify
   6 # it under the terms of the GNU General Public License as published by
   7 # the Free Software Foundation; either version 2 of the License, or
   8 # (at your option) any later version.
   9 #
  10 # This program is distributed in the hope that it will be useful,
  11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13 # GNU General Public License for more details.
  14 #
  15 # You should have received a copy of the GNU General Public License along
  16 # with this program; if not, write to the Free Software Foundation, Inc.,
  17 # 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18 # http://www.gnu.org/copyleft/gpl.html
  19
  20 /**
  21  * @defgroup Dump Dump
  22  */
  23
  24 /**
  25  * @ingroup SpecialPage Dump
  26  */
  27 class WikiExporter {
  28         var $list_authors = false ; # Return distinct author list (when not returning full history)
  29         var $author_list = "" ;
  30
  31         var $dumpUploads = false;
  32
  33         const FULL = 0;
  34         const CURRENT = 1;
  35         const LOGS = 2;
  36
  37         const BUFFER = 0;
  38         const STREAM = 1;
  39
  40         const TEXT = 0;
  41         const STUB = 1;
  42
  43         /**
  44          * If using WikiExporter::STREAM to stream a large amount of data,
  45          * provide a database connection which is not managed by
  46          * LoadBalancer to read from: some history blob types will
  47          * make additional queries to pull source data while the
  48          * main query is still running.
  49          *
  50          * @param $db Database
  51          * @param $history Mixed: one of WikiExporter::FULL or WikiExporter::CURRENT,
  52          *                 or an associative array:
  53          *                   offset: non-inclusive offset at which to start the query
  54          *                   limit: maximum number of rows to return
  55          *                   dir: "asc" or "desc" timestamp order
  56          * @param $buffer Int: one of WikiExporter::BUFFER or WikiExporter::STREAM
  57          */
  58         function __construct( &$db, $history = WikiExporter::CURRENT,
  59                         $buffer = WikiExporter::BUFFER, $text = WikiExporter::TEXT ) {
  60                 $this->db =& $db;
  61                 $this->history = $history;
  62                 $this->buffer  = $buffer;
  63                 $this->writer  = new XmlDumpWriter();
  64                 $this->sink    = new DumpOutput();
  65                 $this->text    = $text;
  66         }
  67
  68         /**
  69          * Set the DumpOutput or DumpFilter object which will receive
  70          * various row objects and XML output for filtering. Filters
  71          * can be chained or used as callbacks.
  72          *
  73          * @param $sink mixed
  74          */
  75         public function setOutputSink( &$sink ) {
  76                 $this->sink =& $sink;
  77         }
  78
  79         public function openStream() {
  80                 $output = $this->writer->openStream();
  81                 $this->sink->writeOpenStream( $output );
  82         }
  83
  84         public function closeStream() {
  85                 $output = $this->writer->closeStream();
  86                 $this->sink->writeCloseStream( $output );
  87         }
  88
  89         /**
  90          * Dumps a series of page and revision records for all pages
  91          * in the database, either including complete history or only
  92          * the most recent version.
  93          */
  94         public function allPages() {
  95                 return $this->dumpFrom( '' );
  96         }
  97
  98         /**
  99          * Dumps a series of page and revision records for those pages
 100          * in the database falling within the page_id range given.
 101          * @param $start Int: inclusive lower limit (this id is included)
 102          * @param $end   Int: Exclusive upper limit (this id is not included)
 103          *                   If 0, no upper limit.
 104          */
 105         public function pagesByRange( $start, $end ) {
 106                 $condition = 'page_id >= ' . intval( $start );
 107                 if( $end ) {
 108                         $condition .= ' AND page_id < ' . intval( $end );
 109                 }
 110                 return $this->dumpFrom( $condition );
 111         }
 112
 113         /**
 114          * @param $title Title
 115          */
 116         public function pageByTitle( $title ) {
 117                 return $this->dumpFrom(
 118                         'page_namespace=' . $title->getNamespace() .
 119                         ' AND page_title=' . $this->db->addQuotes( $title->getDBkey() ) );
 120         }
 121
 122         public function pageByName( $name ) {
 123                 $title = Title::newFromText( $name );
 124                 if( is_null( $title ) ) {
 125                         return new WikiError( "Can't export invalid title" );
 126                 } else {
 127                         return $this->pageByTitle( $title );
 128                 }
 129         }
 130
 131         public function pagesByName( $names ) {
 132                 foreach( $names as $name ) {
 133                         $this->pageByName( $name );
 134                 }
 135         }
 136
 137         public function allLogs() {
 138                 return $this->dumpFrom( '' );
 139         }
 140
 141         public function logsByRange( $start, $end ) {
 142                 $condition = 'log_id >= ' . intval( $start );
 143                 if( $end ) {
 144                         $condition .= ' AND log_id < ' . intval( $end );
 145                 }
 146                 return $this->dumpFrom( $condition );
 147         }
 148
 149         # Generates the distinct list of authors of an article
 150         # Not called by default (depends on $this->list_authors)
 151         # Can be set by Special:Export when not exporting whole history
 152         protected function do_list_authors( $page , $revision , $cond ) {
 153                 $fname = "do_list_authors" ;
 154                 wfProfileIn( $fname );
 155                 $this->author_list = "<contributors>";
 156                 //rev_deleted
 157                 $nothidden = '(rev_deleted & '.Revision::DELETED_USER.') = 0';
 158
 159                 $sql = "SELECT DISTINCT rev_user_text,rev_user FROM {$page},{$revision}
 160                 WHERE page_id=rev_page AND $nothidden AND " . $cond ;
 161                 $result = $this->db->query( $sql, $fname );
 162                 $resultset = $this->db->resultObject( $result );
 163                 while( $row = $resultset->fetchObject() ) {
 164                         $this->author_list .= "<contributor>" .
 165                                 "<username>" .
 166                                 htmlentities( $row->rev_user_text )  .
 167                                 "</username>" .
 168                                 "<id>" .
 169                                 $row->rev_user .
 170                                 "</id>" .
 171                                 "</contributor>";
 172                 }
 173                 wfProfileOut( $fname );
 174                 $this->author_list .= "</contributors>";
 175         }
 176
 177         protected function dumpFrom( $cond = '' ) {
 178                 $fname = 'WikiExporter::dumpFrom';
 179                 wfProfileIn( $fname );
 180
 181                 # For logs dumps...
 182                 if( $this->history & self::LOGS ) {
 183                         $where = array( 'user_id = log_user' );
 184                         # Hide private logs
 185                         $where[] = LogEventsList::getExcludeClause( $this->db );
 186                         if( $cond ) $where[] = $cond;
 187                         # Get logging table name for logging.* clause
 188                         $logging = $this->db->tableName('logging');
 189                         $result = $this->db->select( array('logging','user'),
 190                                 array( "{$logging}.*", 'user_name' ), // grab the user name
 191                                 $where,
 192                                 $fname,
 193                                 array( 'ORDER BY' => 'log_id', 'USE INDEX' => array('logging' => 'PRIMARY') )
 194                         );
 195                         $wrapper = $this->db->resultObject( $result );
 196                         $this->outputLogStream( $wrapper );
 197                 # For page dumps...
 198                 } else {
 199                         list($page,$revision,$text) = $this->db->tableNamesN('page','revision','text');
 200
 201                         $order = 'ORDER BY page_id';
 202                         $limit = '';
 203
 204                         if( $this->history == WikiExporter::FULL ) {
 205                                 $join = 'page_id=rev_page';
 206                         } elseif( $this->history == WikiExporter::CURRENT ) {
 207                                 if ( $this->list_authors && $cond != '' )  { // List authors, if so desired
 208                                         $this->do_list_authors ( $page , $revision , $cond );
 209                                 }
 210                                 $join = 'page_id=rev_page AND page_latest=rev_id';
 211                         } elseif ( is_array( $this->history ) ) {
 212                                 $join = 'page_id=rev_page';
 213                                 if ( $this->history['dir'] == 'asc' ) {
 214                                         $op = '>';
 215                                         $order .= ', rev_timestamp';
 216                                 } else {
 217                                         $op = '<';
 218                                         $order .= ', rev_timestamp DESC';
 219                                 }
 220                                 if ( !empty( $this->history['offset'] ) ) {
 221                                         $join .= " AND rev_timestamp $op " . $this->db->addQuotes(
 222                                                 $this->db->timestamp( $this->history['offset'] ) );
 223                                 }
 224                                 if ( !empty( $this->history['limit'] ) ) {
 225                                         $limitNum = intval( $this->history['limit'] );
 226                                         if ( $limitNum > 0 ) {
 227                                                 $limit = "LIMIT $limitNum";
 228                                         }
 229                                 }
 230                         } else {
 231                                 wfProfileOut( $fname );
 232                                 return new WikiError( "$fname given invalid history dump type." );
 233                         }
 234                         $where = ( $cond == '' ) ? '' : "$cond AND";
 235
 236                         if( $this->buffer == WikiExporter::STREAM ) {
 237                                 $prev = $this->db->bufferResults( false );
 238                         }
 239                         if( $cond == '' ) {
 240                                 // Optimization hack for full-database dump
 241                                 $revindex = $pageindex = $this->db->useIndexClause("PRIMARY");
 242                                 $straight = ' /*! STRAIGHT_JOIN */ ';
 243                         } else {
 244                                 $pageindex = '';
 245                                 $revindex = '';
 246                                 $straight = '';
 247                         }
 248                         if( $this->text == WikiExporter::STUB ) {
 249                                 $sql = "SELECT $straight * FROM
 250                                         $page $pageindex,
 251                                         $revision $revindex
 252                                         WHERE $where $join
 253                                         $order $limit";
 254                         } else {
 255                                 $sql = "SELECT $straight * FROM
 256                                         $page $pageindex,
 257                                         $revision $revindex,
 258                                         $text
 259                                         WHERE $where $join AND rev_text_id=old_id
 260                                         $order $limit";
 261                         }
 262                         $result = $this->db->query( $sql, $fname );
 263                         $wrapper = $this->db->resultObject( $result );
 264                         $this->outputPageStream( $wrapper );
 265
 266                         if ( $this->list_authors ) {
 267                                 $this->outputPageStream( $wrapper );
 268                         }
 269
 270                         if( $this->buffer == WikiExporter::STREAM ) {
 271                                 $this->db->bufferResults( $prev );
 272                         }
 273                 }
 274                 wfProfileOut( $fname );
 275         }
 276
 277         /**
 278          * Runs through a query result set dumping page and revision records.
 279          * The result set should be sorted/grouped by page to avoid duplicate
 280          * page records in the output.
 281          *
 282          * The result set will be freed once complete. Should be safe for
 283          * streaming (non-buffered) queries, as long as it was made on a
 284          * separate database connection not managed by LoadBalancer; some
 285          * blob storage types will make queries to pull source data.
 286          *
 287          * @param $resultset ResultWrapper
 288          */
 289         protected function outputPageStream( $resultset ) {
 290                 $last = null;
 291                 while( $row = $resultset->fetchObject() ) {
 292                         if( is_null( $last ) ||
 293                                 $last->page_namespace != $row->page_namespace ||
 294                                 $last->page_title     != $row->page_title ) {
 295                                 if( isset( $last ) ) {
 296                                         $output = '';
 297                                         if( $this->dumpUploads ) {
 298                                                 $output .= $this->writer->writeUploads( $last );
 299                                         }
 300                                         $output .= $this->writer->closePage();
 301                                         $this->sink->writeClosePage( $output );
 302                                 }
 303                                 $output = $this->writer->openPage( $row );
 304                                 $this->sink->writeOpenPage( $row, $output );
 305                                 $last = $row;
 306                         }
 307                         $output = $this->writer->writeRevision( $row );
 308                         $this->sink->writeRevision( $row, $output );
 309                 }
 310                 if( isset( $last ) ) {
 311                         $output = '';
 312                         if( $this->dumpUploads ) {
 313                                 $output .= $this->writer->writeUploads( $last );
 314                         }
 315                         $output .= $this->author_list;
 316                         $output .= $this->writer->closePage();
 317                         $this->sink->writeClosePage( $output );
 318                 }
 319                 $resultset->free();
 320         }
 321
 322         protected function outputLogStream( $resultset ) {
 323                 while( $row = $resultset->fetchObject() ) {
 324                         $output = $this->writer->writeLogItem( $row );
 325                         $this->sink->writeLogItem( $row, $output );
 326                 }
 327                 $resultset->free();
 328         }
 329 }
 330
 331 /**
 332  * @ingroup Dump
 333  */
 334 class XmlDumpWriter {
 335
 336         /**
 337          * Returns the export schema version.
 338          * @return string
 339          */
 340         function schemaVersion() {
 341                 return "0.3"; // FIXME: upgrade to 0.4 when updated XSD is ready, for the revision deletion bits
 342         }
 343
 344         /**
 345          * Opens the XML output stream's root <mediawiki> element.
 346          * This does not include an xml directive, so is safe to include
 347          * as a subelement in a larger XML stream. Namespace and XML Schema
 348          * references are included.
 349          *
 350          * Output will be encoded in UTF-8.
 351          *
 352          * @return string
 353          */
 354         function openStream() {
 355                 global $wgContLanguageCode;
 356                 $ver = $this->schemaVersion();
 357                 return Xml::element( 'mediawiki', array(
 358                         'xmlns'              => "http://www.mediawiki.org/xml/export-$ver/",
 359                         'xmlns:xsi'          => "http://www.w3.org/2001/XMLSchema-instance",
 360                         'xsi:schemaLocation' => "http://www.mediawiki.org/xml/export-$ver/ " .
 361                                                 "http://www.mediawiki.org/xml/export-$ver.xsd",
 362                         'version'            => $ver,
 363                         'xml:lang'           => $wgContLanguageCode ),
 364                         null ) .
 365                         "\n" .
 366                         $this->siteInfo();
 367         }
 368
 369         function siteInfo() {
 370                 $info = array(
 371                         $this->sitename(),
 372                         $this->homelink(),
 373                         $this->generator(),
 374                         $this->caseSetting(),
 375                         $this->namespaces() );
 376                 return "  <siteinfo>\n    " .
 377                         implode( "\n    ", $info ) .
 378                         "\n  </siteinfo>\n";
 379         }
 380
 381         function sitename() {
 382                 global $wgSitename;
 383                 return Xml::element( 'sitename', array(), $wgSitename );
 384         }
 385
 386         function generator() {
 387                 global $wgVersion;
 388                 return Xml::element( 'generator', array(), "MediaWiki $wgVersion" );
 389         }
 390
 391         function homelink() {
 392                 return Xml::element( 'base', array(), Title::newMainPage()->getFullUrl() );
 393         }
 394
 395         function caseSetting() {
 396                 global $wgCapitalLinks;
 397                 // "case-insensitive" option is reserved for future
 398                 $sensitivity = $wgCapitalLinks ? 'first-letter' : 'case-sensitive';
 399                 return Xml::element( 'case', array(), $sensitivity );
 400         }
 401
 402         function namespaces() {
 403                 global $wgContLang;
 404                 $spaces = "<namespaces>\n";
 405                 foreach( $wgContLang->getFormattedNamespaces() as $ns => $title ) {
 406                         $spaces .= '      ' . Xml::element( 'namespace', array( 'key' => $ns ), $title ) . "\n";
 407                 }
 408                 $spaces .= "    </namespaces>";
 409                 return $spaces;
 410         }
 411
 412         /**
 413          * Closes the output stream with the closing root element.
 414          * Call when finished dumping things.
 415          */
 416         function closeStream() {
 417                 return "</mediawiki>\n";
 418         }
 419
 420
 421         /**
 422          * Opens a <page> section on the output stream, with data
 423          * from the given database row.
 424          *
 425          * @param $row object
 426          * @return string
 427          * @access private
 428          */
 429         function openPage( $row ) {
 430                 $out = "  <page>\n";
 431                 $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 432                 $out .= '    ' . Xml::elementClean( 'title', array(), $title->getPrefixedText() ) . "\n";
 433                 $out .= '    ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
 434                 if( '' != $row->page_restrictions ) {
 435                         $out .= '    ' . Xml::element( 'restrictions', array(),
 436                                 strval( $row->page_restrictions ) ) . "\n";
 437                 }
 438                 return $out;
 439         }
 440
 441         /**
 442          * Closes a <page> section on the output stream.
 443          *
 444          * @access private
 445          */
 446         function closePage() {
 447                 return "  </page>\n";
 448         }
 449
 450         /**
 451          * Dumps a <revision> section on the output stream, with
 452          * data filled in from the given database row.
 453          *
 454          * @param $row object
 455          * @return string
 456          * @access private
 457          */
 458         function writeRevision( $row ) {
 459                 $fname = 'WikiExporter::dumpRev';
 460                 wfProfileIn( $fname );
 461
 462                 $out  = "    <revision>\n";
 463                 $out .= "      " . Xml::element( 'id', null, strval( $row->rev_id ) ) . "\n";
 464
 465                 $out .= $this->writeTimestamp( $row->rev_timestamp );
 466
 467                 if( $row->rev_deleted & Revision::DELETED_USER ) {
 468                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 469                 } else {
 470                         $out .= $this->writeContributor( $row->rev_user, $row->rev_user_text );
 471                 }
 472
 473                 if( $row->rev_minor_edit ) {
 474                         $out .=  "      <minor/>\n";
 475                 }
 476                 if( $row->rev_deleted & Revision::DELETED_COMMENT ) {
 477                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 478                 } elseif( $row->rev_comment != '' ) {
 479                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->rev_comment ) ) . "\n";
 480                 }
 481
 482                 if( $row->rev_deleted & Revision::DELETED_TEXT ) {
 483                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 484                 } elseif( isset( $row->old_text ) ) {
 485                         // Raw text from the database may have invalid chars
 486                         $text = strval( Revision::getRevisionText( $row ) );
 487                         $out .= "      " . Xml::elementClean( 'text',
 488                                 array( 'xml:space' => 'preserve' ),
 489                                 strval( $text ) ) . "\n";
 490                 } else {
 491                         // Stub output
 492                         $out .= "      " . Xml::element( 'text',
 493                                 array( 'id' => $row->rev_text_id ),
 494                                 "" ) . "\n";
 495                 }
 496
 497                 $out .= "    </revision>\n";
 498
 499                 wfProfileOut( $fname );
 500                 return $out;
 501         }
 502
 503         /**
 504          * Dumps a <logitem> section on the output stream, with
 505          * data filled in from the given database row.
 506          *
 507          * @param $row object
 508          * @return string
 509          * @access private
 510          */
 511         function writeLogItem( $row ) {
 512                 $fname = 'WikiExporter::writeLogItem';
 513                 wfProfileIn( $fname );
 514
 515                 $out  = "    <logitem>\n";
 516                 $out .= "      " . Xml::element( 'id', null, strval( $row->log_id ) ) . "\n";
 517
 518                 $out .= $this->writeTimestamp( $row->log_timestamp );
 519
 520                 if( $row->log_deleted & LogPage::DELETED_USER ) {
 521                         $out .= "      " . Xml::element( 'contributor', array( 'deleted' => 'deleted' ) ) . "\n";
 522                 } else {
 523                         $out .= $this->writeContributor( $row->log_user, $row->user_name );
 524                 }
 525
 526                 if( $row->log_deleted & LogPage::DELETED_COMMENT ) {
 527                         $out .= "      " . Xml::element( 'comment', array( 'deleted' => 'deleted' ) ) . "\n";
 528                 } elseif( $row->log_comment != '' ) {
 529                         $out .= "      " . Xml::elementClean( 'comment', null, strval( $row->log_comment ) ) . "\n";
 530                 }
 531
 532                 $out .= "      " . Xml::element( 'type', null, strval( $row->log_type ) ) . "\n";
 533                 $out .= "      " . Xml::element( 'action', null, strval( $row->log_action ) ) . "\n";
 534
 535                 if( $row->log_deleted & LogPage::DELETED_ACTION ) {
 536                         $out .= "      " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
 537                 } else {
 538                         $title = Title::makeTitle( $row->log_namespace, $row->log_title );
 539                         $out .= "      " . Xml::elementClean( 'logtitle', null, $title->getPrefixedText() ) . "\n";
 540                         $out .= "      " . Xml::elementClean( 'params',
 541                                 array( 'xml:space' => 'preserve' ),
 542                                 strval( $row->log_params ) ) . "\n";
 543                 }
 544
 545                 $out .= "    </logitem>\n";
 546
 547                 wfProfileOut( $fname );
 548                 return $out;
 549         }
 550
 551         function writeTimestamp( $timestamp ) {
 552                 $ts = wfTimestamp( TS_ISO_8601, $timestamp );
 553                 return "      " . Xml::element( 'timestamp', null, $ts ) . "\n";
 554         }
 555
 556         function writeContributor( $id, $text ) {
 557                 $out = "      <contributor>\n";
 558                 if( $id ) {
 559                         $out .= "        " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
 560                         $out .= "        " . Xml::element( 'id', null, strval( $id ) ) . "\n";
 561                 } else {
 562                         $out .= "        " . Xml::elementClean( 'ip', null, strval( $text ) ) . "\n";
 563                 }
 564                 $out .= "      </contributor>\n";
 565                 return $out;
 566         }
 567
 568         /**
 569          * Warning! This data is potentially inconsistent. :(
 570          */
 571         function writeUploads( $row ) {
 572                 if( $row->page_namespace == NS_IMAGE ) {
 573                         $img = wfFindFile( $row->page_title );
 574                         if( $img ) {
 575                                 $out = '';
 576                                 foreach( array_reverse( $img->getHistory() ) as $ver ) {
 577                                         $out .= $this->writeUpload( $ver );
 578                                 }
 579                                 $out .= $this->writeUpload( $img );
 580                                 return $out;
 581                         }
 582                 }
 583                 return '';
 584         }
 585
 586         function writeUpload( $file ) {
 587                 return "    <upload>\n" .
 588                         $this->writeTimestamp( $file->getTimestamp() ) .
 589                         $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
 590                         "      " . Xml::elementClean( 'comment', null, $file->getDescription() ) . "\n" .
 591                         "      " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
 592                         "      " . Xml::element( 'src', null, $file->getFullUrl() ) . "\n" .
 593                         "      " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
 594                         "    </upload>\n";
 595         }
 596
 597 }
 598
 599
 600 /**
 601  * Base class for output stream; prints to stdout or buffer or whereever.
 602  * @ingroup Dump
 603  */
 604 class DumpOutput {
 605         function writeOpenStream( $string ) {
 606                 $this->write( $string );
 607         }
 608
 609         function writeCloseStream( $string ) {
 610                 $this->write( $string );
 611         }
 612
 613         function writeOpenPage( $page, $string ) {
 614                 $this->write( $string );
 615         }
 616
 617         function writeClosePage( $string ) {
 618                 $this->write( $string );
 619         }
 620
 621         function writeRevision( $rev, $string ) {
 622                 $this->write( $string );
 623         }
 624
 625         function writeLogItem( $rev, $string ) {
 626                 $this->write( $string );
 627         }
 628
 629         /**
 630          * Override to write to a different stream type.
 631          * @return bool
 632          */
 633         function write( $string ) {
 634                 print $string;
 635         }
 636 }
 637
 638 /**
 639  * Stream outputter to send data to a file.
 640  * @ingroup Dump
 641  */
 642 class DumpFileOutput extends DumpOutput {
 643         var $handle;
 644
 645         function DumpFileOutput( $file ) {
 646                 $this->handle = fopen( $file, "wt" );
 647         }
 648
 649         function write( $string ) {
 650                 fputs( $this->handle, $string );
 651         }
 652 }
 653
 654 /**
 655  * Stream outputter to send data to a file via some filter program.
 656  * Even if compression is available in a library, using a separate
 657  * program can allow us to make use of a multi-processor system.
 658  * @ingroup Dump
 659  */
 660 class DumpPipeOutput extends DumpFileOutput {
 661         function DumpPipeOutput( $command, $file = null ) {
 662                 if( !is_null( $file ) ) {
 663                         $command .=  " > " . wfEscapeShellArg( $file );
 664                 }
 665                 $this->handle = popen( $command, "w" );
 666         }
 667 }
 668
 669 /**
 670  * Sends dump output via the gzip compressor.
 671  * @ingroup Dump
 672  */
 673 class DumpGZipOutput extends DumpPipeOutput {
 674         function DumpGZipOutput( $file ) {
 675                 parent::DumpPipeOutput( "gzip", $file );
 676         }
 677 }
 678
 679 /**
 680  * Sends dump output via the bgzip2 compressor.
 681  * @ingroup Dump
 682  */
 683 class DumpBZip2Output extends DumpPipeOutput {
 684         function DumpBZip2Output( $file ) {
 685                 parent::DumpPipeOutput( "bzip2", $file );
 686         }
 687 }
 688
 689 /**
 690  * Sends dump output via the p7zip compressor.
 691  * @ingroup Dump
 692  */
 693 class Dump7ZipOutput extends DumpPipeOutput {
 694         function Dump7ZipOutput( $file ) {
 695                 $command = "7za a -bd -si " . wfEscapeShellArg( $file );
 696                 // Suppress annoying useless crap from p7zip
 697                 // Unfortunately this could suppress real error messages too
 698                 $command .= ' >' . wfGetNull() . ' 2>&1';
 699                 parent::DumpPipeOutput( $command );
 700         }
 701 }
 702
 703
 704
 705 /**
 706  * Dump output filter class.
 707  * This just does output filtering and streaming; XML formatting is done
 708  * higher up, so be careful in what you do.
 709  * @ingroup Dump
 710  */
 711 class DumpFilter {
 712         function DumpFilter( &$sink ) {
 713                 $this->sink =& $sink;
 714         }
 715
 716         function writeOpenStream( $string ) {
 717                 $this->sink->writeOpenStream( $string );
 718         }
 719
 720         function writeCloseStream( $string ) {
 721                 $this->sink->writeCloseStream( $string );
 722         }
 723
 724         function writeOpenPage( $page, $string ) {
 725                 $this->sendingThisPage = $this->pass( $page, $string );
 726                 if( $this->sendingThisPage ) {
 727                         $this->sink->writeOpenPage( $page, $string );
 728                 }
 729         }
 730
 731         function writeClosePage( $string ) {
 732                 if( $this->sendingThisPage ) {
 733                         $this->sink->writeClosePage( $string );
 734                         $this->sendingThisPage = false;
 735                 }
 736         }
 737
 738         function writeRevision( $rev, $string ) {
 739                 if( $this->sendingThisPage ) {
 740                         $this->sink->writeRevision( $rev, $string );
 741                 }
 742         }
 743
 744         function writeLogItem( $rev, $string ) {
 745                 $this->sink->writeRevision( $rev, $string );
 746         }
 747
 748         /**
 749          * Override for page-based filter types.
 750          * @return bool
 751          */
 752         function pass( $page ) {
 753                 return true;
 754         }
 755 }
 756
 757 /**
 758  * Simple dump output filter to exclude all talk pages.
 759  * @ingroup Dump
 760  */
 761 class DumpNotalkFilter extends DumpFilter {
 762         function pass( $page ) {
 763                 return !MWNamespace::isTalk( $page->page_namespace );
 764         }
 765 }
 766
 767 /**
 768  * Dump output filter to include or exclude pages in a given set of namespaces.
 769  * @ingroup Dump
 770  */
 771 class DumpNamespaceFilter extends DumpFilter {
 772         var $invert = false;
 773         var $namespaces = array();
 774
 775         function DumpNamespaceFilter( &$sink, $param ) {
 776                 parent::DumpFilter( $sink );
 777
 778                 $constants = array(
 779                         "NS_MAIN"           => NS_MAIN,
 780                         "NS_TALK"           => NS_TALK,
 781                         "NS_USER"           => NS_USER,
 782                         "NS_USER_TALK"      => NS_USER_TALK,
 783                         "NS_PROJECT"        => NS_PROJECT,
 784                         "NS_PROJECT_TALK"   => NS_PROJECT_TALK,
 785                         "NS_FILE"           => NS_FILE,
 786                         "NS_FILE_TALK"      => NS_FILE_TALK,
 787                         "NS_IMAGE"          => NS_IMAGE,  // NS_IMAGE is an alias for NS_FILE
 788                         "NS_IMAGE_TALK"     => NS_IMAGE_TALK,
 789                         "NS_MEDIAWIKI"      => NS_MEDIAWIKI,
 790                         "NS_MEDIAWIKI_TALK" => NS_MEDIAWIKI_TALK,
 791                         "NS_TEMPLATE"       => NS_TEMPLATE,
 792                         "NS_TEMPLATE_TALK"  => NS_TEMPLATE_TALK,
 793                         "NS_HELP"           => NS_HELP,
 794                         "NS_HELP_TALK"      => NS_HELP_TALK,
 795                         "NS_CATEGORY"       => NS_CATEGORY,
 796                         "NS_CATEGORY_TALK"  => NS_CATEGORY_TALK );
 797
 798                 if( $param{0} == '!' ) {
 799                         $this->invert = true;
 800                         $param = substr( $param, 1 );
 801                 }
 802
 803                 foreach( explode( ',', $param ) as $key ) {
 804                         $key = trim( $key );
 805                         if( isset( $constants[$key] ) ) {
 806                                 $ns = $constants[$key];
 807                                 $this->namespaces[$ns] = true;
 808                         } elseif( is_numeric( $key ) ) {
 809                                 $ns = intval( $key );
 810                                 $this->namespaces[$ns] = true;
 811                         } else {
 812                                 throw new MWException( "Unrecognized namespace key '$key'\n" );
 813                         }
 814                 }
 815         }
 816
 817         function pass( $page ) {
 818                 $match = isset( $this->namespaces[$page->page_namespace] );
 819                 return $this->invert xor $match;
 820         }
 821 }
 822
 823
 824 /**
 825  * Dump output filter to include only the last revision in each page sequence.
 826  * @ingroup Dump
 827  */
 828 class DumpLatestFilter extends DumpFilter {
 829         var $page, $pageString, $rev, $revString;
 830
 831         function writeOpenPage( $page, $string ) {
 832                 $this->page = $page;
 833                 $this->pageString = $string;
 834         }
 835
 836         function writeClosePage( $string ) {
 837                 if( $this->rev ) {
 838                         $this->sink->writeOpenPage( $this->page, $this->pageString );
 839                         $this->sink->writeRevision( $this->rev, $this->revString );
 840                         $this->sink->writeClosePage( $string );
 841                 }
 842                 $this->rev = null;
 843                 $this->revString = null;
 844                 $this->page = null;
 845                 $this->pageString = null;
 846         }
 847
 848         function writeRevision( $rev, $string ) {
 849                 if( $rev->rev_id == $this->page->page_latest ) {
 850                         $this->rev = $rev;
 851                         $this->revString = $string;
 852                 }
 853         }
 854 }
 855
 856 /**
 857  * Base class for output stream; prints to stdout or buffer or whereever.
 858  * @ingroup Dump
 859  */
 860 class DumpMultiWriter {
 861         function DumpMultiWriter( $sinks ) {
 862                 $this->sinks = $sinks;
 863                 $this->count = count( $sinks );
 864         }
 865
 866         function writeOpenStream( $string ) {
 867                 for( $i = 0; $i < $this->count; $i++ ) {
 868                         $this->sinks[$i]->writeOpenStream( $string );
 869                 }
 870         }
 871
 872         function writeCloseStream( $string ) {
 873                 for( $i = 0; $i < $this->count; $i++ ) {
 874                         $this->sinks[$i]->writeCloseStream( $string );
 875                 }
 876         }
 877
 878         function writeOpenPage( $page, $string ) {
 879                 for( $i = 0; $i < $this->count; $i++ ) {
 880                         $this->sinks[$i]->writeOpenPage( $page, $string );
 881                 }
 882         }
 883
 884         function writeClosePage( $string ) {
 885                 for( $i = 0; $i < $this->count; $i++ ) {
 886                         $this->sinks[$i]->writeClosePage( $string );
 887                 }
 888         }
 889
 890         function writeRevision( $rev, $string ) {
 891                 for( $i = 0; $i < $this->count; $i++ ) {
 892                         $this->sinks[$i]->writeRevision( $rev, $string );
 893                 }
 894         }
 895 }
 896
 897 function xmlsafe( $string ) {
 898         $fname = 'xmlsafe';
 899         wfProfileIn( $fname );
 900
 901         /**
 902          * The page may contain old data which has not been properly normalized.
 903          * Invalid UTF-8 sequences or forbidden control characters will make our
 904          * XML output invalid, so be sure to strip them out.
 905          */
 906         $string = UtfNormal::cleanUp( $string );
 907
 908         $string = htmlspecialchars( $string );
 909         wfProfileOut( $fname );
 910         return $string;
 911 }