maintenance/refreshLinks.inc

   1 <?php
   2 /**
   3  * @todo document
   4  * @package MediaWiki
   5  * @subpackage Maintenance
   6  */
   7
   8 /** */
   9 define( "REPORTING_INTERVAL", 100 );
  10 #define( "REPORTING_INTERVAL", 1 );
  11
  12 function refreshLinks( $start, $newOnly = false, $maxLag = false, $end = 0 ) {
  13         global $wgUser, $wgParser, $wgUseImageResize, $wgUseTidy;
  14
  15         $fname = 'refreshLinks';
  16         $dbr =& wfGetDB( DB_SLAVE );
  17         $dbw =& wfGetDB( DB_MASTER );
  18         $start = intval( $start );
  19
  20         # Don't generate TeX PNGs (lack of a sensible current directory causes errors anyway)
  21         $wgUser->setOption('math', MW_MATH_SOURCE);
  22
  23         # Don't generate extension images (e.g. Timeline)
  24         $wgParser->mTagHooks = array();
  25
  26         # Don't generate thumbnail images
  27         $wgUseImageResize = false;
  28         $wgUseTidy = false;
  29
  30         if ( $newOnly ) {
  31                 print "Refreshing links from ";
  32                 $res = $dbr->select( 'page',
  33                         array( 'page_id' ),
  34                         array(
  35                                 'page_is_new' => 1,
  36                                 "page_id > $start" ),
  37                         $fname
  38                 );
  39                 $num = $dbr->numRows( $res );
  40                 print "$num new articles...\n";
  41
  42                 $i = 0;
  43                 while ( $row = $dbr->fetchObject( $res ) ) {
  44                         if ( !( ++$i % REPORTING_INTERVAL ) ) {
  45                                 print "$i\n";
  46                                 wfWaitForSlaves( $maxLag );
  47                         }
  48
  49                         fixLinksFromArticle( $row->page_id );
  50                 }
  51         } else {
  52                 print "Refreshing link table.\n";
  53                 if ( !$end ) {
  54                         $end = $dbr->selectField( 'page', 'max(page_id)', false );
  55                 }
  56                 print("Starting from page_id $start of $end.\n");
  57
  58                 for ($id = $start; $id <= $end; $id++) {
  59
  60                         if ( !($id % REPORTING_INTERVAL) ) {
  61                                 print "$id\n";
  62                                 wfWaitForSlaves( $maxLag );
  63                         }
  64                         fixLinksFromArticle( $id );
  65                 }
  66
  67
  68         }
  69 }
  70
  71 function fixLinksFromArticle( $id ) {
  72         global $wgTitle, $wgArticle, $wgOut, $wgParser;
  73
  74         $wgTitle = Title::newFromID( $id );
  75         $dbw =& wfGetDB( DB_MASTER );
  76
  77         $linkCache =& LinkCache::singleton();
  78         $linkCache->clear();
  79
  80         if ( is_null( $wgTitle ) ) {
  81                 return;
  82         }
  83         $dbw->begin();
  84
  85         $revision = Revision::newFromTitle( $wgTitle );
  86         if ( !$revision ) {
  87                 return;
  88         }
  89
  90         $options = new ParserOptions;
  91         $parserOutput = $wgParser->parse( $revision->getText(), $wgTitle, $options, true, true, $revision->getId() );
  92         $update = new LinksUpdate( $wgTitle, $parserOutput );
  93         $update->doUpdate();
  94         $dbw->immediateCommit();
  95 }
  96
  97 function deleteLinksFromNonexistent( $maxLag = 0 ) {
  98         $fname = 'deleteLinksFromNonexistent';
  99
 100         wfWaitForSlaves( $maxLag );
 101
 102         $dbw =& wfGetDB( DB_WRITE );
 103
 104         $linksTables = array(
 105                 'pagelinks' => 'pl_from',
 106                 'imagelinks' => 'il_from',
 107                 'categorylinks' => 'cl_from',
 108                 'templatelinks' => 'tl_from',
 109                 'externallinks' => 'el_from',
 110         );
 111
 112         $page = $dbw->tableName( 'page' );
 113
 114
 115         foreach ( $linksTables as $table => $field ) {
 116                 if ( !$dbw->ping() ) {
 117                         print "DB disconnected, reconnecting...";
 118                         while ( !$dbw->ping() ) {
 119                                 print ".";
 120                                 sleep(10);
 121                         }
 122                         print "\n";
 123                 }
 124
 125                 $pTable = $dbw->tableName( $table );
 126                 $sql = "DELETE $pTable FROM $pTable LEFT JOIN $page ON page_id=$field WHERE page_id IS NULL";
 127
 128                 print "Deleting $table from non-existent articles...";
 129                 $dbw->query( $sql, $fname );
 130                 print " fixed " .$dbw->affectedRows() . " row(s)\n";
 131         }
 132 }
 133
 134 ?>