maintenance/refreshLinks.inc

   1 <?php
   2 /**
   3  * @todo document
   4  * @package MediaWiki
   5  * @subpackage Maintenance
   6  */
   7
   8 /** */
   9 define( "REPORTING_INTERVAL", 100 );
  10
  11 function refreshLinks( $start, $newOnly = false, $maxLag = false ) {
  12         global $wgUser, $wgParser, $wgUseImageResize;
  13
  14         $fname = 'refreshLinks';
  15         $dbr =& wfGetDB( DB_SLAVE );
  16         $dbw =& wfGetDB( DB_MASTER );
  17         $start = intval( $start );
  18
  19         # Don't generate TeX PNGs (lack of a sensible current directory causes errors anyway)
  20         $wgUser->setOption('math', MW_MATH_SOURCE);
  21
  22         # Don't generate extension images (e.g. Timeline)
  23         $wgParser->mTagHooks = array();
  24
  25         # Don't generate thumbnail images
  26         $wgUseImageResize = false;
  27
  28         if ( $newOnly ) {
  29                 print "Refreshing links from ";
  30                 $res = $dbr->select( 'page',
  31                         array( 'page_id' ),
  32                         array(
  33                                 'page_is_new' => 1,
  34                                 "page_id > $start" ),
  35                         $fname
  36                 );
  37                 $num = $dbr->numRows( $res );
  38                 print "$num new articles...\n";
  39
  40                 $i = 0;
  41                 while ( $row = $dbr->fetchObject( $res ) ) {
  42                         if ( !( ++$i % REPORTING_INTERVAL ) ) {
  43                                 print "$i\n";
  44                                 wfWaitForSlaves( $maxLag );
  45                         }
  46
  47                         fixLinksFromArticle( $row->page_id );
  48                 }
  49         } else {
  50                 print "Refreshing link table.\n";
  51                 $end = $dbr->selectField( 'page', 'max(page_id)', false );
  52                 print("Starting from page_id $start of $end.\n");
  53
  54                 for ($id = $start; $id <= $end; $id++) {
  55
  56                         if ( !($id % REPORTING_INTERVAL) ) {
  57                                 print "$id\n";
  58                                 wfWaitForSlaves( $maxLag );
  59                         }
  60                         fixLinksFromArticle( $id );
  61                 }
  62
  63
  64         }
  65 }
  66
  67 function fixLinksFromArticle( $id ) {
  68         global $wgTitle, $wgArticle, $wgLinkCache, $wgOut;
  69
  70         $wgTitle = Title::newFromID( $id );
  71         $dbw =& wfGetDB( DB_MASTER );
  72
  73         if ( is_null( $wgTitle ) ) {
  74                 return;
  75         }
  76         $dbw->begin();
  77
  78         $wgArticle = new Article( $wgTitle );
  79         $text = $wgArticle->getContent( true );
  80         $wgLinkCache = new LinkCache;
  81         $wgLinkCache->forUpdate( true );
  82
  83         global $wgLinkHolders;
  84         $wgLinkHolders = array(
  85                 'namespaces' => array(),
  86                 'dbkeys' => array(),
  87                 'queries' => array(),
  88                 'texts' => array(),
  89                 'titles' => array()
  90         );
  91
  92
  93         # Parse the text and replace links with placeholders
  94         $wgOut->addWikiText( $text );
  95
  96         # Look up the links in the DB and add them to the link cache
  97         $wgOut->clearHTML();
  98
  99         $linksUpdate = new LinksUpdate( $id, $wgTitle->getPrefixedDBkey() );
 100         $linksUpdate->doDumbUpdate();
 101         $dbw->immediateCommit();
 102 }
 103
 104 function deleteLinksFromNonexistent( $maxLag = 0 ) {
 105         $fname = 'deleteLinksFromNonexistent';
 106
 107         wfWaitForSlaves( $maxLag );
 108
 109         $dbw =& wfGetDB( DB_WRITE );
 110
 111         $linksTables = array(
 112                 'pagelinks' => 'pl_from',
 113                 'imagelinks' => 'il_from',
 114                 'categorylinks' => 'cl_from',
 115         );
 116
 117         $page = $dbw->tableName( 'page' );
 118
 119
 120         foreach ( $linksTables as $table => $field ) {
 121                 if ( !$dbw->ping() ) {
 122                         print "DB disconnected, reconnecting...";
 123                         while ( !$dbw->ping() ) {
 124                                 print ".";
 125                                 sleep(10);
 126                         }
 127                         print "\n";
 128                 }
 129
 130                 $pTable = $dbw->tableName( $table );
 131                 $sql = "DELETE $pTable FROM $pTable LEFT JOIN $page ON page_id=$field WHERE page_id IS NULL";
 132
 133                 print "Deleting $table from non-existent articles...";
 134                 $dbw->query( $sql, $fname );
 135                 print " fixed " .$dbw->affectedRows() . " row(s)\n";
 136         }
 137 }
 138
 139 ?>