maintenance/refreshLinks.inc

   1 <?php
   2 require_once 'counter.php';
   3 /**
   4  * @todo document
   5  * @package MediaWiki
   6  * @subpackage Maintenance
   7  */
   8
   9 /** */
  10 define( 'REPORTING_INTERVAL', 10 );
  11
  12 function refreshLinks( $start, $newOnly = false, $maxLag = false ) {
  13         global $wgUser, $wgParser, $wgUseImageResize;
  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
  29         if ( $newOnly ) {
  30                 print "Refreshing links from ";
  31                 $res = $dbr->select( 'page',
  32                         array( 'page_id' ),
  33                         array(
  34                                 'page_is_new' => 1,
  35                                 "page_id > $start" ),
  36                         $fname
  37                 );
  38                 $num = $dbr->numRows( $res );
  39                 print "$num new articles...\n";
  40
  41                 $i = 0;
  42                 while ( $row = $dbr->fetchObject( $res ) ) {
  43                         if ( !( ++$i % REPORTING_INTERVAL ) ) {
  44                                 print_c( $i - 1, $i );
  45                                 wfWaitForSlaves( $maxLag );
  46                         }
  47
  48                         fixLinksFromArticle( $row->page_id );
  49                 }
  50         } else {
  51                 print "Refreshing link table.\n";
  52                 $end = $dbr->selectField( 'page', 'max(page_id)', false );
  53                 print("Starting from page_id $start of $end.\n");
  54
  55                 for ($id = $start; $id <= $end; $id++) {
  56
  57                         if ( !($id % REPORTING_INTERVAL) ) {
  58                                 print_c(  $id - 1, $id );
  59                                 wfWaitForSlaves( $maxLag );
  60                         }
  61                         fixLinksFromArticle( $id );
  62                 }
  63
  64
  65         }
  66 }
  67
  68 function fixLinksFromArticle( $id ) {
  69         global $wgTitle, $wgArticle, $wgLinkCache, $wgOut;
  70
  71         $wgTitle = Title::newFromID( $id );
  72         $dbw =& wfGetDB( DB_MASTER );
  73
  74         if ( is_null( $wgTitle ) ) {
  75                 return;
  76         }
  77         $dbw->begin();
  78
  79         $wgArticle = new Article( $wgTitle );
  80         $text = $wgArticle->getContent( true );
  81         $wgLinkCache = new LinkCache;
  82         $wgLinkCache->forUpdate( true );
  83
  84         global $wgLinkHolders;
  85         $wgLinkHolders = array(
  86                 'namespaces' => array(),
  87                 'dbkeys' => array(),
  88                 'queries' => array(),
  89                 'texts' => array(),
  90                 'titles' => array()
  91         );
  92
  93
  94         # Parse the text and replace links with placeholders
  95         $wgOut->addWikiText( $text );
  96
  97         # Look up the links in the DB and add them to the link cache
  98         $wgOut->transformBuffer();
  99         $wgOut->clearHTML();
 100
 101         $linksUpdate = new LinksUpdate( $id, $wgTitle->getPrefixedDBkey() );
 102         $linksUpdate->doDumbUpdate();
 103         $dbw->immediateCommit();
 104 }
 105
 106 function deleteLinksFromNonexistent( $maxLag = 0 ) {
 107         $fname = 'deleteLinksFromNonexistent';
 108
 109         wfWaitForSlaves( $maxLag );
 110
 111         $dbw =& wfGetDB( DB_WRITE );
 112
 113         $linksTables = array(
 114                 'pagelinks' => 'pl_from',
 115                 'imagelinks' => 'il_from',
 116                 'categorylinks' => 'cl_from',
 117         );
 118
 119         $page = $dbw->tableName( 'page' );
 120
 121
 122         foreach ( $linksTables as $table => $field ) {
 123                 if ( !$dbw->ping() ) {
 124                         print "DB disconnected, reconnecting...";
 125                         while ( !$dbw->ping() ) {
 126                                 print ".";
 127                                 sleep(10);
 128                         }
 129                         print "\n";
 130                 }
 131
 132                 $pTable = $dbw->tableName( $table );
 133                 $sql = "DELETE $pTable FROM $pTable LEFT JOIN $page ON page_id=$field WHERE page_id IS NULL";
 134
 135                 print "Deleting $table from non-existent articles...";
 136                 $dbw->query( $sql, $fname );
 137                 print " fixed " .$dbw->affectedRows() . " row(s)\n";
 138         }
 139 }
 140
 141 ?>