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", 3);
  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', array( 'page_id' ),
  31                   array( 'page_is_new' => 1, "page_id > $start" ), $fname );
  32                 $num = $dbr->numRows( $res );
  33                 print "$num new articles...\n";
  34
  35                 $i = 0;
  36                 while ( $row = $dbr->fetchObject( $res ) ) {
  37                         if ( !( ++$i % REPORTING_INTERVAL ) ) {
  38                                 print "$i\n";
  39                                 wfWaitForSlaves( $maxLag );
  40                         }
  41
  42                         fixLinksFromArticle( $row->page_id );
  43                 }
  44         } else {
  45                 print "Refreshing link table.\n";
  46                 $end = $dbr->selectField( 'page', 'max(page_id)', false );
  47                 print("Starting from page_id $start of $end.\n");
  48
  49                 for ($id = $start; $id <= $end; $id++) {
  50
  51                         if ( !($id % REPORTING_INTERVAL) ) {
  52                                 print "$id\n";
  53                                 wfWaitForSlaves( $maxLag );
  54                         }
  55                         fixLinksFromArticle( $id );
  56                 }
  57
  58
  59         }
  60 }
  61
  62 function fixLinksFromArticle( $id ) {
  63         global $wgTitle, $wgArticle, $wgLinkCache, $wgOut;
  64
  65         $wgTitle = Title::newFromID( $id );
  66         $dbw =& wfGetDB( DB_MASTER );
  67
  68         if ( is_null( $wgTitle ) ) {
  69                 return;
  70         }
  71         $dbw->begin();
  72
  73         $wgArticle = new Article( $wgTitle );
  74         $text = $wgArticle->getContent( true );
  75         $wgLinkCache = new LinkCache;
  76         $wgLinkCache->forUpdate( true );
  77
  78         global $wgLinkHolders;
  79         $wgLinkHolders = array(
  80                 'namespaces' => array(),
  81                 'dbkeys' => array(),
  82                 'queries' => array(),
  83                 'texts' => array(),
  84                 'titles' => array()
  85         );
  86
  87
  88         # Parse the text and replace links with placeholders
  89         $wgOut->addWikiText( $text );
  90
  91         # Look up the links in the DB and add them to the link cache
  92         $wgOut->transformBuffer();
  93         $wgOut->clearHTML();
  94
  95         $linksUpdate = new LinksUpdate( $id, $wgTitle->getPrefixedDBkey() );
  96         $linksUpdate->doDumbUpdate();
  97         $dbw->immediateCommit();
  98 }
  99
 100 function deleteLinksFromNonexistent( $maxLag = 0 ) {
 101         $fname = 'deleteLinksFromNonexistent';
 102
 103         wfWaitForSlaves( $maxLag );
 104
 105         $dbw =& wfGetDB( DB_WRITE );
 106
 107         $linksTables = array(
 108                 'pagelinks' => 'pl_from',
 109                 'imagelinks' => 'il_from',
 110                 'categorylinks' => 'cl_from',
 111         );
 112
 113         $page = $dbw->tableName( 'page' );
 114
 115
 116         foreach ( $linksTables as $table => $field ) {
 117                 if ( !$dbw->ping() ) {
 118                         print "DB disconnected, reconnecting...";
 119                         while ( !$dbw->ping() ) {
 120                                 print ".";
 121                                 sleep(10);
 122                         }
 123                         print "\n";
 124                 }
 125
 126                 $pTable = $dbw->tableName( $table );
 127                 global $wgDBmysql4, $wgDBtype;
 128                 if( $wgDBmysql4 || $wgDBtype != 'mysql' ) {
 129                         $sql = "DELETE $pTable FROM $pTable LEFT JOIN $page ON page_id=$field WHERE page_id IS NULL";
 130                 } else {
 131                         # Hack-around for MySQL 3.x, which lacks support
 132                         # for multi-table deletes.
 133
 134                         $sql = "SELECT DISTINCT $field AS id FROM $pTable LEFT JOIN $page ON page_id=$field WHERE page_id IS NULL";
 135                         echo "Looking in $table from non-existent articles...";
 136                         $result = $dbw->query( $sql );
 137                         $ids = array();
 138                         while( $row = $dbw->fetchObject( $result ) ) {
 139                                 $ids[] = $row->id;
 140                         }
 141                         $dbw->freeResult( $result );
 142
 143                         if( empty( $ids ) ) {
 144                                 echo " none.\n";
 145                                 continue;
 146                         }
 147                         echo " found.\n";
 148                         $sql = "DELETE FROM $pTable WHERE $field IN (" . implode( ",", $ids ) . ")";
 149                 }
 150
 151                 print "Deleting $table from non-existent articles...";
 152                 $dbw->query( $sql, $fname );
 153                 print " fixed " .$dbw->affectedRows() . " row(s)\n";
 154         }
 155 }
 156
 157 ?>