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         $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         $wgUseTidy = 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 "$i\n";
  45                                 wfWaitForSlaves( $maxLag );
  46                         }
  47
  48                         fixLinksFromArticle( $row->page_id );
  49                 }
  50         } else {
  51                 print "Refreshing link table.\n";
  52                 if ( !$end ) {
  53                         $end = $dbr->selectField( 'page', 'max(page_id)', false );
  54                 }
  55                 print("Starting from page_id $start of $end.\n");
  56
  57                 for ($id = $start; $id <= $end; $id++) {
  58
  59                         if ( !($id % REPORTING_INTERVAL) ) {
  60                                 print "$id\n";
  61                                 wfWaitForSlaves( $maxLag );
  62                         }
  63                         fixLinksFromArticle( $id );
  64                 }
  65         }
  66 }
  67
  68 function fixLinksFromArticle( $id ) {
  69         global $wgTitle, $wgParser;
  70
  71         $wgTitle = Title::newFromID( $id );
  72         $dbw =& wfGetDB( DB_MASTER );
  73
  74         $linkCache =& LinkCache::singleton();
  75         $linkCache->clear();
  76
  77         if ( is_null( $wgTitle ) ) {
  78                 return;
  79         }
  80         $dbw->begin();
  81
  82         $revision = Revision::newFromTitle( $wgTitle );
  83         if ( !$revision ) {
  84                 return;
  85         }
  86
  87         $options = new ParserOptions;
  88         $parserOutput = $wgParser->parse( $revision->getText(), $wgTitle, $options, true, true, $revision->getId() );
  89         $update = new LinksUpdate( $wgTitle, $parserOutput, false );
  90         $update->doUpdate();
  91         $dbw->immediateCommit();
  92 }
  93
  94 function deleteLinksFromNonexistent( $maxLag = 0 ) {
  95         $fname = 'deleteLinksFromNonexistent';
  96
  97         wfWaitForSlaves( $maxLag );
  98
  99         $dbw =& wfGetDB( DB_WRITE );
 100
 101         $linksTables = array(
 102                 'pagelinks' => 'pl_from',
 103                 'imagelinks' => 'il_from',
 104                 'categorylinks' => 'cl_from',
 105                 'templatelinks' => 'tl_from',
 106                 'externallinks' => 'el_from',
 107         );
 108
 109         $page = $dbw->tableName( 'page' );
 110
 111
 112         foreach ( $linksTables as $table => $field ) {
 113                 if ( !$dbw->ping() ) {
 114                         print "DB disconnected, reconnecting...";
 115                         while ( !$dbw->ping() ) {
 116                                 print ".";
 117                                 sleep(10);
 118                         }
 119                         print "\n";
 120                 }
 121
 122                 $pTable = $dbw->tableName( $table );
 123                 $sql = "DELETE $pTable FROM $pTable LEFT JOIN $page ON page_id=$field WHERE page_id IS NULL";
 124
 125                 print "Deleting $table from non-existent articles...";
 126                 $dbw->query( $sql, $fname );
 127                 print " fixed " .$dbw->affectedRows() . " row(s)\n";
 128         }
 129 }
 130
 131 ?>