maintenance/refreshLinks.inc

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