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, $redirectsOnly = false ) {
  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->clearTagHooks();
  23
  24         # Don't generate thumbnail images
  25         $wgUseImageResize = false;
  26         $wgUseTidy = false;
  27
  28         $what = ($redirectsOnly)? "redirects" : "links";
  29
  30         if ( $newOnly ) {
  31                 print "Refreshing $what from ";
  32                 $res = $dbr->select( 'page',
  33                         array( 'page_id' ),
  34                         array(
  35                                 'page_is_new' => 1,
  36                                 "page_id > $start" ),
  37                         $fname
  38                 );
  39                 $num = $dbr->numRows( $res );
  40                 print "$num new articles...\n";
  41
  42                 $i = 0;
  43                 while ( $row = $dbr->fetchObject( $res ) ) {
  44                         if ( !( ++$i % REPORTING_INTERVAL ) ) {
  45                                 print "$i\n";
  46                                 wfWaitForSlaves( $maxLag );
  47                         }
  48                         if($redirectsOnly)
  49                                 fixRedirect( $row->page_id );
  50                         else
  51                                 fixLinksFromArticle( $row->page_id );
  52                 }
  53         } else {
  54                 print "Refreshing $what table.\n";
  55                 if ( !$end ) {
  56                         $end = $dbr->selectField( 'page', 'max(page_id)', false );
  57                 }
  58                 print("Starting from page_id $start of $end.\n");
  59
  60                 for ($id = $start; $id <= $end; $id++) {
  61
  62                         if ( !($id % REPORTING_INTERVAL) ) {
  63                                 print "$id\n";
  64                                 wfWaitForSlaves( $maxLag );
  65                         }
  66                         if($redirectsOnly)
  67                                 fixRedirect( $id );
  68                         else
  69                                 fixLinksFromArticle( $id );
  70                 }
  71         }
  72 }
  73
  74 function fixRedirect( $id ){
  75         global $wgTitle, $wgArticle;
  76
  77         $wgTitle = Title::newFromID( $id );
  78         $dbw = wfGetDB( DB_MASTER );
  79
  80         if ( is_null( $wgTitle ) ) {
  81                 return;
  82         }
  83         $wgArticle = new Article($wgTitle);
  84
  85         $rt = $wgArticle->followRedirect();
  86
  87         if($rt == false || !is_object($rt))
  88                 return;
  89
  90         $wgArticle->updateRedirectOn($dbw,$rt);
  91 }
  92
  93 function fixLinksFromArticle( $id ) {
  94         global $wgTitle, $wgParser;
  95
  96         $wgTitle = Title::newFromID( $id );
  97         $dbw = wfGetDB( DB_MASTER );
  98
  99         $linkCache =& LinkCache::singleton();
 100         $linkCache->clear();
 101
 102         if ( is_null( $wgTitle ) ) {
 103                 return;
 104         }
 105         $dbw->begin();
 106
 107         $revision = Revision::newFromTitle( $wgTitle );
 108         if ( !$revision ) {
 109                 return;
 110         }
 111
 112         $options = new ParserOptions;
 113         $parserOutput = $wgParser->parse( $revision->getText(), $wgTitle, $options, true, true, $revision->getId() );
 114         $update = new LinksUpdate( $wgTitle, $parserOutput, false );
 115         $update->doUpdate();
 116         $dbw->immediateCommit();
 117 }
 118
 119 function deleteLinksFromNonexistent( $maxLag = 0 ) {
 120         $fname = 'deleteLinksFromNonexistent';
 121
 122         wfWaitForSlaves( $maxLag );
 123
 124         $dbw = wfGetDB( DB_WRITE );
 125
 126         $linksTables = array(
 127                 'pagelinks' => 'pl_from',
 128                 'imagelinks' => 'il_from',
 129                 'categorylinks' => 'cl_from',
 130                 'templatelinks' => 'tl_from',
 131                 'externallinks' => 'el_from',
 132         );
 133
 134         $page = $dbw->tableName( 'page' );
 135
 136
 137         foreach ( $linksTables as $table => $field ) {
 138                 if ( !$dbw->ping() ) {
 139                         print "DB disconnected, reconnecting...";
 140                         while ( !$dbw->ping() ) {
 141                                 print ".";
 142                                 sleep(10);
 143                         }
 144                         print "\n";
 145                 }
 146
 147                 $pTable = $dbw->tableName( $table );
 148                 $sql = "DELETE $pTable FROM $pTable LEFT JOIN $page ON page_id=$field WHERE page_id IS NULL";
 149
 150                 print "Deleting $table from non-existent articles...";
 151                 $dbw->query( $sql, $fname );
 152                 print " fixed " .$dbw->affectedRows() . " row(s)\n";
 153         }
 154 }
 155
 156 ?>