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, $oldRedirectsOnly = 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( $oldRedirectsOnly ) {
  31                 # This entire code path is cut-and-pasted from below.  Hurrah.
  32                 $res = $dbr->query(
  33                         "SELECT page_id ".
  34                         "FROM page ".
  35                         "LEFT JOIN redirect ON page_id=rd_from ".
  36                         "WHERE page_is_redirect=1 AND rd_from IS NULL AND ".
  37                         ($end == 0 ? "page_id >= $start"
  38                                    : "page_id BETWEEN $start AND $end"),
  39                         $fname
  40                 );
  41                 $num = $dbr->numRows( $res );
  42                 print "Refreshing $num old redirects from $start...\n";
  43
  44                 while( $row = $dbr->fetchObject( $res ) ) {
  45                         if ( !( ++$i % REPORTING_INTERVAL ) ) {
  46                                 print "$i\n";
  47                                 wfWaitForSlaves( $maxLag );
  48                         }
  49                         fixRedirect( $row->page_id );
  50                 }
  51         } elseif( $newOnly ) {
  52                 print "Refreshing $what from ";
  53                 $res = $dbr->select( 'page',
  54                         array( 'page_id' ),
  55                         array(
  56                                 'page_is_new' => 1,
  57                                 "page_id >= $start" ),
  58                         $fname
  59                 );
  60                 $num = $dbr->numRows( $res );
  61                 print "$num new articles...\n";
  62
  63                 $i = 0;
  64                 while ( $row = $dbr->fetchObject( $res ) ) {
  65                         if ( !( ++$i % REPORTING_INTERVAL ) ) {
  66                                 print "$i\n";
  67                                 wfWaitForSlaves( $maxLag );
  68                         }
  69                         if($redirectsOnly)
  70                                 fixRedirect( $row->page_id );
  71                         else
  72                                 fixLinksFromArticle( $row->page_id );
  73                 }
  74         } else {
  75                 print "Refreshing $what table.\n";
  76                 if ( !$end ) {
  77                         $end = $dbr->selectField( 'page', 'max(page_id)', false );
  78                 }
  79                 print("Starting from page_id $start of $end.\n");
  80
  81                 for ($id = $start; $id <= $end; $id++) {
  82
  83                         if ( !($id % REPORTING_INTERVAL) ) {
  84                                 print "$id\n";
  85                                 wfWaitForSlaves( $maxLag );
  86                         }
  87                         if($redirectsOnly)
  88                                 fixRedirect( $id );
  89                         else
  90                                 fixLinksFromArticle( $id );
  91                 }
  92         }
  93 }
  94
  95 function fixRedirect( $id ){
  96         global $wgTitle, $wgArticle;
  97
  98         $wgTitle = Title::newFromID( $id );
  99         $dbw = wfGetDB( DB_MASTER );
 100
 101         if ( is_null( $wgTitle ) ) {
 102                 return;
 103         }
 104         $wgArticle = new Article($wgTitle);
 105
 106         $rt = $wgArticle->followRedirect();
 107
 108         if($rt == false || !is_object($rt))
 109                 return;
 110
 111         $wgArticle->updateRedirectOn($dbw,$rt);
 112 }
 113
 114 function fixLinksFromArticle( $id ) {
 115         global $wgTitle, $wgParser;
 116
 117         $wgTitle = Title::newFromID( $id );
 118         $dbw = wfGetDB( DB_MASTER );
 119
 120         $linkCache =& LinkCache::singleton();
 121         $linkCache->clear();
 122
 123         if ( is_null( $wgTitle ) ) {
 124                 return;
 125         }
 126         $dbw->begin();
 127
 128         $revision = Revision::newFromTitle( $wgTitle );
 129         if ( !$revision ) {
 130                 return;
 131         }
 132
 133         $options = new ParserOptions;
 134         $parserOutput = $wgParser->parse( $revision->getText(), $wgTitle, $options, true, true, $revision->getId() );
 135         $update = new LinksUpdate( $wgTitle, $parserOutput, false );
 136         $update->doUpdate();
 137         $dbw->immediateCommit();
 138 }
 139
 140 function deleteLinksFromNonexistent( $maxLag = 0 ) {
 141         $fname = 'deleteLinksFromNonexistent';
 142
 143         wfWaitForSlaves( $maxLag );
 144
 145         $dbw = wfGetDB( DB_MASTER );
 146
 147         $linksTables = array(
 148                 'pagelinks' => 'pl_from',
 149                 'imagelinks' => 'il_from',
 150                 'categorylinks' => 'cl_from',
 151                 'templatelinks' => 'tl_from',
 152                 'externallinks' => 'el_from',
 153         );
 154
 155         $page = $dbw->tableName( 'page' );
 156
 157
 158         foreach ( $linksTables as $table => $field ) {
 159                 if ( !$dbw->ping() ) {
 160                         print "DB disconnected, reconnecting...";
 161                         while ( !$dbw->ping() ) {
 162                                 print ".";
 163                                 sleep(10);
 164                         }
 165                         print "\n";
 166                 }
 167
 168                 $pTable = $dbw->tableName( $table );
 169                 $sql = "DELETE $pTable FROM $pTable LEFT JOIN $page ON page_id=$field WHERE page_id IS NULL";
 170
 171                 print "Deleting $table from non-existent articles...";
 172                 $dbw->query( $sql, $fname );
 173                 print " fixed " .$dbw->affectedRows() . " row(s)\n";
 174         }
 175 }
 176
 177 ?>