maintenance/refreshLinks.inc

   1 <?php
   2 /**
   3  * @todo document
   4  * @file
   5  * @ingroup Maintenance
   6  */
   7
   8 function refreshLinks( $start, $newOnly = false, $maxLag = false, $end = 0, $redirectsOnly = false, $oldRedirectsOnly = false ) {
   9         global $wgUser, $wgParser, $wgUseTidy;
  10
  11         $reportingInterval = 100;
  12         $fname = 'refreshLinks';
  13         $dbr = wfGetDB( DB_SLAVE );
  14         $start = intval( $start );
  15
  16         # Don't generate TeX PNGs (lack of a sensible current directory causes errors anyway)
  17         $wgUser->setOption('math', MW_MATH_SOURCE);
  18
  19         # Don't generate extension images (e.g. Timeline)
  20         $wgParser->clearTagHooks();
  21
  22         # Don't use HTML tidy
  23         $wgUseTidy = false;
  24
  25         $what = $redirectsOnly ? "redirects" : "links";
  26
  27         if( $oldRedirectsOnly ) {
  28                 # This entire code path is cut-and-pasted from below.  Hurrah.
  29                 $res = $dbr->query(
  30                         "SELECT page_id ".
  31                         "FROM page ".
  32                         "LEFT JOIN redirect ON page_id=rd_from ".
  33                         "WHERE page_is_redirect=1 AND rd_from IS NULL AND ".
  34                         ($end == 0 ? "page_id >= $start"
  35                                    : "page_id BETWEEN $start AND $end"),
  36                         $fname
  37                 );
  38                 $num = $dbr->numRows( $res );
  39                 print "Refreshing $num old redirects from $start...\n";
  40
  41                 while( $row = $dbr->fetchObject( $res ) ) {
  42                         if ( !( ++$i % $reportingInterval ) ) {
  43                                 print "$i\n";
  44                                 wfWaitForSlaves( $maxLag );
  45                         }
  46                         fixRedirect( $row->page_id );
  47                 }
  48         } elseif( $newOnly ) {
  49                 print "Refreshing $what from ";
  50                 $res = $dbr->select( 'page',
  51                         array( 'page_id' ),
  52                         array(
  53                                 'page_is_new' => 1,
  54                                 "page_id >= $start" ),
  55                         $fname
  56                 );
  57                 $num = $dbr->numRows( $res );
  58                 print "$num new articles...\n";
  59
  60                 $i = 0;
  61                 while ( $row = $dbr->fetchObject( $res ) ) {
  62                         if ( !( ++$i % $reportingInterval ) ) {
  63                                 print "$i\n";
  64                                 wfWaitForSlaves( $maxLag );
  65                         }
  66                         if($redirectsOnly)
  67                                 fixRedirect( $row->page_id );
  68                         else
  69                                 fixLinksFromArticle( $row->page_id );
  70                 }
  71         } else {
  72                 print "Refreshing $what table.\n";
  73                 if ( !$end ) {
  74                         $end = $dbr->selectField( 'page', 'max(page_id)', false );
  75                 }
  76                 print("Starting from page_id $start of $end.\n");
  77
  78                 for ($id = $start; $id <= $end; $id++) {
  79
  80                         if ( !($id % $reportingInterval) ) {
  81                                 print "$id\n";
  82                                 wfWaitForSlaves( $maxLag );
  83                         }
  84                         if($redirectsOnly)
  85                                 fixRedirect( $id );
  86                         else
  87                                 fixLinksFromArticle( $id );
  88                 }
  89         }
  90 }
  91
  92 function fixRedirect( $id ){
  93         global $wgTitle, $wgArticle;
  94
  95         $wgTitle = Title::newFromID( $id );
  96         $dbw = wfGetDB( DB_MASTER );
  97
  98         if ( is_null( $wgTitle ) ) {
  99                 return;
 100         }
 101         $wgArticle = new Article($wgTitle);
 102
 103         $rt = $wgArticle->followRedirect();
 104
 105         if($rt == false || !is_object($rt))
 106                 return;
 107
 108         $wgArticle->updateRedirectOn($dbw,$rt);
 109 }
 110
 111 function fixLinksFromArticle( $id ) {
 112         global $wgTitle, $wgParser;
 113
 114         $wgTitle = Title::newFromID( $id );
 115         $dbw = wfGetDB( DB_MASTER );
 116
 117         $linkCache =& LinkCache::singleton();
 118         $linkCache->clear();
 119
 120         if ( is_null( $wgTitle ) ) {
 121                 return;
 122         }
 123         $dbw->begin();
 124
 125         $revision = Revision::newFromTitle( $wgTitle );
 126         if ( !$revision ) {
 127                 return;
 128         }
 129
 130         $options = new ParserOptions;
 131         $parserOutput = $wgParser->parse( $revision->getText(), $wgTitle, $options, true, true, $revision->getId() );
 132         $update = new LinksUpdate( $wgTitle, $parserOutput, false );
 133         $update->doUpdate();
 134         $dbw->immediateCommit();
 135 }
 136
 137 function deleteLinksFromNonexistent( $maxLag = 0 ) {
 138         $fname = 'deleteLinksFromNonexistent';
 139
 140         wfWaitForSlaves( $maxLag );
 141
 142         $dbw = wfGetDB( DB_MASTER );
 143
 144         $linksTables = array(
 145                 'pagelinks' => 'pl_from',
 146                 'imagelinks' => 'il_from',
 147                 'categorylinks' => 'cl_from',
 148                 'templatelinks' => 'tl_from',
 149                 'externallinks' => 'el_from',
 150         );
 151
 152         $page = $dbw->tableName( 'page' );
 153
 154
 155         foreach ( $linksTables as $table => $field ) {
 156                 if ( !$dbw->ping() ) {
 157                         print "DB disconnected, reconnecting...";
 158                         while ( !$dbw->ping() ) {
 159                                 print ".";
 160                                 sleep(10);
 161                         }
 162                         print "\n";
 163                 }
 164
 165                 $pTable = $dbw->tableName( $table );
 166                 $sql = "DELETE $pTable FROM $pTable LEFT JOIN $page ON page_id=$field WHERE page_id IS NULL";
 167
 168                 print "Deleting $table from non-existent articles...";
 169                 $dbw->query( $sql, $fname );
 170                 print " fixed " .$dbw->affectedRows() . " row(s)\n";
 171         }
 172 }
 173
 174 ?>