maintenance/rebuildlinks.inc

   1 <?
   2
   3 # Functions for rebuilding the link tracking tables; must
   4 # be included within a script that also includes the Setup.
   5 # See rebuildlinks.php, for example.
   6 #
   7
   8 function rebuildLinkTablesPass1()
   9 {
  10         global $wgLang;
  11         $count = 0;
  12         print "Rebuilding link tables (pass 1).\n";
  13
  14         $sql = "DROP TABLE IF EXISTS rebuildlinks";
  15         wfQuery( $sql, DB_WRITE );
  16
  17         $sql = "CREATE TABLE rebuildlinks (
  18   rl_f_id int(8) unsigned NOT NULL default 0,
  19   rl_f_title varchar(255) binary NOT NULL default '',
  20   rl_to varchar(255) binary NOT NULL default '',
  21   INDEX rl_to (rl_to) ) TYPE=MyISAM";
  22         wfQuery( $sql, DB_WRITE );
  23
  24         $sql = "LOCK TABLES cur READ, rebuildlinks WRITE, interwiki READ";
  25         wfQuery( $sql, DB_WRITE );
  26
  27         $sql = "DELETE FROM rebuildlinks";
  28         wfQuery( $sql, DB_WRITE );
  29
  30         $sql = "SELECT cur_id,cur_namespace,cur_title,cur_text FROM cur";
  31         $res = wfQuery( $sql, DB_WRITE );
  32         $total = wfNumRows( $res );
  33
  34         $tc = Title::legalChars();
  35         while ( $row = wfFetchObject( $res ) ) {
  36                 $id = $row->cur_id;
  37                 $ns = $wgLang->getNsText( $row->cur_namespace );
  38                 if ( "" == $ns ) {
  39                         $title = addslashes( $row->cur_title );
  40                 } else {
  41                         $title = addslashes( "$ns:{$row->cur_title}" );
  42                 }
  43                 $text = $row->cur_text;
  44                 $numlinks = preg_match_all( "/\\[\\[([{$tc}]+)(]|\\|)/", $text,
  45                   $m, PREG_PATTERN_ORDER );
  46
  47                 if ( 0 != $numlinks ) {
  48                         $first = true;
  49                         $sql = "INSERT INTO rebuildlinks (rl_f_id,rl_f_title,rl_to) VALUES ";
  50                         for ( $i = 0; $i < $numlinks; ++$i ) {
  51                                 $nt = Title::newFromText( $m[1][$i] );
  52                                 if (! $nt)
  53                                 {
  54                                         $txt = $m[1][$i];
  55                                         print "error in '$ns:{$row->cur_title}' :\t'$txt'\n";
  56                                         continue;
  57                                 }
  58
  59                                 if (!$first)
  60                                         $sql .= ",";
  61                                 else
  62                                         $first = false;
  63
  64                                 $dest = addslashes( $nt->getPrefixedDBkey() );
  65                                 $sql .= "({$id},'{$title}','{$dest}')";
  66                         }
  67
  68                         if (! $first) { wfQuery( $sql, DB_WRITE  ); }
  69                 }
  70                 if ( ( ++$count % 1000 ) == 0 ) {
  71                         print "$count of $total articles scanned.\n";
  72                 }
  73         }
  74         print "$total articles scanned.\n";
  75         mysql_free_result( $res );
  76
  77         $sql = "UNLOCK TABLES";
  78         wfQuery( $sql, DB_WRITE );
  79 }
  80
  81 function rebuildLinkTablesPass2()
  82 {
  83         global $wgLang;
  84         $count = 0;
  85         print "Rebuilding link tables (pass 2).\n";
  86
  87         $sql = "LOCK TABLES cur READ, rebuildlinks READ, interwiki READ, " .
  88           "links WRITE, brokenlinks WRITE, imagelinks WRITE";
  89         wfQuery( $sql, DB_WRITE );
  90
  91         $sql = "DELETE FROM links";
  92         wfQuery( $sql, DB_WRITE );
  93
  94         $sql = "DELETE FROM brokenlinks";
  95         wfQuery( $sql, DB_WRITE );
  96
  97         $sql = "DELETE FROM links";
  98         wfQuery( $sql, DB_WRITE );
  99
 100         $ins = $wgLang->getNsText( Namespace::getImage() );
 101         $inslen = strlen($ins)+1;
 102         $sql = "SELECT rl_f_title,rl_to FROM rebuildlinks " .
 103           "WHERE rl_to LIKE '$ins:%'";
 104         $res = wfQuery( $sql, DB_WRITE );
 105
 106         $sql = "INSERT INTO imagelinks (il_from,il_to) VALUES ";
 107         $first = true;
 108         while ( $row = wfFetchObject( $res ) )
 109         {
 110                 $iname = addslashes( substr( $row->rl_to, $inslen ) );
 111                 $pname = addslashes( $row->rl_f_title );
 112
 113                 if ( ! $first )
 114                         $sql .= ",";
 115                 else
 116                         $first = false;
 117
 118                 $sql .= "('{$pname}','{$iname}')";
 119         }
 120         wfFreeResult( $res );
 121         if ( ! $first ) { wfQuery( $sql, DB_WRITE ); }
 122
 123         $sql = "SELECT DISTINCT rl_to FROM rebuildlinks ORDER BY rl_to";
 124         $res = wfQuery( $sql, DB_WRITE );
 125         $count = 0;
 126         $total = wfNumRows( $res );
 127
 128         while ( $row = wfFetchObject( $res ) ) {
 129                 if ( 0 == strncmp( "$ins:", $row->rl_to, $inslen ) ) { continue; }
 130
 131                 $nt = Title::newFromDBkey( $row->rl_to );
 132                 if (! $nt)
 133                 {
 134                         print "error pass2: '{$row->rl_to}'\n";
 135                         continue;
 136                 }
 137                 $id = $nt->getArticleID();
 138                 $to = addslashes( $row->rl_to );
 139
 140                 if ( 0 == $id ) {
 141                         $sql = "SELECT rl_f_id FROM rebuildlinks WHERE rl_to='{$to}'";
 142                         $res2 = wfQuery( $sql, DB_WRITE );
 143
 144                         $sql = "INSERT INTO brokenlinks (bl_from,bl_to) VALUES ";
 145                         $first = true;
 146                         while ( $row2 = wfFetchObject( $res2 ) )
 147                         {
 148                                 if (! $first)
 149                                         $sql .= ",";
 150                                 else
 151                                         $first = false;
 152
 153                                 $from = $row2->rl_f_id;
 154                                 $sql .= "({$from},'{$to}')";
 155                         }
 156                         wfFreeResult( $res2 );
 157                         if ( ! $first ) { wfQuery( $sql, DB_WRITE ); }
 158                 } else {
 159                         $sql = "SELECT rl_f_title FROM rebuildlinks WHERE rl_to='{$to}'";
 160                         $res2 = wfQuery( $sql, DB_WRITE );
 161
 162                         $sql = "INSERT INTO links (l_from,l_to) VALUES ";
 163                         $first = true;
 164                         while ( $row2 = wfFetchObject( $res2 ) )
 165                         {
 166                                 if (! $first)
 167                                         $sql .= ",";
 168                                 else
 169                                         $first = false;
 170
 171                                 $from = addslashes( $row2->rl_f_title );
 172                                 $sql .= "('{$from}',{$id})";
 173                         }
 174                         wfFreeResult( $res2 );
 175                         if ( ! $first ) { wfQuery( $sql, DB_WRITE ); }
 176                 }
 177                 if ( ( ++$count % 1000 ) == 0 ) {
 178                         print "$count of $total titles processed.\n";
 179                 }
 180         }
 181         wfFreeResult( $res );
 182
 183         $sql = "UNLOCK TABLES";
 184         wfQuery( $sql, DB_WRITE );
 185
 186         $sql = "DROP TABLE rebuildlinks";
 187         wfQuery( $sql, DB_WRITE );
 188 }
 189 ?>