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 # Turn this on if you've got memory to burn
   9 $wgUseMemoryTables = false;
  10
  11 function rebuildLinkTablesPass1()
  12 {
  13         global $wgLang, $wgUseMemoryTables;
  14         $count = 0;
  15         print "Rebuilding link tables (pass 1).\n";
  16
  17         $sql = "DROP TABLE IF EXISTS rebuildlinks";
  18         wfQuery( $sql, DB_WRITE );
  19
  20         $sql = "CREATE TABLE rebuildlinks (
  21   rl_f_id int(8) unsigned NOT NULL default 0,
  22   rl_f_title varchar(255) binary NOT NULL default '',
  23   rl_to varchar(255) binary NOT NULL default '',
  24   INDEX rl_to (rl_to) )";
  25         if( $wgUseMemoryTables ) $sql .= " TYPE=heap";
  26         wfQuery( $sql, DB_WRITE );
  27
  28         $sql = "LOCK TABLES cur READ, rebuildlinks WRITE, interwiki READ, user_newtalk READ";
  29         wfQuery( $sql, DB_WRITE );
  30
  31         $sql = "DELETE FROM rebuildlinks";
  32         wfQuery( $sql, DB_WRITE );
  33
  34         $sql = "SELECT cur_id,cur_namespace,cur_title,cur_text FROM cur";
  35         $res = wfQuery( $sql, DB_WRITE );
  36         $total = wfNumRows( $res );
  37
  38         $tc = Title::legalChars();
  39         while ( $row = wfFetchObject( $res ) ) {
  40                 $id = $row->cur_id;
  41                 $ns = $wgLang->getNsText( $row->cur_namespace );
  42                 if ( "" == $ns ) {
  43                         $title = addslashes( $row->cur_title );
  44                 } else {
  45                         $title = addslashes( "$ns:{$row->cur_title}" );
  46                 }
  47                 $text = $row->cur_text;
  48                 $numlinks = preg_match_all( "/\\[\\[([{$tc}]+)(]|\\|)/", $text,
  49                   $m, PREG_PATTERN_ORDER );
  50
  51                 if ( 0 != $numlinks ) {
  52                         $first = true;
  53                         $sql = "INSERT INTO rebuildlinks (rl_f_id,rl_f_title,rl_to) VALUES ";
  54                         for ( $i = 0; $i < $numlinks; ++$i ) {
  55                                 if( preg_match( '/^(http|https|ftp|mailto|news):/', $m[1][$i] ) ) {
  56                                         # an URL link; not for us!
  57                                         continue;
  58                                 }
  59                                 # FIXME: Handle subpage links
  60                                 $nt = Title::newFromText( $m[1][$i] );
  61                                 if (! $nt)
  62                                 {
  63                                         $txt = $m[1][$i];
  64                                         print "error in '$ns:{$row->cur_title}' :\t'$txt'\n";
  65                                         continue;
  66                                 }
  67                                 if( $nt->getInterwiki() != "" ) {
  68                                         # Interwiki links are not stored in the link tables
  69                                         continue;
  70                                 }
  71                                 if( $nt->getNamespace() == Namespace::getSpecial() ) {
  72                                         # Special links not stored in link tables
  73                                         continue;
  74                                 }
  75                                 if( $nt->getNamespace() == Namespace::getMedia() ) {
  76                                         # treat media: links as image: links
  77                                         $nt = Title::makeTitle( Namespace::getImage(), $nt->getDBkey() );
  78                                 }
  79
  80                                 if (!$first)
  81                                         $sql .= ",";
  82                                 else
  83                                         $first = false;
  84
  85                                 $dest = addslashes( $nt->getPrefixedDBkey() );
  86                                 $sql .= "({$id},'{$title}','{$dest}')";
  87                         }
  88
  89                         if (! $first) { wfQuery( $sql, DB_WRITE  ); }
  90                 }
  91                 if ( ( ++$count % 1000 ) == 0 ) {
  92                         print "$count of $total articles scanned.\n";
  93                 }
  94         }
  95         print "$total articles scanned.\n";
  96         mysql_free_result( $res );
  97
  98         $sql = "UNLOCK TABLES";
  99         wfQuery( $sql, DB_WRITE );
 100 }
 101
 102 function rebuildLinkTablesPass2()
 103 {
 104         global $wgLang;
 105         $count = 0;
 106         print "Rebuilding link tables (pass 2).\n";
 107
 108         $sql = "LOCK TABLES cur READ, rebuildlinks READ, interwiki READ, " .
 109           "links WRITE, brokenlinks WRITE, imagelinks WRITE";
 110         wfQuery( $sql, DB_WRITE );
 111
 112         $sql = "DELETE FROM links";
 113         wfQuery( $sql, DB_WRITE );
 114
 115         $sql = "DELETE FROM brokenlinks";
 116         wfQuery( $sql, DB_WRITE );
 117
 118         $sql = "DELETE FROM imagelinks";
 119         wfQuery( $sql, DB_WRITE );
 120
 121         $ins = $wgLang->getNsText( Namespace::getImage() );
 122         $inslen = strlen($ins)+1;
 123         $sql = "SELECT rl_f_title,rl_to FROM rebuildlinks " .
 124           "WHERE rl_to LIKE '$ins:%'";
 125         $res = wfQuery( $sql, DB_WRITE );
 126
 127         $sql = "INSERT INTO imagelinks (il_from,il_to) VALUES ";
 128         $first = true;
 129         while ( $row = wfFetchObject( $res ) )
 130         {
 131                 $iname = addslashes( substr( $row->rl_to, $inslen ) );
 132                 $pname = addslashes( $row->rl_f_title );
 133
 134                 if ( ! $first )
 135                         $sql .= ",";
 136                 else
 137                         $first = false;
 138
 139                 $sql .= "('{$pname}','{$iname}')";
 140         }
 141         wfFreeResult( $res );
 142         if ( ! $first ) { wfQuery( $sql, DB_WRITE ); }
 143
 144         $sql = "SELECT DISTINCT rl_to FROM rebuildlinks ORDER BY rl_to";
 145         $res = wfQuery( $sql, DB_WRITE );
 146         $count = 0;
 147         $total = wfNumRows( $res );
 148
 149         while ( $row = wfFetchObject( $res ) ) {
 150                 if ( 0 == strncmp( "$ins:", $row->rl_to, $inslen ) ) { continue; }
 151
 152                 $nt = Title::newFromDBkey( $row->rl_to );
 153                 if (! $nt)
 154                 {
 155                         print "error pass2: '{$row->rl_to}'\n";
 156                         continue;
 157                 }
 158                 $id = $nt->getArticleID();
 159                 $to = addslashes( $row->rl_to );
 160
 161                 if ( 0 == $id ) {
 162                         $sql = "SELECT DISTINCT rl_f_id FROM rebuildlinks WHERE rl_to='{$to}'";
 163                         $res2 = wfQuery( $sql, DB_WRITE );
 164
 165                         $sql = "INSERT INTO brokenlinks (bl_from,bl_to) VALUES ";
 166                         $first = true;
 167                         while ( $row2 = wfFetchObject( $res2 ) )
 168                         {
 169                                 if (! $first)
 170                                         $sql .= ",";
 171                                 else
 172                                         $first = false;
 173
 174                                 $from = $row2->rl_f_id;
 175                                 $sql .= "({$from},'{$to}')";
 176                         }
 177                         wfFreeResult( $res2 );
 178                         if ( ! $first ) { wfQuery( $sql, DB_WRITE ); }
 179                 } else {
 180                         $sql = "SELECT DISTINCT rl_f_title FROM rebuildlinks WHERE rl_to='{$to}'";
 181                         $res2 = wfQuery( $sql, DB_WRITE );
 182
 183                         $sql = "INSERT INTO links (l_from,l_to) VALUES ";
 184                         $first = true;
 185                         while ( $row2 = wfFetchObject( $res2 ) )
 186                         {
 187                                 if (! $first)
 188                                         $sql .= ",";
 189                                 else
 190                                         $first = false;
 191
 192                                 $from = addslashes( $row2->rl_f_title );
 193                                 $sql .= "('{$from}',{$id})";
 194                         }
 195                         wfFreeResult( $res2 );
 196                         if ( ! $first ) { wfQuery( $sql, DB_WRITE ); }
 197                 }
 198                 if ( ( ++$count % 1000 ) == 0 ) {
 199                         print "$count of $total titles processed.\n";
 200                 }
 201         }
 202         wfFreeResult( $res );
 203
 204         $sql = "UNLOCK TABLES";
 205         wfQuery( $sql, DB_WRITE );
 206
 207         $sql = "DROP TABLE rebuildlinks";
 208         wfQuery( $sql, DB_WRITE );
 209 }
 210 ?>