maintenance/convertLinks.inc

   1 <?php
   2
   3 function convertLinks() {
   4         global $wgLang, $wgDBserver, $wgDBadminuser, $wgDBadminpassword, $wgDBname;
   5
   6         $numRows = $tuplesAdded = $numBadLinks = $curRowsRead = 0; #counters etc
   7         $totalTuplesInserted = 0; # total tuples INSERTed into links_temp
   8
   9         $reportCurReadProgress = true; #whether or not to give progress reports while reading IDs from cur table
  10         $curReadReportInterval = 1000; #number of rows between progress reports
  11
  12         $reportLinksConvProgress = true; #whether or not to give progress reports during conversion
  13         $linksConvInsertInterval = 1000; #number of rows per INSERT
  14
  15         $initialRowOffset = 0;
  16         #$finalRowOffset = 0; # not used yet; highest row number from links table to process
  17
  18         # Overwrite the old links table with the new one.  If this is set to false,
  19         # the new table will be left at links_temp.
  20         $overwriteLinksTable = true;
  21
  22         # Don't create keys, and so allow duplicates in the new links table.
  23         # This gives a huge speed improvement for very large links tables which are MyISAM. (What about InnoDB?)
  24         $noKeys = false;
  25
  26
  27         $logPerformance = false; # output performance data to a file
  28         $perfLogFilename = "convLinksPerf.txt";
  29         #--------------------------------------------------------------------
  30
  31         $res = wfQuery( "SELECT COUNT(*) AS count FROM links", DB_WRITE );
  32         $row = wfFetchObject($res);
  33         $numRows = $row->count;
  34         wfFreeResult( $res );
  35
  36         if ( $numRows == 0 ) {
  37                 print "No rows to convert. Updating schema...\n";
  38                 createTempTable();
  39         } else {
  40                 $row = wfFetchObject( $res );
  41                 if ( is_numeric( $row->l_from ) ) {
  42                         print "Schema already converted\n";
  43                         exit;
  44                 }
  45
  46                 if ( $logPerformance ) { $fh = fopen ( $perfLogFilename, "w" ); }
  47                 $baseTime = $startTime = getMicroTime();
  48                 # Create a title -> cur_id map
  49                 print "Loading IDs from cur table...\n";
  50                 performanceLog ( "Reading $numRows rows from cur table...\n" );
  51                 performanceLog ( "rows read vs seconds elapsed:\n" );
  52                 wfBufferSQLResults( false );
  53                 $res = wfQuery( "SELECT cur_namespace,cur_title,cur_id FROM cur", DB_WRITE );
  54                 $ids = array();
  55
  56                 while ( $row = wfFetchObject( $res ) ) {
  57                         $title = $row->cur_title;
  58                         if ( $row->cur_namespace ) {
  59                                 $title = $wgLang->getNsText( $row->cur_namespace ) . ":$title";
  60                         }
  61                         $ids[$title] = $row->cur_id;
  62                         $curRowsRead++;
  63                         if ($reportCurReadProgress) {
  64                                 if (($curRowsRead % $curReadReportInterval) == 0) {
  65                                         performanceLog( $curRowsRead . " " . (getMicroTime() - $baseTime) . "\n" );
  66                                         print "\t$curRowsRead rows of cur table read.\n";
  67                                 }
  68                         }
  69                 }
  70                 wfFreeResult( $res );
  71                 wfBufferSQLResults( true );
  72                 print "Finished loading IDs.\n\n";
  73                 performanceLog( "Took " . (getMicroTime() - $baseTime) . " seconds to load IDs.\n\n" );
  74         #--------------------------------------------------------------------
  75
  76                 # Now, step through the links table (in chunks of $linksConvInsertInterval rows),
  77                 # convert, and write to the new table.
  78                 createTempTable();
  79                 performanceLog( "Resetting timer.\n\n" );
  80                 $baseTime = getMicroTime();
  81                 print "Processing $numRows rows from links table...\n";
  82                 performanceLog( "Processing $numRows rows from links table...\n" );
  83                 performanceLog( "rows inserted vs seconds elapsed:\n" );
  84
  85                 for ($rowOffset = $initialRowOffset; $rowOffset < $numRows; $rowOffset += $linksConvInsertInterval) {
  86                         $sqlRead = "SELECT * FROM links LIMIT $linksConvInsertInterval OFFSET $rowOffset";
  87                         $res = wfQuery($sqlRead, DB_READ);
  88                         if ( $noKeys ) {
  89                                 $sqlWrite = array("INSERT INTO links_temp(l_from,l_to) VALUES ");
  90                         } else {
  91                                 $sqlWrite = array("INSERT IGNORE INTO links_temp(l_from,l_to) VALUES ");
  92                         }
  93
  94                         $tuplesAdded = 0; # no tuples added to INSERT yet
  95                         while ( $row = wfFetchObject($res) ) {
  96                                 $fromTitle = $row->l_from;
  97                                 if ( array_key_exists( $fromTitle, $ids ) ) { # valid title
  98                                         $from = $ids[$fromTitle];
  99                                         $to = $row->l_to;
 100                                         if ( $tuplesAdded != 0 ) {
 101                                                 $sqlWrite[] = ",";
 102                                         }
 103                                         $sqlWrite[] = "($from,$to)";
 104                                         $tuplesAdded++;
 105                                 } else { # invalid title
 106                                         $numBadLinks++;
 107                                 }
 108                         }
 109                         wfFreeResult($res);
 110                         #print "rowOffset: $rowOffset\ttuplesAdded: $tuplesAdded\tnumBadLinks: $numBadLinks\n";
 111                         if ( $tuplesAdded != 0  ) {
 112                                 if ($reportLinksConvProgress) {
 113                                         print "Inserting $tuplesAdded tuples into links_temp...";
 114                                 }
 115                                 wfQuery( implode("",$sqlWrite) , DB_WRITE );
 116                                 $totalTuplesInserted += $tuplesAdded;
 117                                 if ($reportLinksConvProgress)
 118                                         print " done. Total $totalTuplesInserted tuples inserted.\n";
 119                                         performanceLog( $totalTuplesInserted . " " . (getMicroTime() - $baseTime) . "\n"  );
 120                         }
 121                 }
 122                 print "$totalTuplesInserted valid titles and $numBadLinks invalid titles were processed.\n\n";
 123                 performanceLog( "$totalTuplesInserted valid titles and $numBadLinks invalid titles were processed.\n" );
 124                 performanceLog( "Total execution time: " . (getMicroTime() - $startTime) . " seconds.\n" );
 125                 if ( $logPerformance ) { fclose ( $fh ); }
 126         }
 127         #--------------------------------------------------------------------
 128
 129         if ( $overwriteLinksTable ) {
 130                 $dbConn = Database::newFromParams( $wgDBserver, $wgDBadminuser, $wgDBadminpassword, $wgDBname );
 131                 if (!($dbConn->isOpen())) {
 132                         print "Opening connection to database failed.\n";
 133                         exit;
 134                 }
 135                 # Check for existing links_backup, and delete it if it exists.
 136                 print "Dropping backup links table if it exists...";
 137                 $dbConn->query( "DROP TABLE IF EXISTS links_backup", DB_WRITE);
 138                 print " done.\n";
 139
 140                 # Swap in the new table, and move old links table to links_backup
 141                 print "Swapping tables 'links' to 'links_backup'; 'links_temp' to 'links'...";
 142                 $dbConn->query( "RENAME TABLE links TO links_backup, links_temp TO links", DB_WRITE );
 143                 print " done.\n\n";
 144
 145                 $dbConn->close();
 146                 print "Conversion complete. The old table remains at links_backup;\n";
 147                 print "delete at your leisure.\n";
 148         } else {
 149                 print "Conversion complete.  The converted table is at links_temp;\n";
 150                 print "the original links table is unchanged.\n";
 151         }
 152
 153 #--------------------------------------------------------------------
 154
 155 function createTempTable() {
 156         global $wgDBserver, $wgDBadminuser, $wgDBadminpassword, $wgDBname;
 157         global $noKeys;
 158         $dbConn = Database::newFromParams( $wgDBserver, $wgDBadminuser, $wgDBadminpassword, $wgDBname );
 159
 160         if (!($dbConn->isOpen())) {
 161                 print "Opening connection to database failed.\n";
 162                 exit;
 163         }
 164
 165         print "Dropping temporary links table if it exists...";
 166         $dbConn->query( "DROP TABLE IF EXISTS links_temp", DB_WRITE);
 167         print " done.\n";
 168
 169         print "Creating temporary links table...";
 170         if ( $noKeys ) {
 171                 $dbConn->query( "CREATE TABLE links_temp ( " .
 172                 "l_from int(8) unsigned NOT NULL default '0', " .
 173                 "l_to int(8) unsigned NOT NULL default '0')", DB_WRITE);
 174         } else {
 175                 $dbConn->query( "CREATE TABLE links_temp ( " .
 176                 "l_from int(8) unsigned NOT NULL default '0', " .
 177                 "l_to int(8) unsigned NOT NULL default '0', " .
 178                 "UNIQUE KEY l_from(l_from,l_to), " .
 179                 "KEY (l_to))", DB_WRITE);
 180         }
 181         print " done.\n\n";
 182 }
 183
 184 function performanceLog( $text ) {
 185         global $logPerformance, $fh;
 186         if ( $logPerformance ) {
 187                 fwrite( $fh, $text );
 188         }
 189 }
 190
 191 function getMicroTime() { # return time in seconds, with microsecond accuracy
 192         list($usec, $sec) = explode(" ", microtime());
 193         return ((float)$usec + (float)$sec);
 194 }
 195
 196
 197
 198 ?>