maintenance/convertLinks.php

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