maintenance/convertLinks.inc

   1 <?php
   2 /**
   3  * @todo document
   4  * @package MediaWiki
   5  * @subpackage Maintenance
   6  */
   7
   8 /** */
   9 function convertLinks() {
  10         print "Converting links table to ID-ID...\n";
  11
  12         global $wgLang, $wgDBserver, $wgDBadminuser, $wgDBadminpassword, $wgDBname;
  13         global $noKeys, $logPerformance, $fh;
  14
  15         $numRows = $tuplesAdded = $numBadLinks = $curRowsRead = 0; #counters etc
  16         $totalTuplesInserted = 0; # total tuples INSERTed into links_temp
  17
  18         $reportCurReadProgress = true; #whether or not to give progress reports while reading IDs from cur table
  19         $curReadReportInterval = 1000; #number of rows between progress reports
  20
  21         $reportLinksConvProgress = true; #whether or not to give progress reports during conversion
  22         $linksConvInsertInterval = 1000; #number of rows per INSERT
  23
  24         $initialRowOffset = 0;
  25         #$finalRowOffset = 0; # not used yet; highest row number from links table to process
  26
  27         # Overwrite the old links table with the new one.  If this is set to false,
  28         # the new table will be left at links_temp.
  29         $overwriteLinksTable = true;
  30
  31         # Don't create keys, and so allow duplicates in the new links table.
  32         # This gives a huge speed improvement for very large links tables which are MyISAM. (What about InnoDB?)
  33         $noKeys = false;
  34
  35
  36         $logPerformance = false; # output performance data to a file
  37         $perfLogFilename = "convLinksPerf.txt";
  38         #--------------------------------------------------------------------
  39
  40         $dbw =& wfGetDB( DB_MASTER );
  41         extract( $dbw->tableNames( 'cur', 'links', 'links_temp', 'links_backup' ) );
  42
  43         $res = $dbw->query( "SELECT l_from FROM $links LIMIT 1" );
  44         if ( $dbw->fieldType( $res, 0 ) == "int" ) {
  45                 print "Schema already converted\n";
  46                 return;
  47         }
  48
  49         $res = $dbw->query( "SELECT COUNT(*) AS count FROM $links" );
  50         $row = $dbw->fetchObject($res);
  51         $numRows = $row->count;
  52         $dbw->freeResult( $res );
  53
  54         if ( $numRows == 0 ) {
  55                 print "Updating schema (no rows to convert)...\n";
  56                 createTempTable();
  57         } else {
  58                 if ( $logPerformance ) { $fh = fopen ( $perfLogFilename, "w" ); }
  59                 $baseTime = $startTime = getMicroTime();
  60                 # Create a title -> cur_id map
  61                 print "Loading IDs from $cur table...\n";
  62                 performanceLog ( "Reading $numRows rows from cur table...\n" );
  63                 performanceLog ( "rows read vs seconds elapsed:\n" );
  64
  65                 $dbw->setBufferResults( false );
  66                 $res = $dbw->query( "SELECT cur_namespace,cur_title,cur_id FROM $cur" );
  67                 $ids = array();
  68
  69                 while ( $row = $dbw->fetchObject( $res ) ) {
  70                         $title = $row->cur_title;
  71                         if ( $row->cur_namespace ) {
  72                                 $title = $wgLang->getNsText( $row->cur_namespace ) . ":$title";
  73                         }
  74                         $ids[$title] = $row->cur_id;
  75                         $curRowsRead++;
  76                         if ($reportCurReadProgress) {
  77                                 if (($curRowsRead % $curReadReportInterval) == 0) {
  78                                         performanceLog( $curRowsRead . " " . (getMicroTime() - $baseTime) . "\n" );
  79                                         print "\t$curRowsRead rows of $cur table read.\n";
  80                                 }
  81                         }
  82                 }
  83                 $dbw->freeResult( $res );
  84                 $dbw->setBufferResults( true );
  85                 print "Finished loading IDs.\n\n";
  86                 performanceLog( "Took " . (getMicroTime() - $baseTime) . " seconds to load IDs.\n\n" );
  87         #--------------------------------------------------------------------
  88
  89                 # Now, step through the links table (in chunks of $linksConvInsertInterval rows),
  90                 # convert, and write to the new table.
  91                 createTempTable();
  92                 performanceLog( "Resetting timer.\n\n" );
  93                 $baseTime = getMicroTime();
  94                 print "Processing $numRows rows from $links table...\n";
  95                 performanceLog( "Processing $numRows rows from $links table...\n" );
  96                 performanceLog( "rows inserted vs seconds elapsed:\n" );
  97
  98                 for ($rowOffset = $initialRowOffset; $rowOffset < $numRows; $rowOffset += $linksConvInsertInterval) {
  99                         $sqlRead = "SELECT * FROM $links ".wfLimitResult($linksConvInsertInterval,$rowOffset);
 100                         $res = $dbw->query($sqlRead);
 101                         if ( $noKeys ) {
 102                                 $sqlWrite = array("INSERT INTO $links_temp (l_from,l_to) VALUES ");
 103                         } else {
 104                                 $sqlWrite = array("INSERT IGNORE INTO $links_temp (l_from,l_to) VALUES ");
 105                         }
 106
 107                         $tuplesAdded = 0; # no tuples added to INSERT yet
 108                         while ( $row = $dbw->fetchObject($res) ) {
 109                                 $fromTitle = $row->l_from;
 110                                 if ( array_key_exists( $fromTitle, $ids ) ) { # valid title
 111                                         $from = $ids[$fromTitle];
 112                                         $to = $row->l_to;
 113                                         if ( $tuplesAdded != 0 ) {
 114                                                 $sqlWrite[] = ",";
 115                                         }
 116                                         $sqlWrite[] = "($from,$to)";
 117                                         $tuplesAdded++;
 118                                 } else { # invalid title
 119                                         $numBadLinks++;
 120                                 }
 121                         }
 122                         $dbw->freeResult($res);
 123                         #print "rowOffset: $rowOffset\ttuplesAdded: $tuplesAdded\tnumBadLinks: $numBadLinks\n";
 124                         if ( $tuplesAdded != 0  ) {
 125                                 if ($reportLinksConvProgress) {
 126                                         print "Inserting $tuplesAdded tuples into $links_temp...";
 127                                 }
 128                                 $dbw->query( implode("",$sqlWrite) );
 129                                 $totalTuplesInserted += $tuplesAdded;
 130                                 if ($reportLinksConvProgress)
 131                                         print " done. Total $totalTuplesInserted tuples inserted.\n";
 132                                         performanceLog( $totalTuplesInserted . " " . (getMicroTime() - $baseTime) . "\n"  );
 133                         }
 134                 }
 135                 print "$totalTuplesInserted valid titles and $numBadLinks invalid titles were processed.\n\n";
 136                 performanceLog( "$totalTuplesInserted valid titles and $numBadLinks invalid titles were processed.\n" );
 137                 performanceLog( "Total execution time: " . (getMicroTime() - $startTime) . " seconds.\n" );
 138                 if ( $logPerformance ) { fclose ( $fh ); }
 139         }
 140         #--------------------------------------------------------------------
 141
 142         if ( $overwriteLinksTable ) {
 143                 $dbConn = Database::newFromParams( $wgDBserver, $wgDBadminuser, $wgDBadminpassword, $wgDBname );
 144                 if (!($dbConn->isOpen())) {
 145                         print "Opening connection to database failed.\n";
 146                         return;
 147                 }
 148                 # Check for existing links_backup, and delete it if it exists.
 149                 print "Dropping backup links table if it exists...";
 150                 $dbConn->query( "DROP TABLE IF EXISTS $links_backup", DB_MASTER);
 151                 print " done.\n";
 152
 153                 # Swap in the new table, and move old links table to links_backup
 154                 print "Swapping tables '$links' to '$links_backup'; '$links_temp' to '$links'...";
 155                 $dbConn->query( "RENAME TABLE links TO $links_backup, $links_temp TO $links", DB_MASTER );
 156                 print " done.\n\n";
 157
 158                 $dbConn->close();
 159                 print "Conversion complete. The old table remains at $links_backup;\n";
 160                 print "delete at your leisure.\n";
 161         } else {
 162                 print "Conversion complete.  The converted table is at $links_temp;\n";
 163                 print "the original links table is unchanged.\n";
 164         }
 165 }
 166
 167 #--------------------------------------------------------------------
 168
 169 function createTempTable() {
 170         global $wgDBserver, $wgDBadminuser, $wgDBadminpassword, $wgDBname;
 171         global $noKeys;
 172         $dbConn = Database::newFromParams( $wgDBserver, $wgDBadminuser, $wgDBadminpassword, $wgDBname );
 173
 174         if (!($dbConn->isOpen())) {
 175                 print "Opening connection to database failed.\n";
 176                 return;
 177         }
 178         $links_temp = $dbConn->tableName( 'links_temp' );
 179
 180         print "Dropping temporary links table if it exists...";
 181         $dbConn->query( "DROP TABLE IF EXISTS $links_temp");
 182         print " done.\n";
 183
 184         print "Creating temporary links table...";
 185         if ( $noKeys ) {
 186                 $dbConn->query( "CREATE TABLE $links_temp ( " .
 187                 "l_from int(8) unsigned NOT NULL default '0', " .
 188                 "l_to int(8) unsigned NOT NULL default '0')");
 189         } else {
 190                 $dbConn->query( "CREATE TABLE $links_temp ( " .
 191                 "l_from int(8) unsigned NOT NULL default '0', " .
 192                 "l_to int(8) unsigned NOT NULL default '0', " .
 193                 "UNIQUE KEY l_from(l_from,l_to), " .
 194                 "KEY (l_to))");
 195         }
 196         print " done.\n\n";
 197 }
 198
 199 function performanceLog( $text ) {
 200         global $logPerformance, $fh;
 201         if ( $logPerformance ) {
 202                 fwrite( $fh, $text );
 203         }
 204 }
 205
 206 function getMicroTime() { # return time in seconds, with microsecond accuracy
 207         list($usec, $sec) = explode(" ", microtime());
 208         return ((float)$usec + (float)$sec);
 209 }
 210
 211
 212
 213 ?>