maintenance/convertLinks.inc

   1 <?php
   2 /**
   3  * @todo document
   4  * @addtogroup Maintenance
   5  */
   6
   7 /** */
   8 function convertLinks() {
   9         global $wgDBtype;
  10         if( $wgDBtype == 'postgres' ) {
  11                 print "Links table already ok on Postgres.\n";
  12                 return;
  13         }
  14
  15         print "Converting links table to ID-ID...\n";
  16
  17         global $wgLang, $wgDBserver, $wgDBadminuser, $wgDBadminpassword, $wgDBname;
  18         global $noKeys, $logPerformance, $fh;
  19
  20         $tuplesAdded = $numBadLinks = $curRowsRead = 0; #counters etc
  21         $totalTuplesInserted = 0; # total tuples INSERTed into links_temp
  22
  23         $reportCurReadProgress = true; #whether or not to give progress reports while reading IDs from cur table
  24         $curReadReportInterval = 1000; #number of rows between progress reports
  25
  26         $reportLinksConvProgress = true; #whether or not to give progress reports during conversion
  27         $linksConvInsertInterval = 1000; #number of rows per INSERT
  28
  29         $initialRowOffset = 0;
  30         #$finalRowOffset = 0; # not used yet; highest row number from links table to process
  31
  32         # Overwrite the old links table with the new one.  If this is set to false,
  33         # the new table will be left at links_temp.
  34         $overwriteLinksTable = true;
  35
  36         # Don't create keys, and so allow duplicates in the new links table.
  37         # This gives a huge speed improvement for very large links tables which are MyISAM. (What about InnoDB?)
  38         $noKeys = false;
  39
  40
  41         $logPerformance = false; # output performance data to a file
  42         $perfLogFilename = "convLinksPerf.txt";
  43         #--------------------------------------------------------------------
  44
  45         $dbw = wfGetDB( DB_MASTER );
  46         list ($cur, $links, $links_temp, $links_backup) = $dbw->tableNamesN( 'cur', 'links', 'links_temp', 'links_backup' );
  47
  48         $res = $dbw->query( "SELECT l_from FROM $links LIMIT 1" );
  49         if ( $dbw->fieldType( $res, 0 ) == "int" ) {
  50                 print "Schema already converted\n";
  51                 return;
  52         }
  53
  54         $res = $dbw->query( "SELECT COUNT(*) AS count FROM $links" );
  55         $row = $dbw->fetchObject($res);
  56         $numRows = $row->count;
  57         $dbw->freeResult( $res );
  58
  59         if ( $numRows == 0 ) {
  60                 print "Updating schema (no rows to convert)...\n";
  61                 createTempTable();
  62         } else {
  63                 if ( $logPerformance ) { $fh = fopen ( $perfLogFilename, "w" ); }
  64                 $baseTime = $startTime = getMicroTime();
  65                 # Create a title -> cur_id map
  66                 print "Loading IDs from $cur table...\n";
  67                 performanceLog ( "Reading $numRows rows from cur table...\n" );
  68                 performanceLog ( "rows read vs seconds elapsed:\n" );
  69
  70                 $dbw->bufferResults( false );
  71                 $res = $dbw->query( "SELECT cur_namespace,cur_title,cur_id FROM $cur" );
  72                 $ids = array();
  73
  74                 while ( $row = $dbw->fetchObject( $res ) ) {
  75                         $title = $row->cur_title;
  76                         if ( $row->cur_namespace ) {
  77                                 $title = $wgLang->getNsText( $row->cur_namespace ) . ":$title";
  78                         }
  79                         $ids[$title] = $row->cur_id;
  80                         $curRowsRead++;
  81                         if ($reportCurReadProgress) {
  82                                 if (($curRowsRead % $curReadReportInterval) == 0) {
  83                                         performanceLog( $curRowsRead . " " . (getMicroTime() - $baseTime) . "\n" );
  84                                         print "\t$curRowsRead rows of $cur table read.\n";
  85                                 }
  86                         }
  87                 }
  88                 $dbw->freeResult( $res );
  89                 $dbw->bufferResults( true );
  90                 print "Finished loading IDs.\n\n";
  91                 performanceLog( "Took " . (getMicroTime() - $baseTime) . " seconds to load IDs.\n\n" );
  92         #--------------------------------------------------------------------
  93
  94                 # Now, step through the links table (in chunks of $linksConvInsertInterval rows),
  95                 # convert, and write to the new table.
  96                 createTempTable();
  97                 performanceLog( "Resetting timer.\n\n" );
  98                 $baseTime = getMicroTime();
  99                 print "Processing $numRows rows from $links table...\n";
 100                 performanceLog( "Processing $numRows rows from $links table...\n" );
 101                 performanceLog( "rows inserted vs seconds elapsed:\n" );
 102
 103                 for ($rowOffset = $initialRowOffset; $rowOffset < $numRows; $rowOffset += $linksConvInsertInterval) {
 104                         $sqlRead = "SELECT * FROM $links ";
 105                         $sqlRead = $dbw->limitResult($sqlRead, $linksConvInsertInterval,$rowOffset);
 106                         $res = $dbw->query($sqlRead);
 107                         if ( $noKeys ) {
 108                                 $sqlWrite = array("INSERT INTO $links_temp (l_from,l_to) VALUES ");
 109                         } else {
 110                                 $sqlWrite = array("INSERT IGNORE INTO $links_temp (l_from,l_to) VALUES ");
 111                         }
 112
 113                         $tuplesAdded = 0; # no tuples added to INSERT yet
 114                         while ( $row = $dbw->fetchObject($res) ) {
 115                                 $fromTitle = $row->l_from;
 116                                 if ( array_key_exists( $fromTitle, $ids ) ) { # valid title
 117                                         $from = $ids[$fromTitle];
 118                                         $to = $row->l_to;
 119                                         if ( $tuplesAdded != 0 ) {
 120                                                 $sqlWrite[] = ",";
 121                                         }
 122                                         $sqlWrite[] = "($from,$to)";
 123                                         $tuplesAdded++;
 124                                 } else { # invalid title
 125                                         $numBadLinks++;
 126                                 }
 127                         }
 128                         $dbw->freeResult($res);
 129                         #print "rowOffset: $rowOffset\ttuplesAdded: $tuplesAdded\tnumBadLinks: $numBadLinks\n";
 130                         if ( $tuplesAdded != 0  ) {
 131                                 if ($reportLinksConvProgress) {
 132                                         print "Inserting $tuplesAdded tuples into $links_temp...";
 133                                 }
 134                                 $dbw->query( implode("",$sqlWrite) );
 135                                 $totalTuplesInserted += $tuplesAdded;
 136                                 if ($reportLinksConvProgress)
 137                                         print " done. Total $totalTuplesInserted tuples inserted.\n";
 138                                         performanceLog( $totalTuplesInserted . " " . (getMicroTime() - $baseTime) . "\n"  );
 139                         }
 140                 }
 141                 print "$totalTuplesInserted valid titles and $numBadLinks invalid titles were processed.\n\n";
 142                 performanceLog( "$totalTuplesInserted valid titles and $numBadLinks invalid titles were processed.\n" );
 143                 performanceLog( "Total execution time: " . (getMicroTime() - $startTime) . " seconds.\n" );
 144                 if ( $logPerformance ) { fclose ( $fh ); }
 145         }
 146         #--------------------------------------------------------------------
 147
 148         if ( $overwriteLinksTable ) {
 149                 $dbConn = Database::newFromParams( $wgDBserver, $wgDBadminuser, $wgDBadminpassword, $wgDBname );
 150                 if (!($dbConn->isOpen())) {
 151                         print "Opening connection to database failed.\n";
 152                         return;
 153                 }
 154                 # Check for existing links_backup, and delete it if it exists.
 155                 print "Dropping backup links table if it exists...";
 156                 $dbConn->query( "DROP TABLE IF EXISTS $links_backup", DB_MASTER);
 157                 print " done.\n";
 158
 159                 # Swap in the new table, and move old links table to links_backup
 160                 print "Swapping tables '$links' to '$links_backup'; '$links_temp' to '$links'...";
 161                 $dbConn->query( "RENAME TABLE links TO $links_backup, $links_temp TO $links", DB_MASTER );
 162                 print " done.\n\n";
 163
 164                 $dbConn->close();
 165                 print "Conversion complete. The old table remains at $links_backup;\n";
 166                 print "delete at your leisure.\n";
 167         } else {
 168                 print "Conversion complete.  The converted table is at $links_temp;\n";
 169                 print "the original links table is unchanged.\n";
 170         }
 171 }
 172
 173 #--------------------------------------------------------------------
 174
 175 function createTempTable() {
 176         global $wgDBserver, $wgDBadminuser, $wgDBadminpassword, $wgDBname;
 177         global $noKeys;
 178         $dbConn = Database::newFromParams( $wgDBserver, $wgDBadminuser, $wgDBadminpassword, $wgDBname );
 179
 180         if (!($dbConn->isOpen())) {
 181                 print "Opening connection to database failed.\n";
 182                 return;
 183         }
 184         $links_temp = $dbConn->tableName( 'links_temp' );
 185
 186         print "Dropping temporary links table if it exists...";
 187         $dbConn->query( "DROP TABLE IF EXISTS $links_temp");
 188         print " done.\n";
 189
 190         print "Creating temporary links table...";
 191         if ( $noKeys ) {
 192                 $dbConn->query( "CREATE TABLE $links_temp ( " .
 193                 "l_from int(8) unsigned NOT NULL default '0', " .
 194                 "l_to int(8) unsigned NOT NULL default '0')");
 195         } else {
 196                 $dbConn->query( "CREATE TABLE $links_temp ( " .
 197                 "l_from int(8) unsigned NOT NULL default '0', " .
 198                 "l_to int(8) unsigned NOT NULL default '0', " .
 199                 "UNIQUE KEY l_from(l_from,l_to), " .
 200                 "KEY (l_to))");
 201         }
 202         print " done.\n\n";
 203 }
 204
 205 function performanceLog( $text ) {
 206         global $logPerformance, $fh;
 207         if ( $logPerformance ) {
 208                 fwrite( $fh, $text );
 209         }
 210 }
 211
 212 function getMicroTime() { # return time in seconds, with microsecond accuracy
 213         list($usec, $sec) = explode(" ", microtime());
 214         return ((float)$usec + (float)$sec);
 215 }
 216
 217
 218
 219 ?>