maintenance/convertLinks.inc

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