diff options
Diffstat (limited to 'maintenance/convertLinks.inc')
-rw-r--r-- | maintenance/convertLinks.inc | 218 |
1 files changed, 0 insertions, 218 deletions
diff --git a/maintenance/convertLinks.inc b/maintenance/convertLinks.inc deleted file mode 100644 index 7c7b8aff..00000000 --- a/maintenance/convertLinks.inc +++ /dev/null @@ -1,218 +0,0 @@ -<?php -/** - * @file - * @todo document - * @ingroup Maintenance - */ - -/** */ -function convertLinks() { - global $wgDBtype; - if( $wgDBtype == 'postgres' ) { - wfOut( "Links table already ok on Postgres.\n" ); - return; - } - - wfOut( "Converting links table to ID-ID...\n" ); - - global $wgLang, $wgDBserver, $wgDBadminuser, $wgDBadminpassword, $wgDBname; - global $noKeys, $logPerformance, $fh; - - $tuplesAdded = $numBadLinks = $curRowsRead = 0; #counters etc - $totalTuplesInserted = 0; # total tuples INSERTed into links_temp - - $reportCurReadProgress = true; #whether or not to give progress reports while reading IDs from cur table - $curReadReportInterval = 1000; #number of rows between progress reports - - $reportLinksConvProgress = true; #whether or not to give progress reports during conversion - $linksConvInsertInterval = 1000; #number of rows per INSERT - - $initialRowOffset = 0; - #$finalRowOffset = 0; # not used yet; highest row number from links table to process - - # Overwrite the old links table with the new one. If this is set to false, - # the new table will be left at links_temp. - $overwriteLinksTable = true; - - # Don't create keys, and so allow duplicates in the new links table. - # This gives a huge speed improvement for very large links tables which are MyISAM. (What about InnoDB?) - $noKeys = false; - - - $logPerformance = false; # output performance data to a file - $perfLogFilename = "convLinksPerf.txt"; - #-------------------------------------------------------------------- - - $dbw = wfGetDB( DB_MASTER ); - list ($cur, $links, $links_temp, $links_backup) = $dbw->tableNamesN( 'cur', 'links', 'links_temp', 'links_backup' ); - - // Get database-agnostic limit clause - $sql_limit = $dbw->limitResult( "SELECT l_from FROM $links", 1 ); - $res = $dbw->query( $sql_limit ); - if ( $dbw->fieldType( $res, 0 ) == "int" ) { - wfOut( "Schema already converted\n" ); - return; - } - - $res = $dbw->query( "SELECT COUNT(*) AS count FROM $links" ); - $row = $dbw->fetchObject($res); - $numRows = $row->count; - $dbw->freeResult( $res ); - - if ( $numRows == 0 ) { - wfOut( "Updating schema (no rows to convert)...\n" ); - createTempTable(); - } else { - if ( $logPerformance ) { $fh = fopen ( $perfLogFilename, "w" ); } - $baseTime = $startTime = getMicroTime(); - # Create a title -> cur_id map - wfOut( "Loading IDs from $cur table...\n" ); - performanceLog ( "Reading $numRows rows from cur table...\n" ); - performanceLog ( "rows read vs seconds elapsed:\n" ); - - $dbw->bufferResults( false ); - $res = $dbw->query( "SELECT cur_namespace,cur_title,cur_id FROM $cur" ); - $ids = array(); - - while ( $row = $dbw->fetchObject( $res ) ) { - $title = $row->cur_title; - if ( $row->cur_namespace ) { - $title = $wgLang->getNsText( $row->cur_namespace ) . ":$title"; - } - $ids[$title] = $row->cur_id; - $curRowsRead++; - if ($reportCurReadProgress) { - if (($curRowsRead % $curReadReportInterval) == 0) { - performanceLog( $curRowsRead . " " . (getMicroTime() - $baseTime) . "\n" ); - wfOut( "\t$curRowsRead rows of $cur table read.\n" ); - } - } - } - $dbw->freeResult( $res ); - $dbw->bufferResults( true ); - wfOut( "Finished loading IDs.\n\n" ); - performanceLog( "Took " . (getMicroTime() - $baseTime) . " seconds to load IDs.\n\n" ); - #-------------------------------------------------------------------- - - # Now, step through the links table (in chunks of $linksConvInsertInterval rows), - # convert, and write to the new table. - createTempTable(); - performanceLog( "Resetting timer.\n\n" ); - $baseTime = getMicroTime(); - wfOut( "Processing $numRows rows from $links table...\n" ); - performanceLog( "Processing $numRows rows from $links table...\n" ); - performanceLog( "rows inserted vs seconds elapsed:\n" ); - - for ($rowOffset = $initialRowOffset; $rowOffset < $numRows; $rowOffset += $linksConvInsertInterval) { - $sqlRead = "SELECT * FROM $links "; - $sqlRead = $dbw->limitResult($sqlRead, $linksConvInsertInterval,$rowOffset); - $res = $dbw->query($sqlRead); - if ( $noKeys ) { - $sqlWrite = array("INSERT INTO $links_temp (l_from,l_to) VALUES "); - } else { - $sqlWrite = array("INSERT IGNORE INTO $links_temp (l_from,l_to) VALUES "); - } - - $tuplesAdded = 0; # no tuples added to INSERT yet - while ( $row = $dbw->fetchObject($res) ) { - $fromTitle = $row->l_from; - if ( array_key_exists( $fromTitle, $ids ) ) { # valid title - $from = $ids[$fromTitle]; - $to = $row->l_to; - if ( $tuplesAdded != 0 ) { - $sqlWrite[] = ","; - } - $sqlWrite[] = "($from,$to)"; - $tuplesAdded++; - } else { # invalid title - $numBadLinks++; - } - } - $dbw->freeResult($res); - #wfOut( "rowOffset: $rowOffset\ttuplesAdded: $tuplesAdded\tnumBadLinks: $numBadLinks\n" ); - if ( $tuplesAdded != 0 ) { - if ($reportLinksConvProgress) { - wfOut( "Inserting $tuplesAdded tuples into $links_temp..." ); - } - $dbw->query( implode("",$sqlWrite) ); - $totalTuplesInserted += $tuplesAdded; - if ($reportLinksConvProgress) - wfOut( " done. Total $totalTuplesInserted tuples inserted.\n" ); - performanceLog( $totalTuplesInserted . " " . (getMicroTime() - $baseTime) . "\n" ); - } - } - wfOut( "$totalTuplesInserted valid titles and $numBadLinks invalid titles were processed.\n\n" ); - performanceLog( "$totalTuplesInserted valid titles and $numBadLinks invalid titles were processed.\n" ); - performanceLog( "Total execution time: " . (getMicroTime() - $startTime) . " seconds.\n" ); - if ( $logPerformance ) { fclose ( $fh ); } - } - #-------------------------------------------------------------------- - - if ( $overwriteLinksTable ) { - $dbConn = Database::newFromParams( $wgDBserver, $wgDBadminuser, $wgDBadminpassword, $wgDBname ); - if (!($dbConn->isOpen())) { - wfOut( "Opening connection to database failed.\n" ); - return; - } - # Check for existing links_backup, and delete it if it exists. - wfOut( "Dropping backup links table if it exists..." ); - $dbConn->query( "DROP TABLE IF EXISTS $links_backup", DB_MASTER); - wfOut( " done.\n" ); - - # Swap in the new table, and move old links table to links_backup - wfOut( "Swapping tables '$links' to '$links_backup'; '$links_temp' to '$links'..." ); - $dbConn->query( "RENAME TABLE links TO $links_backup, $links_temp TO $links", DB_MASTER ); - wfOut( " done.\n\n" ); - - $dbConn->close(); - wfOut( "Conversion complete. The old table remains at $links_backup;\n" ); - wfOut( "delete at your leisure.\n" ); - } else { - wfOut( "Conversion complete. The converted table is at $links_temp;\n" ); - wfOut( "the original links table is unchanged.\n" ); - } -} - -#-------------------------------------------------------------------- - -function createTempTable() { - global $wgDBserver, $wgDBadminuser, $wgDBadminpassword, $wgDBname; - global $noKeys; - $dbConn = Database::newFromParams( $wgDBserver, $wgDBadminuser, $wgDBadminpassword, $wgDBname ); - - if (!($dbConn->isOpen())) { - wfOut( "Opening connection to database failed.\n" ); - return; - } - $links_temp = $dbConn->tableName( 'links_temp' ); - - wfOut( "Dropping temporary links table if it exists..." ); - $dbConn->query( "DROP TABLE IF EXISTS $links_temp"); - wfOut( " done.\n" ); - - wfOut( "Creating temporary links table..." ); - if ( $noKeys ) { - $dbConn->query( "CREATE TABLE $links_temp ( " . - "l_from int(8) unsigned NOT NULL default '0', " . - "l_to int(8) unsigned NOT NULL default '0')"); - } else { - $dbConn->query( "CREATE TABLE $links_temp ( " . - "l_from int(8) unsigned NOT NULL default '0', " . - "l_to int(8) unsigned NOT NULL default '0', " . - "UNIQUE KEY l_from(l_from,l_to), " . - "KEY (l_to))"); - } - wfOut( " done.\n\n" ); -} - -function performanceLog( $text ) { - global $logPerformance, $fh; - if ( $logPerformance ) { - fwrite( $fh, $text ); - } -} - -function getMicroTime() { # return time in seconds, with microsecond accuracy - list($usec, $sec) = explode(" ", microtime()); - return ((float)$usec + (float)$sec); -} |