X-Git-Url: https://git.cyclocoop.org/?a=blobdiff_plain;f=maintenance%2FconvertLinks.php;h=5f7b02e4237d599a7c844d64ce8c32f30569766b;hb=fcdf2518605954fe1bcf88598fc14bd95d525882;hp=e792319ddae4c2da713744c4c6ad3ae2ef60edc6;hpb=e819779400cb435b68af78542f2d4041acdb0e0a;p=lhc%2Fweb%2Fwiklou.git diff --git a/maintenance/convertLinks.php b/maintenance/convertLinks.php index e792319dda..5f7b02e423 100644 --- a/maintenance/convertLinks.php +++ b/maintenance/convertLinks.php @@ -1,7 +1,6 @@ ID) to the new schema (ID->ID) - * The wiki should be put into read-only mode while this script executes + * Convert from the old links schema (string->ID) to the new schema (ID->ID). * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -18,21 +17,37 @@ * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. * http://www.gnu.org/copyleft/gpl.html * + * @file * @ingroup Maintenance */ -require_once( dirname( __FILE__ ) . '/Maintenance.php' ); +require_once( __DIR__ . '/Maintenance.php' ); +/** + * Maintenance script to convert from the old links schema (string->ID) + * to the new schema (ID->ID). + * + * The wiki should be put into read-only mode while this script executes. + * + * @ingroup Maintenance + */ class ConvertLinks extends Maintenance { + private $logPerformance; public function __construct() { parent::__construct(); - $this->mDescription = "Convert from the old links schema (string->ID) to the new schema (ID->ID) + $this->mDescription = "Convert from the old links schema (string->ID) to the new schema (ID->ID). The wiki should be put into read-only mode while this script executes"; + + $this->addArg( 'logperformance', "Log performance to perfLogFilename.", false ); + $this->addArg( 'perfLogFilename', "Filename where performance is logged if --logperformance was set (defaults to 'convLinksPerf.txt').", false ); + $this->addArg( 'keep-links-table', "Don't overwrite the old links table with the new one, leave the new table at links_temp.", false ); + $this->addArg( 'nokeys', "Don't create keys, and so allow duplicates in the new links table.\n +This gives a huge speed improvement for very large links tables which are MyISAM." /* (What about InnoDB?) */, false ); } public function getDbType() { - return self::DB_ADMIN; + return Maintenance::DB_ADMIN; } public function execute() { @@ -44,9 +59,9 @@ The wiki should be put into read-only mode while this script executes"; return; } - global $wgLang, $noKeys, $logPerformance, $fh; - - $tuplesAdded = $numBadLinks = $curRowsRead = 0; # counters etc + global $wgContLang; + + $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 @@ -58,17 +73,11 @@ The wiki should be put into read-only mode while this script executes"; $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"; + $overwriteLinksTable = !$this->hasOption( 'keep-links-table' ); + $noKeys = $this->hasOption( 'noKeys' ); + $this->logPerformance = $this->hasOption( 'logperformance' ); + $perfLogFilename = $this->getArg( 'perfLogFilename', "convLinksPerf.txt" ); + # -------------------------------------------------------------------- list ( $cur, $links, $links_temp, $links_backup ) = $dbw->tableNamesN( 'cur', 'links', 'links_temp', 'links_backup' ); @@ -77,43 +86,50 @@ The wiki should be put into read-only mode while this script executes"; $this->output( "...have pagelinks; skipping old links table updates\n" ); return; } - + $res = $dbw->query( "SELECT l_from FROM $links LIMIT 1" ); if ( $dbw->fieldType( $res, 0 ) == "int" ) { $this->output( "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 ) { $this->output( "Updating schema (no rows to convert)...\n" ); $this->createTempTable(); } else { - if ( $logPerformance ) { $fh = fopen ( $perfLogFilename, "w" ); } + $fh = false; + if ( $this->logPerformance ) { + $fh = fopen ( $perfLogFilename, "w" ); + if ( !$fh ) { + $this->error( "Couldn't open $perfLogFilename" ); + $this->logPerformance = false; + } + } $baseTime = $startTime = $this->getMicroTime(); # Create a title -> cur_id map $this->output( "Loading IDs from $cur table...\n" ); - $this->performanceLog ( "Reading $numRows rows from cur table...\n" ); - $this->performanceLog ( "rows read vs seconds elapsed:\n" ); + $this->performanceLog ( $fh, "Reading $numRows rows from cur table...\n" ); + $this->performanceLog ( $fh, "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 ) ) { + foreach ( $res as $row ) { $title = $row->cur_title; if ( $row->cur_namespace ) { - $title = $wgLang->getNsText( $row->cur_namespace ) . ":$title"; + $title = $wgContLang->getNsText( $row->cur_namespace ) . ":$title"; } $ids[$title] = $row->cur_id; $curRowsRead++; if ( $reportCurReadProgress ) { if ( ( $curRowsRead % $curReadReportInterval ) == 0 ) { - $this->performanceLog( $curRowsRead . " " . ( $this->getMicroTime() - $baseTime ) . "\n" ); + $this->performanceLog( $fh, $curRowsRead . " " . ( $this->getMicroTime() - $baseTime ) . "\n" ); $this->output( "\t$curRowsRead rows of $cur table read.\n" ); } } @@ -121,19 +137,19 @@ The wiki should be put into read-only mode while this script executes"; $dbw->freeResult( $res ); $dbw->bufferResults( true ); $this->output( "Finished loading IDs.\n\n" ); - $this->performanceLog( "Took " . ( $this->getMicroTime() - $baseTime ) . " seconds to load IDs.\n\n" ); + $this->performanceLog( $fh, "Took " . ( $this->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. $this->createTempTable(); - $this->performanceLog( "Resetting timer.\n\n" ); + $this->performanceLog( $fh, "Resetting timer.\n\n" ); $baseTime = $this->getMicroTime(); $this->output( "Processing $numRows rows from $links table...\n" ); - $this->performanceLog( "Processing $numRows rows from $links table...\n" ); - $this->performanceLog( "rows inserted vs seconds elapsed:\n" ); - + $this->performanceLog( $fh, "Processing $numRows rows from $links table...\n" ); + $this->performanceLog( $fh, "rows inserted vs seconds elapsed:\n" ); + for ( $rowOffset = $initialRowOffset; $rowOffset < $numRows; $rowOffset += $linksConvInsertInterval ) { $sqlRead = "SELECT * FROM $links "; $sqlRead = $dbw->limitResult( $sqlRead, $linksConvInsertInterval, $rowOffset ); @@ -143,9 +159,9 @@ The wiki should be put into read-only mode while this script executes"; } 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 ) ) { + foreach ( $res as $row ) { $fromTitle = $row->l_from; if ( array_key_exists( $fromTitle, $ids ) ) { # valid title $from = $ids[$fromTitle]; @@ -169,27 +185,29 @@ The wiki should be put into read-only mode while this script executes"; $totalTuplesInserted += $tuplesAdded; if ( $reportLinksConvProgress ) $this->output( " done. Total $totalTuplesInserted tuples inserted.\n" ); - $this->performanceLog( $totalTuplesInserted . " " . ( $this->getMicroTime() - $baseTime ) . "\n" ); + $this->performanceLog( $fh, $totalTuplesInserted . " " . ( $this->getMicroTime() - $baseTime ) . "\n" ); } } $this->output( "$totalTuplesInserted valid titles and $numBadLinks invalid titles were processed.\n\n" ); - $this->performanceLog( "$totalTuplesInserted valid titles and $numBadLinks invalid titles were processed.\n" ); - $this->performanceLog( "Total execution time: " . ( $this->getMicroTime() - $startTime ) . " seconds.\n" ); - if ( $logPerformance ) { fclose ( $fh ); } + $this->performanceLog( $fh, "$totalTuplesInserted valid titles and $numBadLinks invalid titles were processed.\n" ); + $this->performanceLog( $fh, "Total execution time: " . ( $this->getMicroTime() - $startTime ) . " seconds.\n" ); + if ( $this->logPerformance ) { + fclose ( $fh ); + } } # -------------------------------------------------------------------- if ( $overwriteLinksTable ) { # Check for existing links_backup, and delete it if it exists. $this->output( "Dropping backup links table if it exists..." ); - $dbw->query( "DROP TABLE IF EXISTS $links_backup", DB_MASTER ); + $dbw->query( "DROP TABLE IF EXISTS $links_backup", __METHOD__ ); $this->output( " done.\n" ); - + # Swap in the new table, and move old links table to links_backup $this->output( "Swapping tables '$links' to '$links_backup'; '$links_temp' to '$links'..." ); - $dbw->query( "RENAME TABLE links TO $links_backup, $links_temp TO $links", DB_MASTER ); + $dbw->query( "RENAME TABLE links TO $links_backup, $links_temp TO $links", __METHOD__ ); $this->output( " done.\n\n" ); - + $dbw->close(); $this->output( "Conversion complete. The old table remains at $links_backup;\n" ); $this->output( "delete at your leisure.\n" ); @@ -200,7 +218,6 @@ The wiki should be put into read-only mode while this script executes"; } private function createTempTable() { - global $noKeys; $dbConn = wfGetDB( DB_MASTER ); if ( !( $dbConn->isOpen() ) ) { @@ -214,7 +231,7 @@ The wiki should be put into read-only mode while this script executes"; $this->output( " done.\n" ); $this->output( "Creating temporary links table..." ); - if ( $noKeys ) { + if ( $this->hasOption( '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')" ); @@ -228,9 +245,8 @@ The wiki should be put into read-only mode while this script executes"; $this->output( " done.\n\n" ); } - private function performanceLog( $text ) { - global $logPerformance, $fh; - if ( $logPerformance ) { + private function performanceLog( $fh, $text ) { + if ( $this->logPerformance ) { fwrite( $fh, $text ); } } @@ -242,4 +258,4 @@ The wiki should be put into read-only mode while this script executes"; } $maintClass = "ConvertLinks"; -require_once( DO_MAINTENANCE ); +require_once( RUN_MAINTENANCE_IF_MAIN );