]> scripts.mit.edu Git - autoinstallsdev/mediawiki.git/blobdiff - maintenance/convertLinks.php
MediaWiki 1.30.2
[autoinstallsdev/mediawiki.git] / maintenance / convertLinks.php
index b7a55d5718f954ec30de52570908d0b5cf9e7578..54c0edaecc8f953f87c968b0441c0ce0f2088492 100644 (file)
@@ -1,7 +1,6 @@
 <?php
 /**
- * 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
+ * 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
  * @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)
-The wiki should be put into read-only mode while this script executes";
+               $this->addDescription(
+                       '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 );
+               $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',
+                       /* (What about InnoDB?) */
+                       "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.",
+                       false
+               );
        }
 
        public function getDbType() {
@@ -44,27 +66,39 @@ This gives a huge speed improvement for very large links tables which are MyISAM
        }
 
        public function execute() {
-               $dbw = wfGetDB( DB_MASTER );
+               $dbw = $this->getDB( DB_MASTER );
 
                $type = $dbw->getType();
                if ( $type != 'mysql' ) {
                        $this->output( "Link table conversion not necessary for $type\n" );
+
                        return;
                }
 
                global $wgContLang;
 
-               $numBadLinks = $curRowsRead = 0; # counters etc
-               $totalTuplesInserted = 0; # total tuples INSERTed into links_temp
+               # counters etc
+               $numBadLinks = $curRowsRead = 0;
+
+               # total tuples INSERTed into links_temp
+               $totalTuplesInserted = 0;
+
+               # whether or not to give progress reports while reading IDs from cur table
+               $reportCurReadProgress = true;
+
+               # number of rows between progress reports
+               $curReadReportInterval = 1000;
 
-               $reportCurReadProgress = true; # whether or not to give progress reports while reading IDs from cur table
-               $curReadReportInterval = 1000; # number of rows between progress reports
+               # whether or not to give progress reports during conversion
+               $reportLinksConvProgress = true;
 
-               $reportLinksConvProgress = true; # whether or not to give progress reports during conversion
-               $linksConvInsertInterval = 1000; # number of rows per INSERT
+               # number of rows per INSERT
+               $linksConvInsertInterval = 1000;
 
                $initialRowOffset = 0;
-               # $finalRowOffset = 0; # not used yet; highest row number from links table to process
+
+               # not used yet; highest row number from links table to process
+               # $finalRowOffset = 0;
 
                $overwriteLinksTable = !$this->hasOption( 'keep-links-table' );
                $noKeys = $this->hasOption( 'noKeys' );
@@ -73,16 +107,19 @@ This gives a huge speed improvement for very large links tables which are MyISAM
 
                # --------------------------------------------------------------------
 
-               list ( $cur, $links, $links_temp, $links_backup ) = $dbw->tableNamesN( 'cur', 'links', 'links_temp', 'links_backup' );
+               list( $cur, $links, $links_temp, $links_backup ) =
+                       $dbw->tableNamesN( 'cur', 'links', 'links_temp', 'links_backup' );
 
-               if( $dbw->tableExists( 'pagelinks' ) ) {
+               if ( $dbw->tableExists( 'pagelinks' ) ) {
                        $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;
                }
 
@@ -97,21 +134,21 @@ This gives a huge speed improvement for very large links tables which are MyISAM
                } else {
                        $fh = false;
                        if ( $this->logPerformance ) {
-                               $fh = fopen ( $perfLogFilename, "w" );
+                               $fh = fopen( $perfLogFilename, "w" );
                                if ( !$fh ) {
                                        $this->error( "Couldn't open $perfLogFilename" );
                                        $this->logPerformance = false;
                                }
                        }
-                       $baseTime = $startTime = $this->getMicroTime();
+                       $baseTime = $startTime = microtime( true );
                        # Create a title -> cur_id map
                        $this->output( "Loading IDs from $cur table...\n" );
-                       $this->performanceLog ( $fh, "Reading $numRows rows from cur table...\n" );
-                       $this->performanceLog ( $fh, "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();
+                       $ids = [];
 
                        foreach ( $res as $row ) {
                                $title = $row->cur_title;
@@ -122,7 +159,10 @@ This gives a huge speed improvement for very large links tables which are MyISAM
                                $curRowsRead++;
                                if ( $reportCurReadProgress ) {
                                        if ( ( $curRowsRead % $curReadReportInterval ) == 0 ) {
-                                               $this->performanceLog( $fh, $curRowsRead . " " . ( $this->getMicroTime() - $baseTime ) . "\n" );
+                                               $this->performanceLog(
+                                                       $fh,
+                                                       $curRowsRead . " " . ( microtime( true ) - $baseTime ) . "\n"
+                                               );
                                                $this->output( "\t$curRowsRead rows of $cur table read.\n" );
                                        }
                                }
@@ -130,7 +170,10 @@ This gives a huge speed improvement for very large links tables which are MyISAM
                        $dbw->freeResult( $res );
                        $dbw->bufferResults( true );
                        $this->output( "Finished loading IDs.\n\n" );
-                       $this->performanceLog( $fh, "Took " . ( $this->getMicroTime() - $baseTime ) . " seconds to load IDs.\n\n" );
+                       $this->performanceLog(
+                               $fh,
+                               "Took " . ( microtime( true ) - $baseTime ) . " seconds to load IDs.\n\n"
+                       );
 
                        # --------------------------------------------------------------------
 
@@ -138,19 +181,21 @@ This gives a huge speed improvement for very large links tables which are MyISAM
                        # convert, and write to the new table.
                        $this->createTempTable();
                        $this->performanceLog( $fh, "Resetting timer.\n\n" );
-                       $baseTime = $this->getMicroTime();
+                       $baseTime = microtime( true );
                        $this->output( "Processing $numRows rows from $links table...\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 ) {
+                       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 " );
+                                       $sqlWrite = [ "INSERT INTO $links_temp (l_from,l_to) VALUES " ];
                                } else {
-                                       $sqlWrite = array( "INSERT IGNORE INTO $links_temp (l_from,l_to) VALUES " );
+                                       $sqlWrite = [ "INSERT IGNORE INTO $links_temp (l_from,l_to) VALUES " ];
                                }
 
                                $tuplesAdded = 0; # no tuples added to INSERT yet
@@ -169,23 +214,35 @@ This gives a huge speed improvement for very large links tables which are MyISAM
                                        }
                                }
                                $dbw->freeResult( $res );
-                               # $this->output( "rowOffset: $rowOffset\ttuplesAdded: $tuplesAdded\tnumBadLinks: $numBadLinks\n" );
-                               if ( $tuplesAdded != 0  ) {
+                               # $this->output( "rowOffset: $rowOffset\ttuplesAdded: "
+                               #       . "$tuplesAdded\tnumBadLinks: $numBadLinks\n" );
+                               if ( $tuplesAdded != 0 ) {
                                        if ( $reportLinksConvProgress ) {
                                                $this->output( "Inserting $tuplesAdded tuples into $links_temp..." );
                                        }
                                        $dbw->query( implode( "", $sqlWrite ) );
                                        $totalTuplesInserted += $tuplesAdded;
-                                       if ( $reportLinksConvProgress )
+                                       if ( $reportLinksConvProgress ) {
                                                $this->output( " done. Total $totalTuplesInserted tuples inserted.\n" );
-                                               $this->performanceLog( $fh, $totalTuplesInserted . " " . ( $this->getMicroTime() - $baseTime ) . "\n"  );
+                                               $this->performanceLog(
+                                                       $fh,
+                                                       $totalTuplesInserted . " " . ( microtime( true ) - $baseTime ) . "\n"
+                                               );
+                                       }
                                }
                        }
-                       $this->output( "$totalTuplesInserted valid titles and $numBadLinks invalid titles were processed.\n\n" );
-                       $this->performanceLog( $fh, "$totalTuplesInserted valid titles and $numBadLinks invalid titles were processed.\n" );
-                       $this->performanceLog( $fh, "Total execution time: " . ( $this->getMicroTime() - $startTime ) . " seconds.\n" );
+                       $this->output( "$totalTuplesInserted valid titles and "
+                               . "$numBadLinks invalid titles were processed.\n\n" );
+                       $this->performanceLog(
+                               $fh,
+                               "$totalTuplesInserted valid titles and $numBadLinks invalid titles were processed.\n"
+                       );
+                       $this->performanceLog(
+                               $fh,
+                               "Total execution time: " . ( microtime( true ) - $startTime ) . " seconds.\n"
+                       );
                        if ( $this->logPerformance ) {
-                               fclose ( $fh );
+                               fclose( $fh );
                        }
                }
                # --------------------------------------------------------------------
@@ -193,15 +250,14 @@ This gives a huge speed improvement for very large links tables which are MyISAM
                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" );
                } else {
@@ -211,10 +267,11 @@ This gives a huge speed improvement for very large links tables which are MyISAM
        }
 
        private function createTempTable() {
-               $dbConn = wfGetDB( DB_MASTER );
+               $dbConn = $this->getDB( DB_MASTER );
 
                if ( !( $dbConn->isOpen() ) ) {
                        $this->output( "Opening connection to database failed.\n" );
+
                        return;
                }
                $links_temp = $dbConn->tableName( 'links_temp' );
@@ -226,14 +283,14 @@ This gives a huge speed improvement for very large links tables which are MyISAM
                $this->output( "Creating temporary links table..." );
                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')" );
+                               "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))" );
+                               "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))" );
                }
                $this->output( " done.\n\n" );
        }
@@ -243,12 +300,7 @@ This gives a huge speed improvement for very large links tables which are MyISAM
                        fwrite( $fh, $text );
                }
        }
-
-       private function getMicroTime() { # return time in seconds, with microsecond accuracy
-               list( $usec, $sec ) = explode( " ", microtime() );
-               return ( (float)$usec + (float)$sec );
-       }
 }
 
 $maintClass = "ConvertLinks";
-require_once( RUN_MAINTENANCE_IF_MAIN );
+require_once RUN_MAINTENANCE_IF_MAIN;