1 files changed, 63 insertions, 0 deletions
diff --git a/maintenance/purgeOldText.inc b/maintenance/purgeOldText.inc
new file mode 100644
index 00000000..0bf6225a
--- /dev/null
+++ b/maintenance/purgeOldText.inc
@@ -0,0 +1,63 @@
+<?php
+
+/**
+ * Support functions for cleaning up redundant text records
+ *
+ * @package MediaWiki
+ * @subpackage Maintenance
+ * @author Rob Church <robchur@gmail.com>
+ */
+
+function PurgeRedundantText( $delete = false ) {
+	
+	# Data should come off the master, wrapped in a transaction
+	$dbw =& wfGetDB( DB_MASTER );
+	$dbw->begin();
+	
+	$tbl_arc = $dbw->tableName( 'archive' );
+	$tbl_rev = $dbw->tableName( 'revision' );
+	$tbl_txt = $dbw->tableName( 'text' );
+	
+	# Get "active" text records from the revisions table
+	echo( "Searching for active text records in revisions table..." );
+	$res = $dbw->query( "SELECT DISTINCTROW rev_text_id FROM $tbl_rev" );
+	while( $row = $dbw->fetchObject( $res ) ) {
+		$cur[] = $row->rev_text_id;
+	}
+	echo( "done.\n" );
+	
+	# Get "active" text records from the archive table
+	echo( "Searching for active text records in archive table..." );
+	$res = $dbw->query( "SELECT DISTINCTROW ar_text_id FROM $tbl_arc" );
+	while( $row = $dbw->fetchObject( $res ) ) {
+		$cur[] = $row->ar_text_id;
+	}
+	echo( "done.\n" );
+	
+	# Get the IDs of all text records not in these sets
+	echo( "Searching for inactive text records..." );
+	$set = implode( ', ', $cur );
+	$res = $dbw->query( "SELECT old_id FROM $tbl_txt WHERE old_id NOT IN ( $set )" );
+	while( $row = $dbw->fetchObject( $res ) ) {
+		$old[] = $row->old_id;
+	}
+	echo( "done.\n" );
+	
+	# Inform the user of what we're going to do
+	$count = count( $old );
+	echo( "$count inactive items found.\n" );
+	
+	# Delete as appropriate
+	if( $delete && $count ) {
+		echo( "Deleting..." );
+		$set = implode( ', ', $old );
+		$dbw->query( "DELETE FROM $tbl_txt WHERE old_id IN ( $set )" );
+		echo( "done.\n" );
+	}
+	
+	# Done
+	$dbw->commit();
+	
+}
+
+?>
+\ No newline at end of file