Maintenance script to delete unused accounts

diff --git a/maintenance/removeUnusedAccounts.inc b/maintenance/removeUnusedAccounts.inc
new file mode 100644 (file)
index 0000000..ac3b5c4
--- /dev/null
+++ b/maintenance/removeUnusedAccounts.inc
@@ -0,0 +1,49 @@
+<?php
+
+/**
+ * Support functions for the removeUnusedAccounts script
+ *
+ * @package MediaWiki
+ * @subpackage Maintenance
+ * @author Rob Church <robchur@gmail.com>
+ */
+
+# Count the number of edits the specified user has made 
+function CountEdits( $user_id ) {
+       # We've *got* to pull this stuff off the master. If the user *has* made an edit, but it hasn't
+       # been replicated to the slaves yet, we'll end up falsely marking them as inactive. This could
+       # (and usually would) lead to their deletion.
+       $dbw =& wfGetDB( DB_MASTER );
+       $sql = 'SELECT COUNT(rev_id) AS count FROM ' . $dbw->tableName( 'revision' ) . ' WHERE rev_user = ' . $user_id;
+       $res = $dbw->query( $sql );
+       $row = $dbw->fetchObject( $res );
+       return( $row->count );  
+}
+
+# Return an array containing all valid user IDs
+function GetUsers() {
+       # We're safe enough pulling this off a slave
+       $dbr =& wfGetDB( DB_SLAVE );
+       $sql = 'SELECT user_id FROM ' . $dbr->tableName( 'user' );
+       $res = $dbr->query( $sql );
+       $users = array();
+       while( $row = $dbr->fetchObject( $res ) ) {
+               $users[] = $row->user_id;
+       }
+       return( $users );
+}
+
+# Delete one or more users
+function DeleteUsers( $users ) {
+       # Need a master, obviously
+       $dbw =& wfGetDB( DB_MASTER );
+       # We'll do it all in one go, for speed
+       $dbw->begin();
+       $table = $dbw->tableName( 'user' );
+       foreach( $users as $user ) {
+               $dbw->query( 'DELETE FROM ' . $table . ' WHERE user_id = ' . $user . ' LIMIT 1' );
+       }
+       $dbw->commit();
+}
+
+?>
\ No newline at end of file

diff --git a/maintenance/removeUnusedAccounts.php b/maintenance/removeUnusedAccounts.php
new file mode 100644 (file)
index 0000000..dd9214f
--- /dev/null
+++ b/maintenance/removeUnusedAccounts.php
@@ -0,0 +1,42 @@
+<?php
+
+/**
+ * Remove unused user accounts from the database
+ * An unused account is one which has made no edits
+ *
+ * @package MediaWiki
+ * @subpackage Maintenance
+ * @author Rob Church <robchur@gmail.com>
+ */
+ 
+require_once( 'commandLine.inc' );
+require_once( 'removeUnusedAccounts.inc' );
+echo( "REMOVE UNUSED ACCOUNTS\nThis script will delete all users who have made no edits.\n\n" );
+
+$count = 0;
+$del = array();
+
+# Right, who needs deleting?
+$users = GetUsers();
+echo( "Found " . count( $users ) . " accounts.\n" );
+echo( "Locating inactive users..." );
+foreach( $users as $user ) {
+       if( $user != 1 ) {      # Don't *touch* the first user account, ever
+               if( CountEdits( $user ) == 0 ) {
+                       # User has no edits, mark them for deletion
+                       $del[] = $user;
+                       $count++;
+               }
+       }
+}
+echo( "done.\n" );
+
+# Purge the inactive accounts we found
+echo( $count . " inactive accounts found. Deleting..." );
+DeleteUsers( $del );
+echo( "done.\n" );
+
+# We're done
+echo( "Complete.\n" );
+
+?>
\ No newline at end of file