I stumbled upon this snippet recently Re: Strip HTML tags again and discovered it works really well (because it allows for broken HTML by checking for allowed tags only). I put in an updated example too because the underlying module behavior has drifted: Re^2: Strip HTML tags again.