Sufficient Conditions For Local Exactness Of The Exact Penalty Function Method In Nonsmooth Optimization