Computational Aspects of Resilient Data Extraction from Semistructured Sources