DiffEditor: Enhancing Speech Editing with Semantic Enrichment and Acoustic Consistency