from irpr.deps import search def retrieve(query: str, top_k=8): return search(query, top_k=top_k) def format_citations(hits): url_to_id = {} uniq = [] for h in hits: u = h["source_url"] if u not in url_to_id: url_to_id[u] = len(url_to_id) + 1 uniq.append((url_to_id[u], u)) return uniq