from __future__ import annotations from typing import Any from pymongo.database import Database from data_studio.identity import payload_hash from data_studio.utils import utc_now def register_protected_benchmark( db: Database, benchmark_id: str, exact_revision: str, task: str, exact_payloads: list[dict[str, Any]], actor: str, ) -> None: fingerprints = sorted({payload_hash(payload) for payload in exact_payloads}) db["protected_benchmarks"].insert_one( { "schema_version": "1.0", "benchmark_id": benchmark_id, "exact_revision": exact_revision, "task": task, "state": "protected", "exact_fingerprints": fingerprints, "created_by": actor, "created_at": utc_now(), } ) def exact_benchmark_matches(db: Database, payloads: list[dict[str, Any]]) -> list[dict[str, str]]: hashes = {payload_hash(payload) for payload in payloads} matches: list[dict[str, str]] = [] for row in db["protected_benchmarks"].find({"state": "protected"}): overlap = hashes.intersection(row.get("exact_fingerprints", [])) for fingerprint in sorted(overlap): matches.append({"benchmark_id": row["benchmark_id"], "fingerprint": fingerprint}) return matches def assert_no_exact_benchmark_matches(db: Database, payloads: list[dict[str, Any]]) -> None: matches = exact_benchmark_matches(db, payloads) if matches: raise ValueError(f"Protected benchmark exact-match leakage detected: {matches}")