65 lines
1.9 KiB
Python
65 lines
1.9 KiB
Python
#!/usr/bin/env python3
|
|
"""
|
|
Validate candidate Agent replay outputs before normalization/scoring.
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import argparse
|
|
import json
|
|
import sys
|
|
from pathlib import Path
|
|
from typing import Any
|
|
|
|
|
|
ROOT = Path(__file__).resolve().parents[2]
|
|
API_SRC = ROOT / "apps" / "api"
|
|
sys.path.insert(0, str(API_SRC))
|
|
|
|
from src.services.agent_replay_contract import ( # noqa: E402
|
|
validate_candidate_replay_contract,
|
|
)
|
|
|
|
|
|
def main() -> int:
|
|
parser = argparse.ArgumentParser(
|
|
description="Validate candidate replay result alignment against inputs."
|
|
)
|
|
parser.add_argument("--inputs", required=True, help="candidate input JSONL")
|
|
parser.add_argument("--results", required=True, help="candidate raw result JSONL")
|
|
parser.add_argument("--candidate-id", help="Expected candidate_id")
|
|
parser.add_argument("--output", help="Contract report JSON path")
|
|
args = parser.parse_args()
|
|
|
|
report = validate_candidate_replay_contract(
|
|
candidate_inputs=_read_jsonl(Path(args.inputs)),
|
|
candidate_results=_read_jsonl(Path(args.results)),
|
|
expected_candidate_id=args.candidate_id,
|
|
).to_dict()
|
|
rendered = json.dumps(report, ensure_ascii=False, indent=2, sort_keys=True)
|
|
|
|
if args.output:
|
|
Path(args.output).write_text(rendered + "\n", encoding="utf-8")
|
|
else:
|
|
print(rendered)
|
|
|
|
return 0 if report["valid"] else 2
|
|
|
|
|
|
def _read_jsonl(path: Path) -> list[dict[str, Any]]:
|
|
records: list[dict[str, Any]] = []
|
|
with path.open(encoding="utf-8") as handle:
|
|
for line_number, line in enumerate(handle, start=1):
|
|
line = line.strip()
|
|
if not line or line.startswith("#"):
|
|
continue
|
|
try:
|
|
records.append(json.loads(line))
|
|
except Exception as exc:
|
|
raise SystemExit(f"{path}:{line_number}: invalid JSONL: {exc}") from exc
|
|
return records
|
|
|
|
|
|
if __name__ == "__main__":
|
|
raise SystemExit(main())
|