"""Build SFT JSONL data from replay trajectories."""

from __future__ import annotations

import argparse
import json

try:
    from .slm_policy import replay_to_sft_rows, write_sft_jsonl
except ImportError:  # pragma: no cover
    from slm_policy import replay_to_sft_rows, write_sft_jsonl


def main() -> None:
    parser = argparse.ArgumentParser()
    parser.add_argument("--replay", nargs="+", default=["artifacts/trajectories.jsonl"])
    parser.add_argument("--output", default="artifacts/sft/browser_actions.jsonl")
    parser.add_argument(
        "--include-failures",
        action="store_true",
        help="Include failed episodes. By default only successful episodes are exported.",
    )
    parser.add_argument(
        "--include-noop",
        action="store_true",
        help="Include noop actions. By default noop/recovery actions are excluded from SFT.",
    )
    args = parser.parse_args()
    excluded = ("ask_oracle",) if args.include_noop else ("ask_oracle", "noop")
    rows = replay_to_sft_rows(
        args.replay,
        success_only=not args.include_failures,
        excluded_action_types=excluded,
    )
    write_sft_jsonl(rows, args.output)
    print(json.dumps({"output": args.output, "rows": len(rows)}, indent=2))


if __name__ == "__main__":
    main()