#!/usr/bin/env python3
"""Extract user-role language from the current Codex rollout without copying hidden context."""

from __future__ import annotations

import hashlib
import json
from datetime import datetime, timezone
from pathlib import Path

SOURCE = Path("/Users/bingshuo/.codex/sessions/2026/08/23/rollout-2026-08-23T00-26-28-01a02a4b-8ab1-7f53-8df8-787cbc2909ad.jsonl")
ROOT = Path("/Volumes/JZAO/铸渊-ICE-GL-ZY001/CODEX-DEV-20260823/DEV-20260823-001-HANDOFF-20260823")
DIRECT = ROOT / "sources/CURRENT-THREAD-BINGSHUO-DIRECT-NATURAL-LANGUAGE.jsonl"
HEARTBEATS = ROOT / "sources/CURRENT-THREAD-AUTOMATION-HEARTBEATS.jsonl"
AUDIT = ROOT / "sources/CURRENT-THREAD-USER-MESSAGE-AUDIT.json"


def sha(data: bytes) -> str:
    return hashlib.sha256(data).hexdigest()


def stable(obj: object) -> str:
    return json.dumps(obj, ensure_ascii=False, sort_keys=True, separators=(",", ":"))


def main() -> None:
    source_bytes = SOURCE.read_bytes()
    direct, heartbeats = [], []
    for line in source_bytes.splitlines():
        envelope = json.loads(line)
        if envelope.get("type") != "response_item":
            continue
        payload = envelope.get("payload") or {}
        if payload.get("type") != "message" or payload.get("role") != "user":
            continue
        parts = [item.get("text", "") for item in payload.get("content", []) if isinstance(item, dict) and item.get("type") == "input_text"]
        text = "\n".join(parts)
        is_heartbeat = text.lstrip().startswith("<heartbeat>")
        record = {
            "schema": "guanghu.user-language-handoff/v1",
            "classification": "AUTOMATION_HEARTBEAT_REPLAY" if is_heartbeat else "S01_BINGSHUO_DIRECT_CURRENT_CHANNEL",
            "source_file": str(SOURCE),
            "source_file_sha256_at_extraction": sha(source_bytes),
            "rollout_ordinal": envelope.get("ordinal"),
            "occurred_at_utc": envelope.get("timestamp"),
            "message_id": payload.get("id"),
            "text_sha256": sha(text.encode("utf-8")),
            "text": text,
        }
        (heartbeats if is_heartbeat else direct).append(record)
    DIRECT.parent.mkdir(parents=True, exist_ok=True)
    DIRECT.write_text("".join(stable(x) + "\n" for x in direct), encoding="utf-8")
    HEARTBEATS.write_text("".join(stable(x) + "\n" for x in heartbeats), encoding="utf-8")
    audit = {
        "schema": "guanghu.user-language-handoff-audit/v1",
        "extracted_at_utc": datetime.now(timezone.utc).isoformat().replace("+00:00", "Z"),
        "source_file": str(SOURCE),
        "source_file_sha256_at_extraction": sha(source_bytes),
        "source_file_size": len(source_bytes),
        "direct_message_count": len(direct),
        "automation_heartbeat_count": len(heartbeats),
        "direct_output_sha256": sha(DIRECT.read_bytes()),
        "heartbeat_output_sha256": sha(HEARTBEATS.read_bytes()),
        "exclusion": "System, developer, assistant, tool calls, tool outputs and hidden reasoning are not copied.",
    }
    AUDIT.write_text(json.dumps(audit, ensure_ascii=False, indent=2) + "\n", encoding="utf-8")
    print(json.dumps(audit, ensure_ascii=False, indent=2))


if __name__ == "__main__":
    main()