#!/usr/bin/env python3
"""Requeue Cassandra workers that were interrupted by a Hermes pod restart."""

from __future__ import annotations

import os
from typing import Any


BOARD = "cassandra"
REASON = "Hermes pod restarted; resume the durable task from its saved context"


def recover_running_tasks(kanban_db: Any) -> list[str]:
    """Reclaim every running Cassandra task and return the recovered task IDs."""
    if not kanban_db.board_exists(BOARD):
        return []

    recovered: list[str] = []
    with kanban_db.scoped_current_board(BOARD):
        connection = kanban_db.connect(board=BOARD)
        try:
            for task in kanban_db.list_tasks(connection):
                if str(getattr(task, "status", "")) != "running":
                    continue
                task_id = str(getattr(task, "id", ""))
                run_id = getattr(task, "current_run_id", None)
                if not task_id or not isinstance(run_id, int) or not kanban_db.reclaim_task(
                    connection,
                    task_id,
                    reason=REASON,
                    expected_run_id=run_id,
                ):
                    continue
                kanban_db.add_comment(
                    connection,
                    task_id,
                    "pod-recovery",
                    "Requeued after the Hermes pod restart; the replacement pod will resume this task.",
                )
                recovered.append(task_id)
        finally:
            connection.close()
    return recovered


def main() -> int:
    """Load Hermes's Kanban API and reclaim interrupted Cassandra work."""
    os.environ.setdefault("HERMES_HOME", "/opt/data")
    from hermes_cli import kanban_db

    recovered = recover_running_tasks(kanban_db)
    if recovered:
        print(f"Requeued {len(recovered)} interrupted Cassandra task(s): {', '.join(recovered)}")
    else:
        print("No interrupted Cassandra tasks needed recovery")
    return 0


if __name__ == "__main__":
    raise SystemExit(main())