client.py

"""Typed client for strategy environment."""

from __future__ import annotations

from typing import Any, Dict

from openenv.core.client_types import StepResult
from openenv.core.env_client import EnvClient

from .models import StrategyAction, StrategyObservation, StrategyState


class StrategyEnv(EnvClient[StrategyAction, StrategyObservation, StrategyState]):
    def _step_payload(self, action: StrategyAction) -> Dict[str, Any]:
        return action.model_dump(exclude_none=True)

    def _parse_result(self, payload: Dict[str, Any]) -> StepResult:
        obs_data = payload.get("observation", {}) or {}
        obs = StrategyObservation(
            done=bool(payload.get("done", obs_data.get("done", False))),
            reward=payload.get("reward", obs_data.get("reward")),
            task_id=str(obs_data.get("task_id", "")),
            difficulty=str(obs_data.get("difficulty", "easy")),
            objective=str(obs_data.get("objective", "")),
            turn=int(obs_data.get("turn", 0)),
            max_turns=int(obs_data.get("max_turns", 12)),
            active_rule=str(obs_data.get("active_rule", "expansion")),
            rule_hint=str(obs_data.get("rule_hint", "")),
            own_resources=int(obs_data.get("own_resources", 0)),
            own_defense=int(obs_data.get("own_defense", 0)),
            own_intel=int(obs_data.get("own_intel", 0)),
            visible_opponent_resources=int(obs_data.get("visible_opponent_resources", 0)),
            visible_opponent_defense=int(obs_data.get("visible_opponent_defense", 0)),
            last_public_event=str(obs_data.get("last_public_event", "")),
            last_agent_action=str(obs_data.get("last_agent_action", "none")),
            last_opponent_action=str(obs_data.get("last_opponent_action", "none")),
            recent_rule_history=list(obs_data.get("recent_rule_history", [])),
            allowed_actions=list(obs_data.get("allowed_actions", [])),
        )
        return StepResult(
            observation=obs,
            reward=payload.get("reward", obs.reward),
            done=bool(payload.get("done", obs.done)),
            info=payload.get("info", {}),
        )

    def _parse_state(self, payload: Dict[str, Any]) -> StrategyState:
        return StrategyState(
            episode_id=payload.get("episode_id"),
            step_count=int(payload.get("step_count", 0)),
            task_id=str(payload.get("task_id", "")),
            difficulty=str(payload.get("difficulty", "easy")),
            objective=str(payload.get("objective", "")),
            turn=int(payload.get("turn", 0)),
            max_turns=int(payload.get("max_turns", 12)),
            active_rule=str(payload.get("active_rule", "expansion")),
            own_resources=int(payload.get("own_resources", 0)),
            own_defense=int(payload.get("own_defense", 0)),
            own_intel=int(payload.get("own_intel", 0)),
            visible_opponent_resources=int(payload.get("visible_opponent_resources", 0)),
            visible_opponent_defense=int(payload.get("visible_opponent_defense", 0)),
            cumulative_reward=float(payload.get("cumulative_reward", 0.0)),
            done=bool(payload.get("done", False)),
            history=list(payload.get("history", [])),
        )