environment.py

from openenv.core.env_server import Environment as BaseEnv
from .schemas import CalculateAction, SubmitAction, AccountantState, Action

class AccountantEng(BaseEnv):
    def __init__(self):
        self.target_answer = 3000
        self.max_steps = 5
        self.messy_text = "Revenue was 5000, costs were 2000. What is the net profit?"
        self.current_step = 0

    def reset(self) -> AccountantState:
        self.current_state = 0
        return AccountantState(
            observation=self.messy_text,
            is_done=False
        )
    
    def step(self, action:Action)->tuple[AccountantState, float, bool, dict]:
        self.current_step+=1
        reward=-0.1
        done=False
        observation=""
        
        if self.current_step==self.max_steps:
            observation="Max steps reached."

        if action.action_type=="calculate":
            try:
                result=eval(action.expression)
                observation = f"Calculation result: {result}"
            except Exception as e:
                reward-=0.5
                
        elif action.action_type=="submit":
            done=True
            if action.net_profit==self.target_answer:
                reward+=1.0
                observation = "Correct! Task complete."
            else:
                reward-=0.5
                observation = f"Incorrect. The target was not {action.net_profit}."        
        if self.current_step==self.max_steps:
            observation="Max steps reached."

        state = AccountantState(observation=observation, is_done=done)
        return state, float(reward), done, {}