Spaces:
Sleeping
Sleeping
File size: 4,689 Bytes
9c4d9c0 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 | """Client library for MissionCtrl environment.
Provides HTTP client functions for interacting with the MissionCtrl environment API.
Contains environment payload structures and example scenarios.
"""
import os
from typing import Any, Dict, Optional
import httpx
from dotenv import load_dotenv
# ---------------------------------------------------------------------------
# Load .env file automatically
# ---------------------------------------------------------------------------
load_dotenv()
# ---------------------------------------------------------------------------
# Environment configuration
# ---------------------------------------------------------------------------
ENV_BASE_URL: str = os.environ.get("ENV_BASE_URL", "http://localhost:7860")
# Known agent types in the environment
KNOWN_AGENTS = (
"PlannerAgent",
"ResearchAgent",
"CoderAgent",
"TesterAgent",
"CommAgent",
)
# Available task tiers
TASKS = ["easy", "medium", "hard", "special"]
# Default max steps per episode
MAX_STEPS = 5
# ---------------------------------------------------------------------------
# HTTP Client
# ---------------------------------------------------------------------------
http = httpx.Client(timeout=60.0)
# ---------------------------------------------------------------------------
# Environment API Functions
# ---------------------------------------------------------------------------
def reset_env(task_id: str, seed: Optional[int] = None) -> Dict[str, Any]:
"""Reset the environment for a specific task.
Args:
task_id: The task tier to run (easy, medium, hard, special)
seed: Optional random seed for reproducibility
Returns:
Dictionary containing the initial observation
"""
payload = {"task_id": task_id}
if seed is not None:
payload["seed"] = seed
resp = http.post(f"{ENV_BASE_URL}/reset", json=payload)
resp.raise_for_status()
return resp.json()
def step_env(action: str) -> Dict[str, Any]:
"""Execute one action in the environment.
Args:
action: The action string to execute (e.g., "APPROVE(task_1)")
Returns:
Dictionary containing the new observation, reward, done flag, and info
"""
resp = http.post(f"{ENV_BASE_URL}/step", json={"action": action})
resp.raise_for_status()
return resp.json()
def get_state() -> Dict[str, Any]:
"""Get the current environment state (read-only).
Returns:
Dictionary containing the current observation
"""
resp = http.get(f"{ENV_BASE_URL}/state")
resp.raise_for_status()
return resp.json()
def get_history() -> list:
"""Get the action history for the current episode.
Returns:
List of past actions and their results
"""
resp = http.get(f"{ENV_BASE_URL}/history")
resp.raise_for_status()
return resp.json()
def record_result(tier: str, score: float, steps: int, history: list,
score_breakdown: Optional[Dict] = None,
hallucination_stats: Optional[Dict] = None) -> Dict[str, str]:
"""Push a completed episode result to the dashboard.
Args:
tier: Task tier (easy, medium, hard, special)
score: Final score for the episode
steps: Number of steps taken
history: Action history for the episode
score_breakdown: Optional detailed score breakdown
hallucination_stats: Optional hallucination detection statistics
Returns:
Confirmation response
"""
payload = {
"tier": tier,
"score": score,
"steps": steps,
"history": history,
"score_breakdown": score_breakdown or {},
"hallucination_stats": hallucination_stats or {},
}
resp = http.post(f"{ENV_BASE_URL}/record", json=payload)
resp.raise_for_status()
return resp.json()
# ---------------------------------------------------------------------------
# Example Usage
# ---------------------------------------------------------------------------
def example_basic_usage():
"""Example showing basic environment interaction."""
# Reset environment for easy task
result = reset_env("easy")
obs = result["observation"]
print(f"Started task with {len(obs['tasks'])} tasks")
# Take a step
step_result = step_env("NOOP")
print(f"Reward: {step_result['reward']}, Done: {step_result['done']}")
# Get current state
state = get_state()
print(f"Current step: {state['time_step']}")
if __name__ == "__main__":
# For backward compatibility, delegate to inference script
from inference import main
main()
|