File size: 1,759 Bytes
8e2a3f7
 
15a68de
8e2a3f7
b0eb85b
8e2a3f7
 
 
15a68de
8e2a3f7
 
 
2e34ede
 
 
00855c5
8e2a3f7
 
 
 
 
 
 
 
 
 
 
 
b0eb85b
 
15a68de
2e34ede
 
 
 
 
8e2a3f7
15a68de
 
 
 
2e34ede
b0eb85b
15a68de
8e2a3f7
61966fb
15a68de
b0eb85b
15a68de
 
 
61966fb
15a68de
2e34ede
00855c5
8e2a3f7
 
 
2e34ede
8e2a3f7
 
 
 
15a68de
8e2a3f7
 
 
 
 
 
00855c5
2e34ede
8e2a3f7
 
 
 
15a68de
00855c5
 
 
 
 
8e2a3f7
 
 
 
 
 
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
from fastapi import FastAPI
from pydantic import BaseModel
from typing import Optional
from environment import SupportEnv
from dataset import easy_cases, medium_cases, hard_cases

app = FastAPI()

# Models 
class Action(BaseModel):
    message: str

class ResetRequest(BaseModel):
    task: str = "easy"

# Parser 
def parse_action(message):
    message = message.lower()

    if "refund" in message:
        return "refund"
    elif "replace" in message:
        return "replace"
    elif "reject" in message:
        return "reject"
    elif "proof" in message:
        return "ask_proof"
    else:
        return "reject"

#  TASKS 
TASKS = {
    "easy": easy_cases,
    "medium": medium_cases,
    "hard": hard_cases
}

# AUTO TASK CYCLING
TASK_SEQUENCE = ["easy", "medium", "hard"]
current_task_index = 0

env = None

# RESET 
@app.post("/reset")
def reset(req: Optional[ResetRequest] = None):
    global env, current_task_index

    # IGNORE BODY → AUTO CYCLE TASKS
    task = TASK_SEQUENCE[current_task_index]
    current_task_index = (current_task_index + 1) % 3

    cases = TASKS[task]
    env = SupportEnv(cases)

    obs = env.reset()

    return {
        "observation": {"echoed_message": obs},
        "reward": 0.0,
        "done": False
    }

#  STEP
@app.post("/step")
def step(action: Action):
    parsed_action = parse_action(action.message)

    obs, reward, done = env.step(parsed_action)

    response = {
        "observation": {"echoed_message": obs},
        "reward": reward,
        "done": done
    }

    #  RETURN SCORE PER TASK
    if done:
        response["score"] = env.get_score()

    return response

def main():
    import uvicorn
    uvicorn.run(app, host="0.0.0.0", port=7860)

if __name__ == "__main__":
    main()