Spaces:
Runtime error
Runtime error
File size: 2,691 Bytes
716048e |
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 |
from src.chatbot_v2.layers.perception import PerceptionLayer
from src.chatbot_v2.layers.cognition import CognitionLayer, CognitiveOutput
from src.chatbot_v2.layers.action import ActionLayer
from src.chatbot_v2.layers.memory import MemoryLayer
from typing import List, Dict
class ChatbotOrchestrator:
"""
DESIGN PATTERN: ReAct (Reason + Act) - Orchestrator
Structure:
1. Loop:
- Update Perception
- Consult Cognition (Reasoning)
- If Action needed -> Call Action Layer -> Loop
- If Final Answer -> Return
"""
def __init__(self):
# Initialize the 'Organs' (Layers)
self.perception = PerceptionLayer()
self.brain = CognitionLayer()
self.hands = ActionLayer()
self.memory = MemoryLayer()
async def run(self, user_input: str, external_history: List[Dict[str, str]] = None):
"""
Runs the agent loop.
Args:
user_input: The new message from the user.
external_history: Existing chat history (e.g. from Streamlit).
"""
# 1. Sync Memory
if external_history:
self.memory.set_history(list(external_history)) # Copy
# 2. Perception Layer (Input)
env_state = self.perception.perceive(user_input)
self.memory.add_entry("user", env_state.user_input)
# Max steps to prevent infinite loops
for step in range(5):
print(f"\n--- Step {step+1} (ReAct Loop) ---")
# 3. Cognition Layer (Reasoning)
history = self.memory.get_history()
decision: CognitiveOutput = await self.brain.decide(history)
print(f"[Think]: {decision.thought}")
# 4. Handling Decision
if decision.final_answer:
print(f"[Final Answer]: {decision.final_answer}")
self.memory.add_entry("assistant", decision.final_answer)
return decision.final_answer
if decision.action:
# 5. Action Layer (Execution)
print(f"[Action Needed]: Call {decision.action} with {decision.action_input}")
tool_result = await self.hands.execute(decision.action, decision.action_input or {})
# 6. Feedback Loop
print(f"[Observation]: {tool_result}")
self.memory.add_entry("system", f"Tool {decision.action} returned: {tool_result}")
return "I apologize, but I got stuck in a loop trying to answer your request."
|