Spaces:
Building
Building
File size: 6,681 Bytes
872c2a9 3f61806 872c2a9 3f61806 7361b6f 3f61806 7361b6f 3f61806 b9ccd0b 3f61806 b9ccd0b 3f61806 b9ccd0b 3f61806 872c2a9 3f61806 b9ccd0b 3f61806 b9ccd0b 3f61806 872c2a9 b9ccd0b 872c2a9 3f61806 872c2a9 3f61806 872c2a9 3f61806 872c2a9 3f61806 b9ccd0b 872c2a9 7361b6f b9ccd0b 7361b6f b9ccd0b 872c2a9 3f61806 872c2a9 7361b6f 24ae72d 7361b6f 872c2a9 3f61806 872c2a9 3f61806 872c2a9 3f61806 872c2a9 3f61806 872c2a9 |
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 |
# /home/bk_anupam/code/LLM_agents/RAG_BOT/message_handler.py
import datetime
from config import Config
from RAG_BOT.logger import logger # Corrected import path based on other files
from langchain_core.messages import HumanMessage
from telebot.types import Message
from langgraph.graph import StateGraph
from RAG_BOT.utils import parse_json_answer
class MessageHandler:
def __init__(self, agent: StateGraph, config: Config):
# Store user session data (could be moved to a database for persistence)
self.config = config
self.sessions = config.USER_SESSIONS
self.agent = agent
def _get_user_session(self, user_id):
if user_id not in self.sessions:
logger.info(f"Creating new session for user {user_id}")
self.sessions[user_id] = {
'last_interaction': datetime.datetime.now(),
'conversation': [],
'context': {},
'language': 'en' # Default language
}
# Ensure essential keys are always present using setdefault
if 'language' not in self.sessions[user_id]:
self.sessions[user_id]['language'] = 'en'
# *** Ensure 'conversation' key exists ***
self.sessions[user_id].setdefault('conversation', [])
return self.sessions[user_id]
def _update_session(self, user_id, message, response):
"""Update the user session with new interaction"""
session = self._get_user_session(user_id)
session['last_interaction'] = datetime.datetime.now()
session['conversation'].append({
'user': message,
'bot': response,
'timestamp': datetime.datetime.now().isoformat()
})
# Limit conversation history (optional)
# Consider making the limit configurable
history_limit = self.config.CONVERSATION_HISTORY_LIMIT or 10
if len(session['conversation']) > history_limit:
session['conversation'] = session['conversation'][-history_limit:]
def process_message(self, incoming_message: Message, language_code: str):
"""
Process the incoming message and generate a response
This is where you implement your custom logic
"""
# language_code is now passed as an argument
user_id = incoming_message.from_user.id
message = incoming_message.text
if not message: # Handle cases like stickers or empty messages if needed
logger.warning(f"Received empty message text from user {user_id}")
return "Sorry, I didn't receive any text."
logger.info(f"Processing message from {user_id}: {message[:100]}...") # Log snippet
# Get user session (still useful for history, etc.)
session = self._get_user_session(user_id)
# Convert message to lowercase for easier matching
message_lower = message.lower().strip()
# --- Updated Greeting Check ---
# Check for exact matches for greetings
greetings = {'hello', 'hi', 'hey'}
if message_lower in greetings:
response = "👋 Hello! I'm your Telegram assistant. How can I help you today?"
# --- End Updated Check ---
elif "help" == message_lower: # Check for exact match for help too
response = ("Here's what I can do:\n"
"- Answer your query (on pdf documents uploaded). Use /query command followed by the query for this\n"
"- Index and store in vector DB uploaded pdf documents. Just send the pdf document as a message\n"
"- Answer any general query \n"
"- Last message - to see your last message\n"
"Just let me know what you need!")
elif "last message" == message_lower: # Exact match
if session['conversation']: # Check if list is not empty
last_user_message = session['conversation'][-1]['user']
response = f"Your last message was: '{last_user_message}'"
else:
response = "You haven't sent any previous messages in this session."
else:
# --- Agent Invocation Logic ---
try:
logger.info(f"Invoking agent for thread_id={str(incoming_message.chat.id)}, lang='{language_code}', query='{message[:50]}...'")
config_thread = {"configurable": {"thread_id": str(incoming_message.chat.id)}}
# Build the initial state for the agent
initial_state = {
"messages": [HumanMessage(content=message)],
"language_code": language_code
}
# It's good practice to stream or use async invoke if available and the agent call might take time
# Using synchronous invoke for now as per the original code
final_state = self.agent.invoke(initial_state, config_thread)
# Extract the answer more robustly
answer = None
if isinstance(final_state, dict) and "messages" in final_state:
final_messages = final_state["messages"]
if isinstance(final_messages, list) and final_messages:
# Get the last message, assuming it's the agent's response
last_msg = final_messages[-1]
if hasattr(last_msg, 'content'):
json_result = parse_json_answer(last_msg.content)
answer = json_result.get("answer") if json_result else None
else:
logger.warning(f"Last message in agent response has no 'content' attribute: {last_msg}")
else:
logger.warning(f"Agent response 'messages' is not a non-empty list: {final_messages}")
else:
logger.warning(f"Unexpected agent response format: {final_state}")
if not answer:
answer = "Sorry, I couldn't retrieve an answer for that." # More specific default
response = answer
except Exception as e:
logger.error(f"Error invoking RAG agent for user {user_id}: {str(e)}", exc_info=True)
response = "Sorry, I encountered an internal error while processing your query."
# --- End Agent Invocation ---
# Update session with this interaction
self._update_session(user_id, message, response)
logger.info(f"Generated response for user {user_id}: {response[:100]}...") # Log response snippet
return response
|