Update main.py
Browse files
main.py
CHANGED
|
@@ -2,53 +2,63 @@ import os
|
|
| 2 |
import logging
|
| 3 |
import asyncio
|
| 4 |
import nest_asyncio
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
from huggingface_hub import InferenceClient, login
|
| 8 |
-
import langid
|
| 9 |
|
| 10 |
# Import Telegram bot components
|
| 11 |
from telegram import Update, Bot
|
| 12 |
from telegram.ext import Application, CommandHandler, MessageHandler, filters, CallbackContext
|
| 13 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14 |
# -------------------------
|
| 15 |
# Configure logging
|
| 16 |
# -------------------------
|
| 17 |
-
logging.basicConfig(
|
| 18 |
-
format="%(asctime)s - %(levelname)s - %(message)s", level=logging.INFO
|
| 19 |
-
)
|
| 20 |
logger = logging.getLogger(__name__)
|
| 21 |
|
| 22 |
# -------------------------
|
| 23 |
-
#
|
| 24 |
# -------------------------
|
| 25 |
-
|
| 26 |
-
if
|
| 27 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 28 |
|
| 29 |
-
login(token=HF_HUB_TOKEN)
|
| 30 |
-
client = InferenceClient(api_key=HF_HUB_TOKEN)
|
| 31 |
-
|
| 32 |
-
# Get Telegram Bot Token
|
| 33 |
-
TOKEN = os.getenv("TELEGRAM_BOT_TOKEN")
|
| 34 |
-
if not TOKEN:
|
| 35 |
-
raise ValueError("Missing Telegram Bot Token. Please set TELEGRAM_BOT_TOKEN.")
|
| 36 |
-
|
| 37 |
-
# Get Webhook Domain (e.g., your-space.hf.space)
|
| 38 |
-
WEBHOOK_DOMAIN = os.getenv("WEBHOOK_DOMAIN")
|
| 39 |
-
if not WEBHOOK_DOMAIN:
|
| 40 |
-
raise ValueError("Missing Webhook Domain. Please set WEBHOOK_DOMAIN.")
|
| 41 |
-
|
| 42 |
-
# Construct the full webhook URL.
|
| 43 |
-
# We use the Telegram bot token as a unique path.
|
| 44 |
-
WEBHOOK_URL = f"https://{WEBHOOK_DOMAIN}/{TOKEN}"
|
| 45 |
|
|
|
|
|
|
|
| 46 |
|
| 47 |
# -------------------------
|
| 48 |
-
#
|
| 49 |
# -------------------------
|
| 50 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 51 |
try:
|
|
|
|
| 52 |
lang, _ = langid.classify(user_input)
|
| 53 |
if lang == "he":
|
| 54 |
return "hebrew"
|
|
@@ -60,94 +70,103 @@ def detect_language(user_input: str):
|
|
| 60 |
logger.error(f"Language detection error: {e}")
|
| 61 |
return "unsupported"
|
| 62 |
|
| 63 |
-
|
| 64 |
-
|
| 65 |
-
|
| 66 |
-
|
| 67 |
-
|
| 68 |
-
|
| 69 |
-
|
| 70 |
if language == "hebrew":
|
| 71 |
-
|
| 72 |
-
content = "תענה בקצרה אבל תשתף את תהליך קבלת ההחלטות שלך, " + text
|
| 73 |
-
model = "microsoft/Phi-3.5-mini-instruct"
|
| 74 |
elif language == "english":
|
| 75 |
-
|
| 76 |
-
content = "keep it short but tell your decision making process, " + text
|
| 77 |
-
model = "mistralai/Mistral-Nemo-Instruct-2407"
|
| 78 |
else:
|
| 79 |
return "Sorry, I only support Hebrew and English."
|
| 80 |
|
| 81 |
-
|
| 82 |
-
|
| 83 |
try:
|
| 84 |
-
#
|
| 85 |
-
|
| 86 |
-
|
| 87 |
-
|
| 88 |
-
|
| 89 |
-
|
| 90 |
-
|
| 91 |
-
)
|
| 92 |
-
return completion.choices[0].message.content
|
| 93 |
except Exception as e:
|
| 94 |
-
logger.error(f"Error generating response: {e}")
|
| 95 |
return "Error: Could not generate response."
|
| 96 |
|
| 97 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 98 |
# -------------------------
|
| 99 |
# Telegram Bot Handlers
|
| 100 |
# -------------------------
|
| 101 |
async def start(update: Update, context: CallbackContext):
|
| 102 |
-
|
|
|
|
|
|
|
| 103 |
await update.message.reply_text("Hello! Tell me your decision-making issue, and I'll try to help.")
|
| 104 |
-
logger.info("
|
| 105 |
|
| 106 |
|
| 107 |
async def handle_message(update: Update, context: CallbackContext):
|
|
|
|
|
|
|
|
|
|
|
|
|
| 108 |
user_text = update.message.text
|
| 109 |
-
logger.info(f"
|
| 110 |
-
|
| 111 |
-
# Generate a response using our Hugging Face chat logic.
|
| 112 |
-
response_text = generate_response(user_text)
|
| 113 |
logger.info(f"Generated response: {response_text}")
|
| 114 |
-
|
| 115 |
await update.message.reply_text(response_text)
|
| 116 |
|
| 117 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 118 |
# -------------------------
|
| 119 |
# Main function to run the Telegram Bot using Webhook mode
|
| 120 |
# -------------------------
|
| 121 |
async def main():
|
| 122 |
-
# Build the
|
| 123 |
-
application = Application.builder().token(
|
| 124 |
|
| 125 |
-
# Add
|
| 126 |
application.add_handler(CommandHandler("start", start))
|
| 127 |
application.add_handler(MessageHandler(filters.TEXT & ~filters.COMMAND, handle_message))
|
| 128 |
|
| 129 |
logger.info("Starting bot in webhook mode...")
|
| 130 |
print("Starting bot in webhook mode...")
|
| 131 |
|
| 132 |
-
# Run the bot
|
| 133 |
-
# The bot will listen on all interfaces (0.0.0.0) at port 7860.
|
| 134 |
-
# Telegram will send updates to WEBHOOK_URL.
|
| 135 |
await application.run_webhook(
|
| 136 |
-
listen="0.0.0.0",
|
| 137 |
-
port=7860,
|
| 138 |
-
url_path=
|
| 139 |
-
webhook_url=WEBHOOK_URL
|
| 140 |
)
|
| 141 |
|
| 142 |
-
|
| 143 |
# -------------------------
|
| 144 |
-
# Run the main function
|
| 145 |
# -------------------------
|
| 146 |
if __name__ == "__main__":
|
| 147 |
-
# Apply nest_asyncio to
|
| 148 |
nest_asyncio.apply()
|
| 149 |
loop = asyncio.get_event_loop()
|
| 150 |
-
|
| 151 |
try:
|
| 152 |
loop.run_until_complete(main())
|
| 153 |
except Exception as e:
|
|
|
|
| 2 |
import logging
|
| 3 |
import asyncio
|
| 4 |
import nest_asyncio
|
| 5 |
+
import requests
|
| 6 |
+
from hugchat import hugchat
|
|
|
|
|
|
|
| 7 |
|
| 8 |
# Import Telegram bot components
|
| 9 |
from telegram import Update, Bot
|
| 10 |
from telegram.ext import Application, CommandHandler, MessageHandler, filters, CallbackContext
|
| 11 |
|
| 12 |
+
# -------------------------
|
| 13 |
+
# Load environment variables
|
| 14 |
+
# -------------------------
|
| 15 |
+
# Expected environment variables:
|
| 16 |
+
# TELEGRAM_BOT_TOKEN - Telegram bot token
|
| 17 |
+
# WEBHOOK_DOMAIN - example: your-space.hf.space
|
| 18 |
+
# HF_EMAIL and HF_PASSWORD
|
| 19 |
+
|
| 20 |
+
TELEGRAM_BOT_TOKEN = os.getenv("TELEGRAM_BOT_TOKEN")
|
| 21 |
+
if not TELEGRAM_BOT_TOKEN:
|
| 22 |
+
raise ValueError("Missing TELEGRAM_BOT_TOKEN environment variable.")
|
| 23 |
+
|
| 24 |
+
WEBHOOK_DOMAIN = os.getenv("WEBHOOK_DOMAIN")
|
| 25 |
+
if not WEBHOOK_DOMAIN:
|
| 26 |
+
raise ValueError("Missing WEBHOOK_DOMAIN environment variable.")
|
| 27 |
+
|
| 28 |
# -------------------------
|
| 29 |
# Configure logging
|
| 30 |
# -------------------------
|
| 31 |
+
logging.basicConfig(format="%(asctime)s - %(levelname)s - %(message)s", level=logging.INFO)
|
|
|
|
|
|
|
| 32 |
logger = logging.getLogger(__name__)
|
| 33 |
|
| 34 |
# -------------------------
|
| 35 |
+
# Set up Hugging Chat integration
|
| 36 |
# -------------------------
|
| 37 |
+
# Logs into Hugging Face or uses a fallback to get cookies.
|
| 38 |
+
if os.getenv("HF_EMAIL") and os.getenv("HF_PASSWORD"):
|
| 39 |
+
from hugchat.login import Login
|
| 40 |
+
logger.info("Logging into Hugging Face using provided credentials.")
|
| 41 |
+
sign = Login(os.getenv("HF_EMAIL"), os.getenv("HF_PASSWORD"))
|
| 42 |
+
cookies = sign.login()
|
| 43 |
+
else:
|
| 44 |
+
logger.info("No HF_EMAIL/HF_PASSWORD provided. Using fallback method to get cookies.")
|
| 45 |
+
cookies = requests.get("https://huggingface.co/chat/").cookies
|
| 46 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 47 |
|
| 48 |
+
# Create a ChatBot instance from the hugchat library.
|
| 49 |
+
chatbot = hugchat.ChatBot(cookies=cookies.get_dict())
|
| 50 |
|
| 51 |
# -------------------------
|
| 52 |
+
# Helper Functions
|
| 53 |
# -------------------------
|
| 54 |
+
|
| 55 |
+
def detect_language(user_input: str) -> str:
|
| 56 |
+
"""
|
| 57 |
+
Detect language using langid.
|
| 58 |
+
Returns 'hebrew' if Hebrew, 'english' if English, otherwise 'unsupported'.
|
| 59 |
+
"""
|
| 60 |
try:
|
| 61 |
+
import langid
|
| 62 |
lang, _ = langid.classify(user_input)
|
| 63 |
if lang == "he":
|
| 64 |
return "hebrew"
|
|
|
|
| 70 |
logger.error(f"Language detection error: {e}")
|
| 71 |
return "unsupported"
|
| 72 |
|
| 73 |
+
|
| 74 |
+
def generate_response_sync(message: str) -> str:
|
| 75 |
+
"""
|
| 76 |
+
Generate a response using Hugging Chat in a synchronous manner.
|
| 77 |
+
This function is blocking; we'll run it in an executor.
|
| 78 |
+
"""
|
| 79 |
+
language = detect_language(message)
|
| 80 |
if language == "hebrew":
|
| 81 |
+
prompt = "תענה בקצרה אבל תשתף את תהליך קבלת ההחלטות שלך, " + message
|
|
|
|
|
|
|
| 82 |
elif language == "english":
|
| 83 |
+
prompt = "keep it short but tell your decision making process, " + message
|
|
|
|
|
|
|
| 84 |
else:
|
| 85 |
return "Sorry, I only support Hebrew and English."
|
| 86 |
|
| 87 |
+
response_queue = ""
|
| 88 |
+
full_response = ""
|
| 89 |
try:
|
| 90 |
+
# The huggingchat.ChatBot.chat() method returns a generator (streaming tokens)
|
| 91 |
+
for resp in chatbot.chat(prompt, _stream_yield_all=True):
|
| 92 |
+
if resp and "token" in resp:
|
| 93 |
+
response_queue += resp["token"]
|
| 94 |
+
# Flush the response if it's long (optional, here we accumulate)
|
| 95 |
+
full_response = response_queue
|
| 96 |
+
return full_response
|
|
|
|
|
|
|
| 97 |
except Exception as e:
|
| 98 |
+
logger.error(f"Error generating response via Hugging Chat: {e}")
|
| 99 |
return "Error: Could not generate response."
|
| 100 |
|
| 101 |
|
| 102 |
+
async def generate_response(message: str) -> str:
|
| 103 |
+
"""
|
| 104 |
+
Asynchronous wrapper around generate_response_sync using run_in_executor.
|
| 105 |
+
"""
|
| 106 |
+
loop = asyncio.get_event_loop()
|
| 107 |
+
response = await loop.run_in_executor(None, generate_response_sync, message)
|
| 108 |
+
return response
|
| 109 |
+
|
| 110 |
+
|
| 111 |
# -------------------------
|
| 112 |
# Telegram Bot Handlers
|
| 113 |
# -------------------------
|
| 114 |
async def start(update: Update, context: CallbackContext):
|
| 115 |
+
"""
|
| 116 |
+
Handler for the /start command.
|
| 117 |
+
"""
|
| 118 |
await update.message.reply_text("Hello! Tell me your decision-making issue, and I'll try to help.")
|
| 119 |
+
logger.info("Received /start command.")
|
| 120 |
|
| 121 |
|
| 122 |
async def handle_message(update: Update, context: CallbackContext):
|
| 123 |
+
"""
|
| 124 |
+
Handler for incoming text messages.
|
| 125 |
+
It generates a response using the Hugging Chat integration.
|
| 126 |
+
"""
|
| 127 |
user_text = update.message.text
|
| 128 |
+
logger.info(f"Received message: {user_text}")
|
| 129 |
+
response_text = await generate_response(user_text)
|
|
|
|
|
|
|
| 130 |
logger.info(f"Generated response: {response_text}")
|
|
|
|
| 131 |
await update.message.reply_text(response_text)
|
| 132 |
|
| 133 |
|
| 134 |
+
# -------------------------
|
| 135 |
+
# Webhook Configuration for Telegram
|
| 136 |
+
# -------------------------
|
| 137 |
+
# Construct the full webhook URL.
|
| 138 |
+
# We use the bot token as the URL path for uniqueness.
|
| 139 |
+
WEBHOOK_URL = f"https://{WEBHOOK_DOMAIN}/{TELEGRAM_BOT_TOKEN}"
|
| 140 |
+
|
| 141 |
# -------------------------
|
| 142 |
# Main function to run the Telegram Bot using Webhook mode
|
| 143 |
# -------------------------
|
| 144 |
async def main():
|
| 145 |
+
# Build the Telegram Application.
|
| 146 |
+
application = Application.builder().token(TELEGRAM_BOT_TOKEN).build()
|
| 147 |
|
| 148 |
+
# Add handlers.
|
| 149 |
application.add_handler(CommandHandler("start", start))
|
| 150 |
application.add_handler(MessageHandler(filters.TEXT & ~filters.COMMAND, handle_message))
|
| 151 |
|
| 152 |
logger.info("Starting bot in webhook mode...")
|
| 153 |
print("Starting bot in webhook mode...")
|
| 154 |
|
| 155 |
+
# Run the bot using webhook mode.
|
|
|
|
|
|
|
| 156 |
await application.run_webhook(
|
| 157 |
+
listen="0.0.0.0", # Listen on all interfaces.
|
| 158 |
+
port=7860, # Port to listen on – must match the exposed port in Dockerfile.
|
| 159 |
+
url_path=TELEGRAM_BOT_TOKEN, # Use the bot token as the URL path.
|
| 160 |
+
webhook_url=WEBHOOK_URL # Full webhook URL that Telegram will call.
|
| 161 |
)
|
| 162 |
|
|
|
|
| 163 |
# -------------------------
|
| 164 |
+
# Run the main function (handling event loop issues)
|
| 165 |
# -------------------------
|
| 166 |
if __name__ == "__main__":
|
| 167 |
+
# Apply nest_asyncio to support nested event loops (useful in HF Spaces).
|
| 168 |
nest_asyncio.apply()
|
| 169 |
loop = asyncio.get_event_loop()
|
|
|
|
| 170 |
try:
|
| 171 |
loop.run_until_complete(main())
|
| 172 |
except Exception as e:
|