cyclop/telegram_inference_bot.py

import json
import os
import importlib
import inspect
import logging
from telegram import Update, __version__ as telegram_version
from telegram.ext import Application, CommandHandler, MessageHandler, filters, ContextTypes
from openai import OpenAI
from dotenv import load_dotenv
from tools.base_tool import BaseTool
from anthropic import Anthropic

# Load environment variables
load_dotenv()

openai_client = OpenAI()

anthropic_client = Anthropic(
    api_key=os.environ.get("ANTHROPIC_API_KEY"),
    default_headers={"anthropic-beta": "max-tokens-3-5-sonnet-2024-07-15"}
)

GPT_4O = "gpt-4o"
GPT_4O_MINI = "gpt-4o-mini"

model_max_tokens = {
    GPT_4O: 4096,
    GPT_4O_MINI: 16384
}

use_smart_model = False
use_anthropic = True

# Set up logging to console and file
logging.basicConfig(level=logging.WARNING, handlers=[
    logging.StreamHandler(),
    logging.FileHandler('logs/output.log', mode='a')
])

# Set up Telegram bot
TELEGRAM_BOT_TOKEN = os.getenv('TELEGRAM_BOT_TOKEN')

# Load system prompt
with open("prompts/developer_prompt.txt", "r") as file:
    system_prompt = file.read().strip()

# Dictionary to store conversation history for each user
conversation_history = {}

# Load tools
tools = []
tools_dir = os.path.join(os.path.dirname(__file__), 'tools')
for filename in os.listdir(tools_dir):
    if filename.endswith('.py') and filename != '__init__.py' and filename != 'base_tool.py':
        module_name = f'tools.{filename[:-3]}'
        module = importlib.import_module(module_name)
        for name, obj in inspect.getmembers(module):
            if inspect.isclass(obj) and issubclass(obj, BaseTool) and obj != BaseTool:
                tools.append(obj())

# Collect all function definitions
functions = []
for tool in tools:
    functions.extend(tool.get_functions())

async def start(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
    logging.info("Bot started")
    await update.message.reply_text("Hello! I'm your AI assistant. How can I help you today? You can send me images and then ask questions about them.")

async def clear(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
    user_id = update.effective_user.id
    if user_id in conversation_history:
        del conversation_history[user_id]
    for tool in tools:
        tool.clear()

    logging.info(f"Cleared conversation history and image for user {user_id}")
    await update.message.reply_text("Conversation history and image cleared. Let's start fresh!")

async def handle_message(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
    try:
        user_id = update.effective_user.id
        user_message = update.message.text

        logging.info(f"Message from user {user_id}: {user_message}")

        if user_id not in conversation_history:
            conversation_history[user_id] = []

        conversation_history[user_id].append({"role": "user", "content": user_message})

        messages = conversation_history[user_id]

        response = get_chat_response(messages)
        tool_calls = []
        if use_anthropic:
            fullMessage = []
            for message_part in response.content:
                fullMessage.append(message_part)
                if message_part.type == "tool_use":
                    tool_calls.append(message_part)
            messages.append({"role": "assistant", "content": fullMessage})
        else:
            assistant_message = response.choices[0].message
            if hasattr(assistant_message, 'function_call') and assistant_message.function_call is not None:
                tool_calls.append(assistant_message.function_call)

        toolUseCount = 0

        while len(tool_calls) > 0 and toolUseCount < 50:
            tool_use_results = []
            while len(tool_calls) > 0:
                tool_call = tool_calls.pop(0)
                function_name = tool_call.name
                
                tool_response = call_tool(tool_call)
                tool_use_results.append({"type": "tool_result", "tool_use_id": tool_call.id, "content": json.dumps(tool_response)})

            
            formatted_result = {}

            if use_anthropic:
                formatted_result = {"role": "user", "content":tool_use_results}
            else:
                formatted_result = {"role": "function", "name": function_name, "content": json.dumps(tool_use_results[0])}

            messages.append(formatted_result)

            response = get_chat_response(messages)
            assistant_message = ""
            if use_anthropic:
                fullMessage = []
                for message_part in response.content:
                    fullMessage.append(message_part)
                    if message_part.type == "tool_use":
                        tool_calls.append(message_part)
                messages.append({"role": "assistant", "content": fullMessage})
            else:
                assistant_message = response.choices[0].message
                conversation_history[user_id].append({"role": "assistant", "content": assistant_message})
                if hasattr(assistant_message, 'function_call') and assistant_message.function_call is not None:
                    tool_calls.append(assistant_message.function_call)
                else:
                    conversation_history[user_id].append({"role": "assistant", "content": assistant_message})
                assistant_reply = assistant_message

            toolUseCount += 1
        

        if (toolUseCount == 0):
            if use_anthropic:
                assistant_reply = response.content
            else:
                assistant_reply = assistant_message
            conversation_history[user_id].append({"role": "assistant", "content": assistant_reply})
        
        if len(conversation_history[user_id]) > 20:
            conversation_history[user_id] = conversation_history[user_id][-20:]

        if use_anthropic:
            await update.message.reply_text(messages[-1]["content"][0].text)
        else:
            await update.message.reply_text(assistant_reply.content)

    except Exception as e:
        logging.error(f"An error occurred: {str(e)}")
        await update.message.reply_text("Sorry, an error occurred while processing your request.")

def call_tool(function_call):
    function_name = function_call.name if use_anthropic else function_call.name
    function_args =  json.dumps(function_call.input) if use_anthropic else function_call.arguments
    for tool in tools:
        if function_name in [f["name"] for f in tool.get_functions()]:
            return tool.execute(function_name, **json.loads(function_args))

def get_chat_response(messages):
    return get_claude_response(messages) if use_anthropic else get_openai_response(messages)

def get_openai_response(messages):
    model = GPT_4O if use_smart_model else GPT_4O_MINI
    response = openai_client.chat.completions.create(
        model=model,
        messages = [{"role": "system", "content": system_prompt}] + messages,
        functions=functions,
        function_call="auto",
        max_tokens=model_max_tokens[model]
    )
    return response

def get_claude_response(messages):
    anthropic_tools = [
            {
                "name": function['name'],
                "description": function['description'],
                "input_schema": function['parameters'] if function['parameters'] not in [None, {}] else {"type": "object", "properties": {"param1": {"type": "string", "description": "Unnecessary"}}, "required": []}
            }
            for function in functions
        ]
    try:
        response = anthropic_client.messages.create(
            model="claude-3-5-sonnet-20240620",
            system=system_prompt,
            messages=messages,
            max_tokens=8192,
            tools=anthropic_tools
        )
    except Exception as e:
        logging.error(f"An error occurred: {str(e)}")
        return None
        
    return response

async def switch(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
    global use_smart_model
    use_smart_model = not use_smart_model
    model = GPT_4O if use_smart_model else GPT_4O_MINI
    logging.info(f"Switched to model: {model}")
    await update.message.reply_text(f"Switched to model: {model}")

async def switch_providers(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
    await clear(update, context)
    global use_anthropic
    use_anthropic = not use_anthropic
    logging.info("Using Anthropic" if use_anthropic else "Using OpenAI")
    await update.message.reply_text("Using Anthropic" if use_anthropic else "Using OpenAI")

async def status(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
    if use_anthropic:
        await update.message.reply_text("Currently using claude-3-5-sonnet-20240620")
    else:
        model = GPT_4O if use_smart_model else GPT_4O_MINI
        await update.message.reply_text(f"Currently using: {model}")

def main() -> None:
    # Create the Application and pass it your bot's token
    application = Application.builder().token(TELEGRAM_BOT_TOKEN).build()

    # Add handlers
    application.add_handler(CommandHandler("start", start))
    application.add_handler(CommandHandler("clear", clear))
    application.add_handler(CommandHandler("switch", switch))
    application.add_handler(CommandHandler("toggle", switch_providers))
    application.add_handler(CommandHandler("status", status))
    application.add_handler(MessageHandler(filters.TEXT & ~filters.COMMAND, handle_message))

    # Start the Bot
    logging.info("Bot is running...")
    application.run_polling()

if __name__ == '__main__':
    main()