twillio_gemini_api.py

# app.py

import os
import requests  # To download the audio file from Twilio's URL
from flask import Flask, request
from dotenv import load_dotenv
# Use the new top-level import and types
from google import genai
from google.genai import types
from twilio.twiml.messaging_response import MessagingResponse

# --- Define Calculator Functions for the Model to Use ---
# The SDK uses the function signature (name, parameters) and docstring
# to tell the model how and when to use these tools.

def add(a: float, b: float):
    """
    Adds two numbers together.

    Args:
        a: The first number.
        b: The second number.
    """
    print(f"Tool Call: add(a={a}, b={b})")
    return a + b

def subtract(a: float, b: float):
    """
    Subtracts the second number from the first.

    Args:
        a: The number to subtract from.
        b: The number to subtract.
    """
    print(f"Tool Call: subtract(a={a}, b={b})")
    return a - b

def multiply(a: float, b: float):
    """
    Multiplies two numbers.

    Args:
        a: The first number.
        b: The second number.
    """
    print(f"Tool Call: multiply(a={a}, b={b})")
    return a * b

def divide(a: float, b: float):
    """
    Divides the first number by the second.

    Args:
        a: The numerator.
        b: The denominator.
    """
    print(f"Tool Call: divide(a={a}, b={b})")
    if b == 0:
        return "Error: Cannot divide by zero."
    return a / b

# A list of all the functions the model can call
calculator_tools = [add, subtract, multiply, divide]


# --- Initialize Flask App and APIs ---

# Load environment variables from .env file
load_dotenv(r"C:\Users\Vaibhav Arora\Documents\MyExperimentsandCodes\APPS_WEBSITES\CANADA_WHOLESALE_PROJECT\GITHUB_REPOS\mvp-vue\wholesale-grocery-app\AIAPPS\.env")

app = Flask(__name__)

# Configure the Gemini API and initialize the client
try:
    client = genai.Client(api_key=os.environ.get("GEMINI_API_KEY"))
    print("Gemini client and calculator tools initialized successfully.")
except Exception as e:
    print(f"Error initializing Gemini client: {e}")
    client = None


# --- Define the Webhook Endpoint ---
@app.route("/sms", methods=['POST'])
def sms_reply():
    """Respond to incoming text or audio messages, using calculator functions if needed."""
    
    twilio_resp = MessagingResponse()

    if not client:
        twilio_resp.message("The AI client is not configured correctly. Please check the server logs.")
        return str(twilio_resp)

    # Prepare the contents list for the Gemini API call
    contents = []
    
    # Check if the incoming message contains media
    num_media = int(request.values.get('NumMedia', 0))

    if num_media > 0:
        media_url = request.values.get('MediaUrl0')
        mime_type = request.values.get('MediaContentType0')

        # Process only if the media is audio
        if 'audio' in mime_type:
            print(f"Received audio message. URL: {media_url}, MIME Type: {mime_type}")
            
            # Download the audio file from the Twilio URL
            audio_response = requests.get(media_url)
            
            if audio_response.status_code == 200:
                audio_bytes = audio_response.content
                audio_part = types.Part.from_bytes(data=audio_bytes, mime_type=mime_type)
                prompt = "Please transcribe this audio. If it contains a calculation or a question, please answer it."
                contents = [prompt, audio_part]
            else:
                error_message = "Sorry, I couldn't download the audio file to process it. Please try again."
                twilio_resp.message(error_message)
                return str(twilio_resp)
        else:
            # Handle non-audio media like images or videos
            twilio_resp.message("Sorry, I can only process text and audio messages.")
            return str(twilio_resp)

    else:
        # Fallback to text message processing
        incoming_msg = request.values.get('Body', '').strip()
        print(f"Received text message: '{incoming_msg}'")
        if not incoming_msg:
            twilio_resp.message("Please send a text or audio message to get a response.")
            return str(twilio_resp)
        contents = [incoming_msg]

    if not contents:
        twilio_resp.message("Could not determine content to process. Please send a message.")
        return str(twilio_resp)
        
    try:
        print("Sending content to Gemini with calculator tools...")

        # Configure the request to use our calculator functions
        config = types.GenerateContentConfig(tools=calculator_tools)

        # The SDK handles the multi-turn process for function calling automatically
        gemini_response = client.models.generate_content(
            model="gemini-2.5-flash",
            contents=contents, # This will contain either text or [prompt, audio_part]
            config=config,
        )
        
        # This is the final text answer after any function calls have been resolved.
        ai_answer = gemini_response.text
        print(f"Gemini final response: '{ai_answer}'")
        
        # Add the Gemini response to the TwiML message
        twilio_resp.message(ai_answer)

    except Exception as e:
        print(f"An error occurred with the Gemini API: {e}")
        # Send a user-friendly error message
        error_message = "Sorry, I'm having trouble connecting to my brain right now. Please try again later."
        twilio_resp.message(error_message)

    return str(twilio_resp)


# --- Run the Flask App ---
if __name__ == "__main__":
    app.run(debug=True, port=5000)