app.py

import os
import gradio as gr
import google.generativeai as genai
import speech_recognition as sr
import tempfile

# -----------------------------------------
# GEMINI CONFIGURATION
# -----------------------------------------
genai.configure(api_key=os.getenv("GEMINI_API_KEY"))
model = genai.GenerativeModel("gemini-1.5-flash")

# -----------------------------------------
# SPEECH RECOGNITION
# -----------------------------------------
recognizer = sr.Recognizer()

def voice_to_text(audio_path):
    try:
        with sr.AudioFile(audio_path) as source:
            audio = recognizer.record(source)
        return recognizer.recognize_google(audio)
    except Exception as e:
        return f"Error converting voice: {str(e)}"

# -----------------------------------------
# CHAT RESPONSE FUNCTION
# -----------------------------------------
def chat_response(message, history):

    if history is None:
        history = []

    # Convert history → Gemini messages format
    gemini_messages = []
    for user_msg, bot_msg in history:
        gemini_messages.append({"role": "user", "content": user_msg})
        gemini_messages.append({"role": "assistant", "content": bot_msg})

    # Add the new user message
    gemini_messages.append({"role": "user", "content": message})

    # ---------------------
    # AUTO-CORRECTION STEP
    # ---------------------
    correction = model.generate_content(
        [{"role": "user", "content": f"Correct this sentence only if needed:\n{message}"}]
    ).text.strip()

    if correction.lower() != message.lower():
        bot_reply = f"❌ Incorrect sentence.\nCorrect form → **{correction}**\nTry again!"
        history.append((message, bot_reply))
        return history

    # ---------------------
    # NORMAL RESPONSE
    # ---------------------
    response = model.generate_content(gemini_messages).text

    history.append((message, response))
    return history

# -----------------------------------------
# HANDLE VOICE INPUT
# -----------------------------------------
def handle_voice(audio_file, history):

    if audio_file is None:
        return history

    text = voice_to_text(audio_file)

    if "error" in text.lower():
        history.append(("Voice Input", text))
        return history

    return chat_response(text, history)


# -----------------------------------------
# GRADIO UI (OLD VERSION COMPATIBLE)
# -----------------------------------------
with gr.Blocks() as app:

    gr.HTML("<h2 style='text-align:center;'>🎤 AI Voice Chatbot (Gemini 1.5 Flash)</h2>")

    chatbot = gr.Chatbot(height=450)

    with gr.Row():
        msg = gr.Textbox(placeholder="Type your message...")
        send = gr.Button("Send")

    audio = gr.Audio(type="filepath", label="🎤 Record/Upload your voice")

    # Text input
    send.click(chat_response, [msg, chatbot], chatbot)
    msg.submit(chat_response, [msg, chatbot], chatbot)

    # Voice input
    audio.change(handle_voice, [audio, chatbot], chatbot)

app.launch()