import gradio as gr
import time
import openai
import json
import os

openai.api_key = os.environ.get('OPENAI_KEY')

def classify_audio(audio):
    # Transcribe the audio to text
    audio_transcript = asr_pipeline(audio)["text"]
    audio_transcript = audio_transcript.lower()

    messages = [
        {"role": "system", "content": "Is this chat a scam, spam or is safe? Only answer in JSON format with 'classification': '' as string and 'reasons': '' as the most plausible reasons why. The reason should be explaning to the potential victim why the conversation is probably a scam"},
        {"role": "user", "content": audio_transcript},
    ]

    # Call the OpenAI API to generate a response
    response = openai.ChatCompletion.create(
        model="gpt-4",  # Replace with the actual GPT-4 model ID
        messages=messages
    )

    # Extract the generated text
    text = response.choices[0].message['content']
    text = json.loads(text)

    # Get the decision and reasons from the JSON dictionary
    decision = text["classification"]
    reasons = text["reasons"]

    # Return the transcription and the prediction as a dictionary
    return audio_transcript, decision, reasons

gr.Interface(
    fn=classify_audio,
    inputs=gr.inputs.Audio(source="upload", type="filepath"),
    outputs=[
        gr.outputs.Textbox(label="Transcription"),
        gr.outputs.Textbox(label="Classification"),
        gr.outputs.Textbox(label="Reason"),
    ],
    live=True
).launch(share=True, debug=True)