import os
import gradio as gr
import openai
from gtts import gTTS


openai.api_key = os.environ["OPEN_AI_KEY"]



#messages = gr.State([
#       {"role": "system", "content": "You are a therapist. Respond in less than 5 sentences."}
#])
messages = [{"role": "system", "content": "You are a therapist. Respond in less than 5 sentences."}]


def transcribe(audio):    
    audio_file = open(audio, "rb")    
    # Call the transcribe method with the file-like object
    transcript = openai.Audio.transcribe("whisper-1", audio_file)


    #msg_contents.append({"role": "user", "content": transcript["text"]})

    #chat_transcript = ""
    #for message in msg_contents:
    #    if (message["role"] != "system"):
    #        chat_transcript += message["role"] + ": " + message["content"] + "\n\n"
    system_response = botResponse(transcript["text"])
    
    return system_response

def botResponse(user_input):    
    response = openai.ChatCompletion.create(
      model="gpt-3.5-turbo",
      #messages=user_input)
      messages=[
                {"role": "system", "content": "You are a therapist. Respond in less than 5 sentences."},
                {"role": "user", "content": user_input}
               ]
    )
  
    system_message = response["choices"][0]["message"]["content"]
    #msg_contents.append({"role": "assistant", "content": system_message})

    #chat_transcript = chat_log
    #for message in msg_contents:
    #    if (message["role"] != "system"):
    #        chat_transcript += message["role"] + ": " + message["content"] + "\n\n"

    return system_message


def giveVoice(bot_message):
    myobj = gTTS(text=bot_message)
    myobj.save("temp.mp3")

    dir = os.getcwd()
    new_path = os.path.join(dir, "temp.mp3")

    return new_path






with gr.Blocks() as demo:
    with gr.Row():
        with gr.Column():
            user_audio = gr.Audio(source="microphone", type="filepath", label="Input Phrase")
            submit_btn = gr.Button(value="Transcribe") 
        with gr.Column():
            #gpt_response = gr.Audio(label="Voice Response")
            gpt_transcript = gr.Text(label="Generate Transcript")
            gpt_response = gr.Audio(label="Voice Response")
    submit_btn.click(transcribe, inputs=user_audio, outputs=gpt_transcript)
    gpt_transcript.change(giveVoice, inputs=gpt_transcript, outputs="audio")
    
    

demo.launch(share=False)