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

def generate_stage_directions(location, situation, api_key):
    prompt = (
        f"Write detailed stage directions for a scene with 5 characters, set in a {location}. "
        "You are reading these directions out loud to an audience, so keep the stage directions conversational. "
        "Do not break it down into different sections. Each character enters one by one, two of them enter as a couple. "
        "As they enter, tell us their name and describe their physical characteristics, their emotional state and their actions, "
        "gestures and movements in the scene. Write detailed stage direction on how they interact with the location they are in "
        "and with each other, with detailed description on their movements, actions and gestures in the scene. Make the overall "
        "scene highly dramatic, full of twists and turns, with lots of movement by the characters that keep changing positions "
        "and moving around. At some point, a {situation} happens in the scene. Show the characters interacting with elements of "
        "the location. Describe in vivid detail their emotion, facial expressions and emotions. You will also write dialogues for "
        "each character. Keep the dialogues short. Keep the scene mostly non-verbal, with only a few dialogues. Make the scene "
        "very dramatic, emotional, thrilling. Keep your response limited to 750 words."
    )
    
    openai.api_key = api_key  # Set the API key from the user input
    
    try:
        response = openai.Completion.create(
            engine="text-davinci-003",
            prompt=prompt,
            max_tokens=750,
            temperature=0.7,
        )
        stage_directions = response.choices[0].text.strip()
        response_audio_path = text_to_audio(stage_directions)
        return response_audio_path
    except Exception as e:
        return str(e)

def text_to_audio(text):
    tts = gTTS(text, lang='en')
    temp_file = tempfile.NamedTemporaryFile(delete=False, suffix='.mp3')
    tts.save(temp_file.name)
    return temp_file.name

# Create Gradio UI
iface = gr.Interface(
    fn=generate_stage_directions,
    inputs=[
        gr.Textbox(label="Location"),
        gr.Textbox(label="Situation"),
        gr.Textbox(label="API Key") 
    ],
    outputs=gr.Audio(type='filepath', label="Stage Directions"),
    live=True,
    title="DramaDirector",
    description="Input a location, situation, and your OpenAI API key to generate stage directions.",
)

iface.launch()