import streamlit as st
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
from gtts import gTTS
import io
from PIL import Image

# Install PyTorch
try:
    import torch
except ImportError:
    st.warning("PyTorch is not installed. Installing PyTorch...")
    import subprocess
    subprocess.run(["pip", "install", "torch"])
    st.success("PyTorch has been successfully installed!")
    import torch

# Load the image captioning model
caption_model = pipeline("image-to-text", model="Salesforce/blip-image-captioning-base")

# Load the GPT-2 model for story generation
story_generator = pipeline("text-generation", model="openai-community/gpt2")

def generate_caption(image):
    # Generate the caption for the uploaded image
    caption = caption_model(image)[0]["generated_text"]
    return caption

def generate_story(caption):
    # Generate the story based on the caption using the GPT-2 model
    prompt = f"Imagine you are a storyteller for young children. Based on the image described as '{caption}', create a interesting and attractive story for children aged 3-10. Keep it positive and happy in tone. Also, add more imagination"
    story = story_generator(prompt, max_length=200, num_return_sequences=1)[0]["generated_text"]
    return story

def convert_to_audio(story):
    # Convert the story to audio using gTTS
    tts = gTTS(text=story, lang="en")
    audio_bytes = io.BytesIO()
    tts.write_to_fp(audio_bytes)
    audio_bytes.seek(0)
    return audio_bytes

def main():
    st.title("Storytelling Application")
    
    # File uploader for the image (restricted to JPG)
    uploaded_image = st.file_uploader("Upload an image", type=["jpg"])
    
    if uploaded_image is not None:
        # Convert the uploaded image to PIL image
        image = Image.open(uploaded_image)
        
        # Display the uploaded image
        st.image(image, caption="Uploaded Image", use_container_width=True)
        
        # Generate the caption for the image
        caption = generate_caption(image)
        st.subheader("Generated Caption:")
        st.write(caption)
        
        # Generate the story based on the caption using the GPT-2 model
        story = generate_story(caption)
        st.subheader("Generated Story:")
        st.write(story)
        
        # Convert the story to audio
        audio_bytes = convert_to_audio(story)
        
        # Display the audio player
        st.audio(audio_bytes, format="audio/mp3")

if __name__ == "__main__":
    main()