import streamlit as st
from transformers import pipeline, AutoTokenizer, AutoModelForCausalLM

def img2text(url):
    image_to_text_model = pipeline("image-to-text", model="Salesforce/blip-image-captioning-base")
    text = image_to_text_model(url)[0]["generated_text"]
    return text

def text2story(text):
    story_generator = pipeline("text-generation", model="openai-community/gpt2")
    prompt = f"Create a short story under 100 words based on: {text}"
    generated = story_generator(prompt, max_length=100)
    story_text = generated[0]['generated_text']
    return story_text

def text2audio(story_text):
    audio_data = pipeline("text-to-speech", model="facebook/mms-tts-eng")
    return audio_data

st.set_page_config(page_title="Once Upon A Time - Storytelling Application", page_icon="📖🏰🦄🧙")
st.header("Create a story of yours with an image!")
uploaded_file = st.file_uploader("Upload an image of your story!")

if uploaded_file is not None:
    print(uploaded_file)
    bytes_data = uploaded_file.getvalue()
    with open(uploaded_file.name, "wb") as file:
        file.write(bytes_data)
    st.image(uploaded_file, caption="Uploaded Image", use_container_width=True)

    st.text('Processing img2text...')
    scenario = img2text(uploaded_file.name)
    st.write(scenario)

    st.text('Generating a story...')
    story = text2story(scenario)
    st.write(story)

    st.text('Generating audio data...')
    audio_data =text2audio(story)

    if st.button("Story Time!"):
        st.audio(audio_data['audio'], format="audio/wav", start_time=0, sample_rate = audio_data['sampling_rate'])