import streamlit as st
from transformers import pipeline
from PIL import Image
import requests
import torch
from diffusers import DiffusionPipeline


def generate_text(prompt, text_generator):
    generated_text = text_generator(prompt, max_length=200, num_return_sequences=1, temperature=0.7)[0]['generated_text']
    return generated_text

def generate_image(prompt, img_gen):
    generated_image = img_gen(prompt)[0]
    return generated_image

def generate_blog_post(keywords, text_generator, img_gen):
    # Text generation
    generated_text = generate_text(f"Write about {keywords}", text_generator)

    # Image generation
    generated_image = generate_image(keywords, img_gen)

    return f"# {keywords}\n\n## Introduction\n{generated_text}\n\n## Body\n{generated_text}\n\n## Conclusion\n{generated_text}\n\nGenerated Image: {generated_image}"

def main():
  

    # Load models
    text_model_name = "EleutherAI/gpt-neo-1.3B"
    text_generator = pipeline("text-generation", model=text_model_name, tokenizer=text_model_name)

    img_gen = DiffusionPipeline.from_pretrained("kandinsky-community/kandinsky-2-2-decoder")
      # Title of the app
    st.title("AI Blog Post Generator")

    # User input for keywords
    user_keywords = st.text_input("Enter keywords for the blog post:")

    # Button to generate blog post
    if st.button("Generate Blog Post"):
        # Generate blog post
        blog_post = generate_blog_post(user_keywords, text_generator, img_gen)

        # Display the generated blog post
        st.markdown(blog_post, unsafe_allow_html=True)

if __name__ == "__main__":
    main()