import streamlit as st import json from datetime import datetime from transformers import pipeline from diffusers import StableDiffusionPipeline import torch from io import BytesIO import base64 # Page configuration st.set_page_config(page_title="ReportRocket: Harvard-Style", page_icon="🚀", layout="wide") # Initialize AI models @st.cache_resource def load_models(): return { 'text_generator': pipeline('text-generation', model='distilgpt2'), 'summarizer': pipeline('summarization', model='facebook/bart-large-cnn'), 'image_generator': StableDiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16).to("cuda") } models = load_models() # Initialize session state if 'project_title' not in st.session_state: st.session_state.project_title = "" if 'project_description' not in st.session_state: st.session_state.project_description = "" if 'report_sections' not in st.session_state: st.session_state.report_sections = { 'Abstract': '', 'Introduction': '', 'Literature Review': '', 'Methodology': '', 'Results': '', 'Discussion': '', 'Conclusion': '', 'References': '' } if 'images' not in st.session_state: st.session_state.images = [] # Function to generate table of contents def generate_table_of_contents(): toc = "# Table of Contents\n\n" for i, section in enumerate(st.session_state.report_sections.keys(), 1): toc += f"{i}. [{section}](#{section.lower().replace(' ', '-')})\n" return toc # Function to generate images using Stable Diffusion @st.cache_data def generate_images(prompt, num_images=3): images = [] for _ in range(num_images): image = models['image_generator'](prompt, num_inference_steps=50).images[0] buffered = BytesIO() image.save(buffered, format="PNG") img_str = base64.b64encode(buffered.getvalue()).decode() images.append(f"data:image/png;base64,{img_str}") return images # Main UI st.title("🚀 ReportRocket: Harvard-Style Project Report Generator") # Project Input Section st.header("Project Details") project_title = st.text_input("Project Title", st.session_state.project_title) project_description = st.text_area("Project Description", st.session_state.project_description, height=150) if st.button("Generate Report Structure and Images"): if project_title and project_description: with st.spinner("Generating report structure and images..."): st.session_state.project_title = project_title st.session_state.project_description = project_description # Generate abstract abstract = models['summarizer'](project_description, max_length=150, min_length=50, do_sample=False)[0]['summary_text'] st.session_state.report_sections['Abstract'] = abstract # Generate other sections (placeholder logic, expand as needed) for section in st.session_state.report_sections: if section != 'Abstract' and not st.session_state.report_sections[section]: st.session_state.report_sections[section] = f"Generated content for {section} based on project description." # Generate images st.session_state.images = generate_images(project_title) st.success("Report structure generated and images created successfully!") else: st.error("Please provide both project title and description.") # Display Table of Contents st.header("Table of Contents") st.markdown(generate_table_of_contents()) # Report Editing Section st.header("Report Sections") selected_section = st.selectbox("Select a section to edit:", list(st.session_state.report_sections.keys())) st.session_state.report_sections[selected_section] = st.text_area( f"Edit {selected_section}:", st.session_state.report_sections[selected_section], height=300 ) # Display Images st.header("Project Images") if st.session_state.images: cols = st.columns(len(st.session_state.images)) for i, img_data in enumerate(st.session_state.images): cols[i].image(img_data) else: st.info("No images generated yet. Generate report structure to create images.") # AI Assistance if st.button("Generate Content for Section"): with st.spinner("Generating content..."): prompt = f"Generate content for {selected_section} of a project titled '{st.session_state.project_title}'" generated_text = models['text_generator'](prompt, max_length=200, num_return_sequences=1)[0]['generated_text'] st.session_state.report_sections[selected_section] += "\n" + generated_text st.success("Content generated. Please review and edit as needed.") st.experimental_rerun() # Additional Features st.header("Additional Features") # ER Diagram Generator (placeholder) if st.button("Generate ER Diagram"): st.info("ER Diagram generation feature is under development.") # Plagiarism Check (placeholder) if st.button("Check for Plagiarism"): st.info("Plagiarism check feature is under development.") # Paraphrasing Tool if st.button("Paraphrase Selected Section"): with st.spinner("Paraphrasing..."): paraphrased_text = models['summarizer'](st.session_state.report_sections[selected_section], max_length=len(st.session_state.report_sections[selected_section].split()), min_length=10, do_sample=True)[0]['summary_text'] st.session_state.report_sections[selected_section] = paraphrased_text st.success("Content paraphrased. Please review and edit as needed.") st.experimental_rerun() # Save and Download if st.button("Save and Download Report"): report_data = { "title": st.session_state.project_title, "description": st.session_state.project_description, "table_of_contents": generate_table_of_contents(), "sections": st.session_state.report_sections, "images": st.session_state.images, "timestamp": datetime.now().isoformat() } json_data = json.dumps(report_data, indent=2) st.download_button( label="Download Report as JSON", file_name=f"{st.session_state.project_title.replace(' ', '_')}_report.json", mime="application/json", data=json_data ) st.success("Report saved and ready for download!") # Footer st.markdown("---") st.markdown("Built with ❤️ using Streamlit and Hugging Face")