import streamlit as st from huggingface_hub import HfApi import asyncio import os import plotly.express as px # Initialize the Hugging Face API api = HfApi() # Directory to save the generated HTML files HTML_DIR = "generated_html_pages" if not os.path.exists(HTML_DIR): os.makedirs(HTML_DIR) # Default list of Hugging Face usernames - where all the magic begins! 🪄 default_users = { "users": [ "awacke1", "rogerxavier", "jonatasgrosman", "kenshinn", "Csplk", "DavidVivancos", "cdminix", "Jaward", "TuringsSolutions", "Severian", "Wauplin", "phosseini", "Malikeh1375", "gokaygokay", "MoritzLaurer", "mrm8488", "TheBloke", "lhoestq", "xw-eric", "Paul", "Muennighoff", "ccdv", "haonan-li", "chansung", "lukaemon", "hails", "pharmapsychotic", "KingNish", "merve", "ameerazam08", "ashleykleynhans" ] } # Asynchronous function to fetch user content - because why wait when you can multitask? 🚀 async def fetch_user_content(username): try: # Fetch models and datasets - the stars of our show! 🌟 models = list(await asyncio.to_thread(api.list_models, author=username)) # Convert generator to list datasets = list(await asyncio.to_thread(api.list_datasets, author=username)) # Convert generator to list return { "username": username, "models": models, "datasets": datasets } except Exception as e: # Oops! Something went wrong - blame it on the gremlins! 😈 return {"username": username, "error": str(e)} # Fetch all users concurrently - more hands (or threads) make light work! 💪 async def fetch_all_users(usernames): tasks = [fetch_user_content(username) for username in usernames] return await asyncio.gather(*tasks) # Generate HTML content for a user and save it to a file - because who doesn't love a good download link? 💾 def generate_html_page(username, models, datasets): html_content = f""" {username}'s Hugging Face Content

{username}'s Hugging Face Profile

🔗 Profile Link

🧠 Models

{model_name}

📚 Datasets

{dataset_name}

""" # Save the HTML content to a file html_file_path = os.path.join(HTML_DIR, f"{username}.html") with open(html_file_path, "w") as html_file: html_file.write(html_content) return html_file_path # Cache the HTML file path using Streamlit's caching decorator @st.cache_data(show_spinner=False) def get_cached_html_file(username): return generate_html_page(username, *get_user_content(username)) # Fetch user content from the API (without caching) def get_user_content(username): user_data = asyncio.run(fetch_user_content(username)) if "error" in user_data: return None, user_data["error"] return user_data["models"], user_data["datasets"] # Streamlit app setup - the nerve center of our operation! 🎛️ st.title("Hugging Face User Content Display - Let's Automate Some Fun! 🎉") # Convert the default users list to a string - because nobody likes typing out long lists! 📝 default_users_str = "\n".join(default_users["users"]) # Text area with default list of usernames - feel free to add your friends! 👥 usernames = st.text_area("Enter Hugging Face usernames (one per line):", value=default_users_str, height=300) # Show User Content button - the big red button! (But actually it's blue) 🖱️ if st.button("Show User Content"): if usernames: username_list = [username.strip() for username in usernames.split('\n') if username.strip()] # Collect statistics for Plotly graphs stats = {"username": [], "models_count": [], "datasets_count": []} st.markdown("### User Content Overview") for username in username_list: with st.container(): # Profile link - because everyone deserves their 15 seconds of fame! 🎤 st.markdown(f"**{username}** [🔗 Profile](https://huggingface.co/{username})") # Generate HTML page and provide download link - because who wouldn't want a custom webpage? 🌐 models, datasets = get_user_content(username) if models is None: st.warning(f"{username}: {datasets} - Looks like the AI needs a coffee break ☕") else: html_file_path = get_cached_html_file(username) st.markdown(f"[📄 Download {username}'s HTML Page]({html_file_path})") # Add to statistics for Plotly graphs stats["username"].append(username) stats["models_count"].append(len(models)) stats["datasets_count"].append(len(datasets)) # Models section with expander - 🧠 because AI models are brainy! 🧠 with st.expander(f"🧠 Models ({len(models)})", expanded=False): if models: for model in models: model_name = model.modelId.split("/")[-1] st.markdown(f"- [{model_name}](https://huggingface.co/{model.modelId})") else: st.markdown("No models found. Did you check under the rug? 🕵️‍♂️") # Datasets section with expander - 📚 because data is the foundation of AI! 📚 with st.expander(f"📚 Datasets ({len(datasets)})", expanded=False): if datasets: for dataset in datasets: dataset_name = dataset.id.split("/")[-1] st.markdown(f"- [{dataset_name}](https://huggingface.co/datasets/{dataset.id})") else: st.markdown("No datasets found. Maybe they’re still baking in the oven? 🍪") st.markdown("---") # Plotly graphs to visualize the number of models and datasets each user has if stats["username"]: st.markdown("### User Content Statistics") # Plotting the number of models per user fig_models = px.bar(x=stats["username"], y=stats["models_count"], labels={'x':'Username', 'y':'Number of Models'}, title="Number of Models per User") st.plotly_chart(fig_models) # Plotting the number of datasets per user fig_datasets = px.bar(x=stats["username"], y=stats["datasets_count"], labels={'x':'Username', 'y':'Number of Datasets'}, title="Number of Datasets per User") st.plotly_chart(fig_datasets) else: st.warning("Please enter at least one username. Don't be shy! 😅") # Sidebar instructions - just in case you get lost! 🗺️ st.sidebar.markdown(""" ## How to use: 1. The text area is pre-filled with a list of Hugging Face usernames. You can edit this list or add more usernames. 2. Click 'Show User Content'. 3. View the user's models and datasets along with a link to their Hugging Face profile. 4. Download an HTML page for each user to use the absolute links offline! 5. Check out the statistics visualizations at the end! """)