app.py

import subprocess
import os
import gradio as gr
import pandas as pd
import time
import threading
from huggingface_hub import HfApi
api = HfApi()

HF_TOKEN = os.getenv('HF_TOKEN')
repo_url = "https://huggingface.co/datasets/Weyaxi/compute-power-leaderboard"
os.system(f"git clone --bare --filter=blob:none {repo_url}")

os.chdir("compute-power-leaderboard.git")

result = subprocess.check_output("git log -1 --pretty=%B", shell=True, universal_newlines=True).replace("Upload", "").replace("/data.csv with huggingface_hub", "").strip().replace(" ", "%20")

os.system(f"wget -Odata.csv https://huggingface.co/datasets/Weyaxi/compute-power-leaderboard/resolve/main/{result}/data.csv")


def clickable(x):
    return f'<a target="_blank" href="https://huggingface.co/{x}" style="color: var(--link-text-color); text-decoration: underline;text-decoration-style: dotted;">{x}</a>'

def apply_headers(df, headers):
    tmp = df.copy()
    tmp.columns = headers

    return tmp


def search(search_text):
    if not search_text:
        return df

    return df[df['👤 Author'].str.contains(search_text, case=False, na=False)]


def restart_space():
  time.sleep(36000)
  api.restart_space(repo_id="Weyaxi/compute-power-leaderboard", token=HF_TOKEN)


df = pd.read_csv("data.csv")

df_author_copy = df.copy()

df["Author"] = df["Author"].apply(lambda x: clickable(x))
df = df.sort_values(by='TFLOPS', ascending=False)
df['Serial Number'] = [i for i in range(1, len(df)+1)]
df = df[['Serial Number', "Author", "TFLOPS", "Type"]];

df['Type'] = df['Type'].apply(lambda x: f'<img src="https://huggingface.co{x}" width="65" height="75">')

df = apply_headers(df, ["🔢 Serial Number", "👤 Author", "🖥️ TFLOPS", "🏷️ Type"])
desc = f"""
🎯 The Leaderboard aims to track users compute power in Huggingface.

## 📄 Information

🛠️ This leaderboard consists of 4000 users scraped from [🤗 Huggingface Leaderboard](https://huggingface.co/spaces/Weyaxi/huggingface-leaderboard).

These 4000 users have been selected based on their [🤗 Huggingface Leaderboard](https://huggingface.co/spaces/Weyaxi/huggingface-leaderboard) positions:

- 🤖 Top 2250 authors in the models category

- 📊 Top 1100 authors in the datasets category

- 🚀 Top 1100 authors in the spaces category

**Note that the majority of this 4000 users didn't inlclude their compute power.**

## 🤝 I want to see someone here

No problem, you can request to add a user [here](https://huggingface.co/spaces/Weyaxi/compute-power-leaderboard/discussions/1).

There is no critique; please request anyone. The number of users in this leaderboard is limited because scraping 250k user's follower count is challenging. 🙂

## Last Update

⌛ This space information is last updated in **{result.replace("%20", " ")}**.
"""

title = """
<div style="text-align:center">
  <h1 id="space-title">🖥️ Compute Power Leaderboard 🖥️</h1>
</div>
"""

with gr.Blocks() as demo:
    gr.Markdown("""<h1 align="center" id="space-title">🖥️ Compute Power Leaderboard 🖥️</h1>""")
    gr.Markdown("""<h1 align="center" id="space-title"><img src="https://huggingface.co/front/assets/hardware/gpu.png" width="75" height="75" style="display:inline-block;"><img src="https://huggingface.co/front/assets/hardware/apple-silicon.svg" width="75" height="75" style="display:inline-block;"><img src="https://huggingface.co/front/assets/hardware/cpu.png" width="75" height="75" style="display:inline-block;"></h1>""")
    gr.Markdown(desc)
    with gr.Column(min_width=320):
        search_bar = gr.Textbox(placeholder="🔍 Search for a author", show_label=False)

    gr_followers = gr.Dataframe(df, interactive=False, datatype=["number", 'markdown', 'number', 'markdown'])

    search_bar.submit(fn=search, inputs=search_bar, outputs=gr_followers)


threading.Thread(target=restart_space).start()
demo.launch()