app.py

import os
import gradio as gr
import pandas as pd
import time
import threading
from huggingface_hub import HfApi
from humanize import naturalsize

api = HfApi()

HF_TOKEN = os.getenv('HF_TOKEN')


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)]

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

df_author_copy = df.copy()

df["author"] = df["author"].apply(lambda x: clickable(x))
df['Total Usage'] = df[['models', 'datasets', 'spaces']].sum(axis=1)
df = df.sort_values(by='Total Usage', ascending=False)

sum_all_author = naturalsize(sum(df['models'].tolist()+df['datasets'].tolist()+df['spaces'].tolist()))

naturalsize_columns = ['Total Usage', 'models', 'datasets', 'spaces']
df[naturalsize_columns] = df[naturalsize_columns].map(naturalsize)

df['Serial Number'] = [i for i in range(1, len(df)+1)]
df = df[['Serial Number', "author", "Total Usage", "models", "datasets", "spaces"]]

df = apply_headers(df, ["🔢 Serial Number", "👤 Author", "⚡️ Total Usage", "🏛️ Models", "📊 Datasets", "🚀 Spaces"])

desc = f"""
🎯 The Leaderboard aims to track authors data usage in 🤗 Huggingface.

## 📄 Information

🛠️ This leaderboard consists of 125k authors scraped from [🤗 Huggingface Leaderboard](https://huggingface.co/spaces/Weyaxi/huggingface-leaderboard).

These 125k authors have been selected based on their [🤗 Huggingface Leaderboard](https://huggingface.co/spaces/Weyaxi/huggingface-leaderboard) positions:

- 🤖 Top 60k authors in the models category

- 📊 Top 60k authors in the datasets category

- 🚀 Top 50k authors in the spaces category

## 📒 Notes

Note that these numbers may not be entirely accurate due to the following reasons:

- I only calculated the data usage from the main branch and did not include deleted files that cannot be directly seen.

- There may be large datasets/models to which I don't have access (either private or gated).

# 📶 Total Data Usage From All Authors

According to this leaderboard, there is a total of {sum_all_author} of data on this platform.
"""
# Write note maybe?


title = """
<div style="text-align:center">
  <h1 id="space-title">💾 Data Leaderboard 💾</h1>
</div>
"""

with gr.Blocks() as demo:
    gr.Markdown("""<h1 align="center" id="space-title">💾 Data Leaderboard 💾</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'])

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


demo.launch()