app.py

import streamlit as st
import numpy as np
import pandas as pd
from datasets import load_dataset

st.set_page_config(layout="wide")

col1, col2 = st.columns([2, 3])  # Adjust the width ratio as needed

sources = [
    "https://huggingface.co/datasets/cfahlgren1/hub-stats",
    "https://huggingface.co/datasets/maxiw/hf-posts",
]

with col1:
    st.header("HuggingFace 🤗 Posts leaderboard")

with col2:
    selected_source = st.selectbox(
        "Data Source:",
        options=sources,
        index=0,
    )

if selected_source == sources[0]:
    try:
        df = pd.read_parquet("hf://datasets/cfahlgren1/hub-stats/posts.parquet")
        # ds = load_dataset("cfahlgren1/hub-stats", "posts")
        # df = pd.DataFrame(ds['train']).info()
        df["Name"] = df.fullname
        df["username"] = df.name
    except Exception as exp:
        st.error(f'''
        ERROR>> in loading {selected_source}
        
        >> {exp}''', icon="🚨")
        selected_source = sources[1]
        st.info(f'''
        This can be solved by "Space Restart"
        
        Switching Sources for now...
        
        New Source: {selected_source}''', icon="ℹ️")
        
        
    

if selected_source == sources[1]:
    df = pd.read_json("hf://datasets/maxiw/hf-posts/posts.jsonl", lines=True)

    df["publishedAt"] = pd.to_datetime(df.publishedAt)
    print(">>> ", df.columns)

    df["Name"] = df.author.apply(lambda x: x["fullname"])
    df["username"] = df.author.apply(lambda x: x["name"])

# Define the metrics
metrics = ["totalUniqueImpressions", "totalReactions", "numComments", "Num of posts"]


# Get min and max dates from the DataFrame
min_date = df["publishedAt"].min().to_pydatetime()
max_date = df["publishedAt"].max().to_pydatetime()

# Create columns for the slider and the selectbox
col1, col2 = st.columns([3, 1])  # Adjust the width ratio as needed

with col1:
    date_range = st.slider(
        "Select Date Range",
        min_value=min_date,
        max_value=max_date,
        value=(min_date, max_date),
        format="DD/MMM/YYYY",
    )

with col2:
    selected_metric = st.selectbox(
        "Sort by:",
        options=metrics,
        index=0,
    )


# Filter the DataFrame based on selected date range
mask = df["publishedAt"].between(*date_range)
df = df[mask]


df["totalReactions"] = df.reactions.apply(lambda x: sum([_["count"] for _ in x]))
df["Num of posts"] = 1

# Ensure metrics columns are integers, handling NaN values
df[metrics] = df[metrics].fillna(0).astype(int)

data = (
    df.groupby(["username", "Name"])[metrics]
    .sum()
    .sort_values(selected_metric, ascending=False)
    .reset_index()
)
data.index = np.arange(1, len(data) + 1)
data.index.name = "Rank"

# Format metrics columns with commas
data[metrics] = data[metrics].applymap(lambda x: f"{x:,}")


def make_clickable(val):
    return f'<a target="_blank" href="https://huggingface.co/{val}">{val}</a>'


df_styled = data.style.format({"username": make_clickable})
st.write(
    f"""<center>{df_styled.to_html(escape=False, index=False)}""",
    unsafe_allow_html=True,
)