plot_graph.py

from huggingface_hub import snapshot_download
import os
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from datetime import datetime
from scipy.ndimage import gaussian_filter1d

snapshot_download("Weyaxi/followers-leaderboard", local_dir="followers-leaderboard", repo_type="dataset", max_workers=32)

global_dataset_dfs = {}

def gr_plot_follower_comparison(author_names, dataset_path="followers-leaderboard", smooth_sigma=1.5):
    if isinstance(author_names, str):
        author_names = [author_names]

    # Load or retrieve cached DataFrames
    if dataset_path not in global_dataset_dfs:
        dfs = {}
        for root, _, files in os.walk(dataset_path):
            if "data.csv" in files:
                file_path = os.path.join(root, "data.csv")
                folder_name = os.path.basename(root)
                date_str = '-'.join(folder_name.split('-')[:3]).split()[0]
                try:
                    date = datetime.strptime(date_str, "%d-%m-%Y")
                    df = pd.read_csv(file_path)
                    dfs[date] = df
                except (ValueError, pd.errors.ParserError):
                    continue
        global_dataset_dfs[dataset_path] = dfs
    else:
        dfs = global_dataset_dfs[dataset_path]

    data = []
    
    for author in author_names:
        date_count = {}
        for date, df in dfs.items():
            if author in df['Author'].values:
                count = df.loc[df['Author'] == author, 'Number of Followers'].iloc[0]
                date_count[date] = count

        if date_count:
            dates, counts = zip(*sorted(date_count.items()))
            counts = np.array(counts, dtype=np.float32)

            # Apply Gaussian smoothing if requested
            if smooth_sigma > 0 and len(counts) > 1:
                counts = gaussian_filter1d(counts, sigma=smooth_sigma)

            for d, c in zip(dates, counts):
                data.append({"x": d, "y": c, "author": author})

    if not data:
        return pd.DataFrame(columns=["x", "y", "author"])

    return pd.DataFrame(data)