Organize Youtube subscriptions into categories

Sort YT Subs.py

import os
import tkinter as tk
from tkinter import filedialog
from webbrowser import open
import pandas as pd


info = """
Youtube Subscription Organizer

0. Go to Google Takeout and export your Youtube data (change HTML option to CSV)
1. Open your subscriptions.csv file (or place this script near it)
2. Use the following instructions to categorize

Enter blank to open channel in browser
Enter x or X or 0 to archive
Enter any words to create new category and add this channel to it
Enter valid category number/exact name to add channel to existing category
Enter "category name | note here" to add optional note about the channel

Tip: You can close this program and reopen to resume, it'll skip categorized items
"""


def select_file():
    """Selects a file to process. If 'subscriptions.csv' exists, it is selected. Otherwise, a file dialog is opened."""
    file_path = "subscriptions.csv"
    if os.path.isfile(file_path):
        return file_path
    else:
        root = tk.Tk()
        root.withdraw()
        file_path = filedialog.askopenfilename(filetypes=[("CSV files", "*.csv")])
        return file_path


def get_output(output_file):
    """Reads the output file and returns its contents as a DataFrame."""
    try:
        return pd.read_csv(output_file)
    except FileNotFoundError:
        return pd.DataFrame(columns=['Channel Title', 'Channel Url', 'Category', 'Note'])


def show_categories(cat):
    """Displays the categories in a compact way."""
    formatted_cat = [f"{i if i != 0 else 'X'}: {item}" for i, item in enumerate(cat)]
    print("Categories -\t" + " | ".join(formatted_cat))


def categorize(categories, channel_id, channel_url, channel_title):
    """Categorizes a channel and returns the category and a note."""
    category = input(f"{channel_title} ({channel_id}) ~> ")
    if category.lower() == 'x': # X/x = archive
        return categories[0]
    elif category == '': # empty = open channel
        open(channel_url)
        return categorize(categories, channel_id, channel_url, channel_title)
    elif category in categories: # category = pass
        return category
    else:
        # 0 = Archive, 1 = last, -2 = penultimate, etc.
        try: # Num = categories[N]
            category_key = int(category)
            return categories[category_key]
        except ValueError: # Str = new category
            note = ''
            if ' | ' in category:
                category, note = category.split(' | ')  
            categories.append(category)
            show_categories(categories)
            if note: return category + ' | ' + note
            return category
        except IndexError: # invalid Num = retry
            print("Incorrect number input.")
            return categorize(categories, channel_id, channel_url, channel_title)


def add_entry(output_file, channel_title, channel_url, category, note):
    """Adds an entry to the output file."""
    df = get_output(output_file)
    new_entry = pd.DataFrame(
        {'Channel Title': [channel_title], 'Channel Url': [channel_url], 'Category': [category], 'Note': [note]})
    df = pd.concat([df, new_entry], ignore_index=True)
    df.to_csv(output_file, index=False)


def main():
    source_file = select_file()
    data = pd.read_csv(source_file)
    output_file = source_file.replace('.csv', '_output.csv')
    output_data = get_output(output_file)

    categories = ["Archive"]
    show_categories(categories)

    for index, row in data.iterrows():
        channel_id = row['Channel Id']
        channel_url = row['Channel Url']
        channel_title = row['Channel Title']
        
        if output_data['Channel Url'].isin([channel_url]).any():
            continue

        category, note = categorize(categories, channel_id, channel_url, channel_title), ""
        if ' | ' in category:
            category, note = category.split(" | ")
        
        add_entry(output_file, channel_title, channel_url, category, note)



if __name__ == "__main__":
    print(info+'\n')
    main()