tags.py

import os
import glob
from collections import Counter

files_path = "./tags2"
save_path = "./tags2/fixed"
ext = "txt"
len_ext = len(ext) + 1
remove_list = ["emataso1"]
add_list = ["iom", "virtual youtuber", "dress"]
add_back = False


if not os.path.exists(save_path):
    os.makedirs(save_path)
    
tags = []

for file_path in glob.glob(f"{files_path}/*.{ext}"):
    current_tags = []

    file_path = file_path.replace("\\", "/")
    filename = file_path.split("/")[-1][:-len_ext]

    with open(file_path, "r", encoding="utf-8") as read_file:
        current_tags = read_file.read().split(", ")

    for remove_tag in remove_list:
        if remove_tag in current_tags:
            current_tags.remove(remove_tag)
    
    filtered_add_list = list(filter(lambda x: x not in current_tags, add_list))
    if not add_back:
        current_tags[:0] = filtered_add_list
    else:
        current_tags.extend(filtered_add_list)

    tags.extend(current_tags)

    with open(f"{save_path}/{filename}.{ext}", "w", encoding="utf-8") as save_file:
        save_file.write(", ".join(current_tags))

counter = dict(Counter(tags))
counter = dict(sorted(counter.items(), key=lambda x: x[1], reverse=True))

with open(f"{save_path}/results.txt", "w", encoding="utf-8") as result_file:
    tags_one_liner = ", ".join(counter.keys())
    tags_multi_line = [f"{i} : {j}" for i, j in counter.items()]
    lines = [tags_one_liner, "\n\n\n", "\n".join(tags_multi_line)]
    
    result_file.writelines(lines)