file_path = "linkss.txt"
imdb_ids_file = "importer/ani-delete-imdb-ids.txt"

with open(imdb_ids_file, "r", encoding="utf-8") as f:
    imdb_ids = set(line.strip() for line in f if line.strip())

with open(file_path, "r", encoding="utf-8") as f:
    lines = f.readlines()

filtered_lines = []
for line in lines:
    if "فیلم" in line and "dubbed" in line and ".vtt" in line:
        continue
    if any(imdb_id in line for imdb_id in imdb_ids):
        continue
    if "tt0413573" in line:
        continue
    if ";ر" in line:
        line = line.replace(";ر", ";")
    if "httpd\\s://" in line:
        line = line.replace("httpd\\s://", "https://")
    if ".2020./" in line:
        line = line.replace(".2020./", ".2020/")
    if ".vtthttps://" in line:
        import re

        # فقط vtt هایی که چسبیده به https هستن
        broken_vtts = re.findall(r'([^;\s]+\.vtt)https://[^;\s]+', line)

        if broken_vtts:
            # فقط اولین vtt رو نگه می‌داریم
            first_vtt = broken_vtts[0]

            # حذف همه vttهای خراب
            line = re.sub(r'[^;\s]+\.vtthttps://[^;\s]+', '', line)

            # اضافه کردن vtt سالم فقط یک‌بار
            line = line.replace(
                first_vtt,
                f'{first_vtt};https://',
                1
            )

            # تمیزکاری ;
            line = re.sub(r';{2,}', ';', line)

    filtered_lines.append(line)

with open(file_path, "w", encoding="utf-8") as f:
    f.writelines(filtered_lines)

print(f"Lines filtered and cleaned, saved to {file_path}")