mac2vendor from wireshark db

#!/usr/bin/env python3


import pickle
from binascii import unhexlify
from pathlib import Path
from urllib.request import urlopen

mac2vendor_file = Path.home() / ".config/mac2vendor.pickle"


def get_mac2vendor() -> dict[bytes, str]:
    if mac2vendor_file.exists():
        with mac2vendor_file.open("rb") as fd:
            return pickle.load(fd)

    url = "https://gitlab.com/wireshark/wireshark/-/raw/master/manuf"
    lines = urlopen(url).read().decode().splitlines()
    mac2vendor = {}

    for row in lines:
        if row.startswith("#") or not row.strip():
            continue

        mac, *_, company = row.strip().split("\t")
        mac = unhexlify(mac.replace(":", "")[:6])
        mac2vendor[mac] = company

    with mac2vendor_file.open("wb") as fd:
        pickle.dump(mac2vendor, fd)

    return mac2vendor


def search(mac: str) -> str | None:
    find_mac = unhexlify(mac.replace(":", "").replace("-", ""))
    for mac_db, name in get_mac2vendor().items():
        if find_mac.startswith(mac_db):
            return name

    return None


print(search("3c:7c:3f:c2:00:00"))