depot/ops/nixos/swann/ee-scrape-data.py

import re
import sys
import textwrap

import bs4
import requests


BYTE_SUFFIXES = {
    "": 1,
    "B": 1,
    "kB": 1024,
    "MB": 1024 * 1024,
    "GB": 1024 * 1024 * 1024,
    "TB": 1024 * 1024 * 1024 * 1024,
}

TIME_SUFFIXES = {
    "Days": 60 * 60 * 24,
    "Day": 60 * 60 * 24,
    "Hrs": 60 * 60,
    "Hr": 60 * 60,
    "Mins": 60,
    "Min": 60,
}


def uniec(n):
    m = re.match(r"^([0-9]+(?:[.][0-9]+)?)([a-zA-Z]*)$", n)
    if not m:
        raise ValueError(f"invalid suffixed-bytes {n}")
    val, suffix = m.groups()
    val = float(val)
    if suffix not in BYTE_SUFFIXES:
        raise ValueError(f"unknown suffix {suffix} from {n}")
    return val * BYTE_SUFFIXES[suffix]


def untime(n):
    remaining = n
    x = 0
    while remaining and (m := re.match(r"^([0-9]+) ([a-zA-Z]+)(?: (.*)|$)", remaining)):
        val, suffix, remaining = m.groups()
        val = float(val)
        if suffix not in TIME_SUFFIXES:
            raise ValueError(f"unknown suffix {suffix} from {n}")
        x += val * TIME_SUFFIXES[suffix]
    return x


def scrape_data():
    resp = requests.get("http://add-on.ee.co.uk/mbbstatus")
    resp.raise_for_status()

    soup = bs4.BeautifulSoup(resp.text, "html5lib")

    data = {}

    account_number_el = soup.select("p.account__details > span")[1]
    data["account_number"] = account_number_el.get_text().strip().replace(" ", "")

    allowance_left_el = soup.select("span.allowance__left")[0]
    allowance_left_str, allowance_total_str = list(allowance_left_el.stripped_strings)
    data["allowance_left"] = uniec(allowance_left_str)
    if allowance_total_str.startswith("left of "):
        allowance_total_str = allowance_total_str[len("left of ") :]
    data["allowance_total"] = uniec(allowance_total_str)

    timespan_str = " ".join(soup.select(".allowance__timespan")[0].stripped_strings)
    if timespan_str.startswith("Lasts for "):
        timespan_str = timespan_str[len("Lasts for ") :]
    data["time_remaining"] = untime(timespan_str)

    return textwrap.dedent(
        f"""
        ee_allowance_left{{account_number="{data['account_number']}"}} {data['allowance_left']}
        ee_allowance_total{{account_number="{data['account_number']}"}} {data['allowance_total']}
        ee_time_remaining{{account_number="{data['account_number']}"}} {data['time_remaining']}
    """
    ).strip()


def main(args):
    if len(args) > 2:
        print(f"{args[0]} (output path)", file=sys.stderr)
        sys.exit(1)

    data = scrape_data()
    if len(args) == 1:
        print(data)
    else:
        with open(args[1], "w") as f:
            print(data, file=f)


if __name__ == "__main__":
    main(sys.argv)