eurotunnel-scrape/web_view.py

#!/usr/bin/python3

import os
from dataclasses import dataclass
from datetime import UTC, datetime, date
from decimal import Decimal

import flask
import lxml.html
import pytz

app = flask.Flask(__name__)
app.debug = True

data_loc = os.path.expanduser("~edward/lib/data/eurotunnel")


def get_filename(direction: str) -> tuple[datetime, str]:
    """Most recent list of outbound prices."""
    end = f"_{direction}.html"
    most_recent = max([f for f in os.listdir(data_loc) if f.endswith(end)])
    filename = os.path.join(data_loc, most_recent)
    timestamp = most_recent.removesuffix(end)
    dt_utc = datetime.strptime(timestamp, "%Y-%m-%d_%H%M%S").replace(tzinfo=UTC)
    dt = dt_utc.astimezone(pytz.timezone("Europe/London"))
    return (dt, filename)


@dataclass
class Train:
    """Eurostar train."""

    dep: str
    arr: str
    price: Decimal | None = None


def get_tickets(filename: str) -> tuple[date, list[Train]]:
    """Get trains and prices."""
    tree = lxml.html.parse(filename)
    root = tree.getroot()

    trains = []
    by_time = {}

    day_div = root.find(".//div[@class='col-md-1 als-item selected']")
    assert day_div
    day_id = day_div.get("data-id")
    assert day_id
    d = date.fromisoformat(day_id[1:])

    for mission in root.findall(".//div[@data-mission]"):
        dep_text = mission.findtext(".//b")
        assert dep_text
        dep = dep_text.replace(":", "")
        arr_text = mission.findtext(".//small")
        assert arr_text
        arr = arr_text.removeprefix("Arrive ").replace(":", "")
        item = Train(dep=dep, arr=arr)
        trains.append(item)
        by_time[dep] = item

    for ticket in root.xpath("//div[contains(@class, 'ticket')]"):
        classes = ticket.get("class").split(" ")
        if "ticket" not in classes:
            continue
        ticket_class = classes[-1]
        if ticket_class != "standard":
            continue

        mission_time = ticket.getparent().getparent().get("data-mission-time")

        onclick = ticket.get("onclick").split(",")
        price = Decimal(onclick[2][1:-1])

        by_time[mission_time].price = price

    return (d, trains)


@app.route("/")
def index() -> str:
    """Index."""
    out_ts, out_filename = get_filename("outbound")
    out_date, out = get_tickets(out_filename)
    out = [t for t in out if t.dep > "0800" and "0700" < t.arr < "2200"]

    back_ts, back_filename = get_filename("return")
    back_date, back = get_tickets(back_filename)
    back = [t for t in back if t.dep > "1100" and "0700" < t.arr < "2200"]

    return flask.render_template(
        "index.html",
        out_ts=out_ts,
        out_date=out_date,
        out=out,
        back_ts=back_ts,
        back_date=back_date,
        back=back,
    )


if __name__ == "__main__":
    app.run(host="0.0.0.0")