agenda/validate_yaml.py

#!/usr/bin/python3
"""Load YAML data to ensure validity."""

import os
import sys
import typing
from datetime import date, timedelta

import yaml
from rich.pretty import pprint

import agenda
import agenda.conference
import agenda.data
import agenda.travel
import agenda.trip
import agenda.types

config = __import__("config.default", fromlist=[""])
data_dir = config.PERSONAL_DATA

currencies = set(config.CURRENCIES + ["GBP"])


def check_currency(item: agenda.types.StrDict) -> None:
    """Throw error if currency is not in config."""
    currency = item.get("currency")
    if not currency or currency in currencies:
        return None
    pprint(item)
    print(f"currency {currency!r} not in {currencies!r}")
    sys.exit(-1)


def check_trips() -> None:
    """Check trips."""
    trip_list = agenda.trip.build_trip_list(data_dir)
    print(len(trip_list), "trips")

    coords, routes = agenda.trip.get_coordinates_and_routes(trip_list, data_dir)
    print(len(coords), "coords")
    print(len(routes), "routes")


def check_flights(airlines: set[str]) -> None:
    """Check flights and ensure they are in chronological order."""
    bookings = agenda.travel.parse_yaml("flights", data_dir)

    prev_first_depart = None
    for booking in bookings:
        assert all(flight["airline"] in airlines for flight in booking["flights"])
        check_currency(booking)

        if prev_first_depart:
            assert (
                booking["flights"][0]["depart"] > prev_first_depart
            ), "Bookings are not in chronological order by first flight's departure."
        prev_first_depart = booking["flights"][0]["depart"]

    print(len(bookings), "flights")


def check_trains() -> None:
    """Check trains."""
    trains = agenda.travel.parse_yaml("trains", data_dir)
    print(len(trains), "trains")


def check_conferences() -> None:
    """Check conferences."""
    filepath = os.path.join(data_dir, "conferences.yaml")
    conferences = [
        agenda.conference.Conference(**conf)
        for conf in yaml.safe_load(open(filepath, "r"))
    ]
    for conf in conferences:
        if not conf.currency or conf.currency in currencies:
            continue
        pprint(conf)
        print(f"currency {conf.currency!r} not in {currencies!r}")
        sys.exit(-1)

    print(len(conferences), "conferences")


def check_events() -> None:
    """Check events."""
    today = date.today()
    last_year = today - timedelta(days=365)
    next_year = today + timedelta(days=2 * 365)

    events = agenda.events_yaml.read(data_dir, last_year, next_year)
    print(len(events), "events")


def check_coordinates(item: agenda.types.StrDict) -> None:
    """Check coordinate are valid."""
    if "latitude" not in item and "longitude" not in item:
        return
    assert "latitude" in item and "longitude" in item
    assert all(isinstance(item[key], (int, float)) for key in ("latitude", "longitude"))


def check_accommodation() -> None:
    """Check accommodation."""
    filepath = os.path.join(data_dir, "accommodation.yaml")
    accommodation_list = yaml.safe_load(open(filepath))

    required_fields = ["type", "name", "country", "location", "trip", "from", "to"]

    for stay in accommodation_list:
        try:
            assert all(field in stay for field in required_fields)
            check_coordinates(stay)
        except AssertionError:
            pprint(stay)
            raise

        check_currency(stay)

    print(len(accommodation_list), "stays")


def check_airports() -> None:
    """Check airports."""
    airports = typing.cast(
        dict[str, agenda.types.StrDict], agenda.travel.parse_yaml("airports", data_dir)
    )
    print(len(airports), "airports")
    for airport in airports.values():
        assert "country" in airport
        assert agenda.get_country(airport["country"])


def check_stations() -> None:
    """Check stations."""
    stations = agenda.travel.parse_yaml("stations", data_dir)
    print(len(stations), "stations")
    for station in stations:
        assert "country" in station
        assert agenda.get_country(station["country"])


def check_airlines() -> list[agenda.types.StrDict]:
    """Check airlines."""
    airlines = agenda.travel.parse_yaml("airlines", data_dir)
    print(len(airlines), "airlines")
    for airline in airlines:
        assert airline.keys() == {"icao", "iata", "name"}
        assert len(airline["icao"]) == 3
        assert len(airline["iata"]) == 2

    return airlines


def check() -> None:
    """Validate personal data YAML files."""
    airlines = check_airlines()
    check_trips()
    check_flights({airline["iata"] for airline in airlines})
    check_trains()
    check_conferences()
    check_events()
    check_accommodation()
    check_airports()
    check_stations()


if __name__ == "__main__":
    check()