paddington-eurostar/tests/test_eurostar_scraper.py

from typing import Any

import pytest
from scraper.eurostar import _parse_graphql, _parse_graphql_leg, search_url


def _gql_response(journeys: list[dict[str, Any]]) -> dict[str, Any]:
    return {"data": {"journeySearch": {"outbound": {"journeys": journeys}}}}


def _journey(
    departs: str,
    arrives: str,
    price: float | None = None,
    seats: int | None = None,
    service_name: str = "",
    carrier: str = "ES",
    plus_price: float | None = None,
    plus_seats: int | None = None,
) -> dict[str, Any]:
    fares: list[dict[str, Any]] = [
        {
            "classOfService": {"code": "STANDARD"},
            "prices": {"displayPrice": price},
            "seats": seats,
            "legs": (
                [{"serviceName": service_name, "serviceType": {"code": carrier}}]
                if service_name
                else []
            ),
        }
    ]
    if plus_price is not None or plus_seats is not None:
        fares.append(
            {
                "classOfService": {"code": "PLUS"},
                "prices": {"displayPrice": plus_price},
                "seats": plus_seats,
                "legs": [],
            }
        )
    return {
        "timing": {"departureTime": departs, "arrivalTime": arrives},
        "fares": fares,
    }


# ---------------------------------------------------------------------------
# _parse_graphql
# ---------------------------------------------------------------------------


def test_parse_graphql_single_journey() -> None:
    data = _gql_response(
        [_journey("09:31", "12:55", price=156, seats=37, service_name="9014")]
    )
    services = _parse_graphql(data, "Paris Gare du Nord")
    assert len(services) == 1
    s = services[0]
    assert s["depart_st_pancras"] == "09:31"
    assert s["arrive_destination"] == "12:55"
    assert s["destination"] == "Paris Gare du Nord"
    assert s["train_number"] == "ES 9014"
    assert s["price"] == 156.0
    assert s["seats"] == 37
    assert s["plus_price"] is None
    assert s["plus_seats"] is None


def test_parse_graphql_standard_premier_price() -> None:
    data = _gql_response(
        [
            _journey(
                "09:31",
                "12:55",
                price=156,
                seats=37,
                service_name="9014",
                plus_price=220,
                plus_seats=12,
            )
        ]
    )
    services = _parse_graphql(data, "Paris Gare du Nord")
    assert len(services) == 1
    s = services[0]
    assert s["price"] == 156.0
    assert s["seats"] == 37
    assert s["plus_price"] == 220.0
    assert s["plus_seats"] == 12


def test_parse_graphql_plus_price_none_when_not_returned() -> None:
    data = _gql_response([_journey("09:31", "12:55", price=156, seats=37)])
    services = _parse_graphql(data, "Paris Gare du Nord")
    assert services[0]["plus_price"] is None
    assert services[0]["plus_seats"] is None


def test_parse_graphql_half_pound_price() -> None:
    data = _gql_response(
        [_journey("09:01", "14:20", price=192.5, seats=25, service_name="9116")]
    )
    services = _parse_graphql(data, "Amsterdam Centraal")
    assert services[0]["price"] == 192.5


def test_parse_graphql_null_price() -> None:
    data = _gql_response([_journey("06:16", "11:09", price=None, seats=0)])
    services = _parse_graphql(data, "Amsterdam Centraal")
    assert services[0]["price"] is None
    assert services[0]["seats"] == 0


def test_parse_graphql_sorted_by_departure() -> None:
    data = _gql_response(
        [
            _journey("10:31", "13:55"),
            _journey("07:31", "10:59"),
        ]
    )
    services = _parse_graphql(data, "Paris Gare du Nord")
    assert services[0]["depart_st_pancras"] == "07:31"
    assert services[1]["depart_st_pancras"] == "10:31"


def test_parse_graphql_deduplicates_same_departure_time() -> None:
    data = _gql_response(
        [
            _journey("06:16", "11:09", price=None, seats=0),
            _journey("06:16", "11:09", price=None, seats=0),
            _journey("06:16", "11:09", price=None, seats=0),
        ]
    )
    services = _parse_graphql(data, "Amsterdam Centraal")
    assert len(services) == 1


def test_parse_graphql_no_legs_gives_empty_train_number() -> None:
    data = _gql_response(
        [_journey("09:31", "12:55", price=156, seats=37, service_name="")]
    )
    services = _parse_graphql(data, "Paris Gare du Nord")
    assert services[0]["train_number"] == ""


def test_parse_graphql_empty_journeys() -> None:
    data = _gql_response([])
    assert _parse_graphql(data, "Paris Gare du Nord") == []


def test_parse_graphql_inbound_leg() -> None:
    data: dict[str, Any] = {
        "data": {
            "journeySearch": {
                "inbound": {
                    "journeys": [
                        _journey(
                            "17:12", "18:30", price=49, seats=43, service_name="9035"
                        )
                    ]
                }
            }
        }
    }
    services = _parse_graphql_leg(data, "Paris Gare du Nord", "inbound", "inbound")

    assert services == [
        {
            "depart_destination": "17:12",
            "arrive_st_pancras": "18:30",
            "destination": "Paris Gare du Nord",
            "train_number": "ES 9035",
            "price": 49.0,
            "seats": 43,
            "plus_price": None,
            "plus_seats": None,
        }
    ]


# ---------------------------------------------------------------------------
# search_url
# ---------------------------------------------------------------------------


def test_search_url() -> None:
    url = search_url("Paris Gare du Nord", "2026-04-10")
    assert url == (
        "https://www.eurostar.com/search/uk-en"
        "?adult=1&origin=7015400&destination=8727100&outbound=2026-04-10"
    )


def test_search_url_return() -> None:
    url = search_url("Paris Gare du Nord", "2026-04-10", return_date="2026-04-17")
    assert url.endswith("&outbound=2026-04-10&inbound=2026-04-17")