Edward Betts
13c4341f3a
Annotated all functions with mypy --strict-compatible types (-> None, dict[str, Any], Generator types, etc.), added # type: ignore for untyped third-party libs (lxml), and reformatted with black. All 18 source files now pass mypy --strict with zero errors. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
272 lines
9 KiB
Python
272 lines
9 KiB
Python
"""
|
|
Fetch Eurostar timetable, prices, and seat availability via the GraphQL API.
|
|
|
|
A single POST to https://site-api.eurostar.com/gateway (operationName
|
|
NewBookingSearch) returns departure time, arrival time, train number,
|
|
Eurostar Standard fare price, and seats remaining at that price for every
|
|
service on the requested date.
|
|
"""
|
|
|
|
import random
|
|
import string
|
|
from typing import Any
|
|
|
|
import requests
|
|
|
|
DEFAULT_UA = (
|
|
"Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 "
|
|
"(KHTML, like Gecko) Chrome/122.0.0.0 Safari/537.36"
|
|
)
|
|
|
|
ST_PANCRAS_STATION_ID = "7015400"
|
|
ORIGIN_STATION_ID = ST_PANCRAS_STATION_ID
|
|
|
|
DESTINATION_STATION_IDS = {
|
|
"Paris Gare du Nord": "8727100",
|
|
"Brussels Midi": "8814001",
|
|
"Lille Europe": "8722326",
|
|
"Amsterdam Centraal": "8400058",
|
|
"Rotterdam Centraal": "8400530",
|
|
"Cologne Hbf": "8015458",
|
|
}
|
|
|
|
_GATEWAY_URL = "https://site-api.eurostar.com/gateway"
|
|
|
|
# Query requesting timing, train identity, and Standard fare price + seats.
|
|
# Variable names and argument names match the site's own query so the
|
|
# server-side query planner sees a familiar shape.
|
|
_GQL_QUERY = (
|
|
"query NewBookingSearch("
|
|
"$origin:String!,$destination:String!,$outbound:String!,"
|
|
"$inbound:String,$currency:Currency!,$adult:Int,"
|
|
"$filteredClassesOfService:[ClassOfServiceEnum]"
|
|
"){"
|
|
"journeySearch("
|
|
"outboundDate:$outbound inboundDate:$inbound origin:$origin destination:$destination"
|
|
" adults:$adult currency:$currency"
|
|
' productFamilies:["PUB"] contractCode:"EIL_ALL"'
|
|
" adults16Plus:0 children:0 youths:0 children4Only:0 children5To11:0"
|
|
" infants:0 adultsWheelchair:0 childrenWheelchair:0 guideDogs:0"
|
|
" wheelchairCompanions:0 nonWheelchairCompanions:0"
|
|
" isAftersales:false multipleFlexibility:true showAllSummatedFares:false"
|
|
" seniorsAges:[] prioritiseShortHaulODTrains:true"
|
|
"){"
|
|
"outbound{"
|
|
"journeys("
|
|
"hideIndirectTrainsWhenDisruptedAndCancelled:false"
|
|
" hideDepartedTrains:true"
|
|
" hideExternalCarrierTrains:true"
|
|
" hideDirectExternalCarrierTrains:true"
|
|
"){"
|
|
"timing{departureTime:departs arrivalTime:arrives}"
|
|
"fares(filteredClassesOfService:$filteredClassesOfService){"
|
|
"classOfService{code}"
|
|
"prices{displayPrice}"
|
|
"seats "
|
|
"legs{serviceName serviceType{code}}"
|
|
"}"
|
|
"}"
|
|
"}"
|
|
"inbound{"
|
|
"journeys("
|
|
"hideIndirectTrainsWhenDisruptedAndCancelled:false"
|
|
" hideDepartedTrains:true"
|
|
" hideExternalCarrierTrains:true"
|
|
" hideDirectExternalCarrierTrains:true"
|
|
"){"
|
|
"timing{departureTime:departs arrivalTime:arrives}"
|
|
"fares(filteredClassesOfService:$filteredClassesOfService){"
|
|
"classOfService{code}"
|
|
"prices{displayPrice}"
|
|
"seats "
|
|
"legs{serviceName serviceType{code}}"
|
|
"}"
|
|
"}"
|
|
"}"
|
|
"}"
|
|
"}"
|
|
)
|
|
|
|
_STANDARD = "STANDARD"
|
|
_STANDARD_PLUS = "PLUS"
|
|
|
|
|
|
def search_url(
|
|
destination: str,
|
|
travel_date: str,
|
|
direction: str = "outbound",
|
|
return_date: str | None = None,
|
|
) -> str:
|
|
dest_id = DESTINATION_STATION_IDS[destination]
|
|
origin = ST_PANCRAS_STATION_ID
|
|
destination_id = dest_id
|
|
outbound = travel_date
|
|
inbound = return_date
|
|
if direction == "inbound":
|
|
origin, destination_id = dest_id, ST_PANCRAS_STATION_ID
|
|
inbound = None
|
|
return (
|
|
f"https://www.eurostar.com/search/uk-en"
|
|
f"?adult=1&origin={origin}&destination={destination_id}&outbound={outbound}"
|
|
+ (f"&inbound={inbound}" if inbound else "")
|
|
)
|
|
|
|
|
|
def _generate_cid() -> str:
|
|
chars = string.ascii_letters + string.digits
|
|
return "SRCH-" + "".join(random.choices(chars, k=22))
|
|
|
|
|
|
def _parse_journeys(
|
|
journeys: list[dict[str, Any]], destination: str, direction: str
|
|
) -> list[dict[str, Any]]:
|
|
"""
|
|
Parse a NewBookingSearch GraphQL response into a list of service dicts.
|
|
|
|
Each dict contains: depart_st_pancras, arrive_destination, destination,
|
|
train_number, price/seats (Standard), plus_price/plus_seats (Standard Premier).
|
|
|
|
The same St Pancras departure can appear multiple times (different
|
|
connecting trains); we keep the entry with the earliest arrival.
|
|
Multi-leg train numbers are joined with ' + ' (e.g. 'ES 9116 + ER 9329').
|
|
"""
|
|
best: dict[str, dict[str, Any]] = {}
|
|
for journey in journeys:
|
|
dep = journey["timing"]["departureTime"]
|
|
arr = journey["timing"]["arrivalTime"]
|
|
std_price = std_seats = plus_price = plus_seats = None
|
|
train_number = ""
|
|
for fare in journey.get("fares") or []:
|
|
cos = fare["classOfService"]["code"]
|
|
p = fare.get("prices")
|
|
price = float(p["displayPrice"]) if p and p.get("displayPrice") else None
|
|
seats = fare.get("seats")
|
|
if not train_number:
|
|
legs = fare.get("legs") or []
|
|
train_number = " + ".join(
|
|
f"{(leg.get('serviceType') or {}).get('code', 'ES')} {leg['serviceName']}"
|
|
for leg in legs
|
|
if leg.get("serviceName")
|
|
)
|
|
if cos == _STANDARD:
|
|
std_price, std_seats = price, seats
|
|
elif cos == _STANDARD_PLUS:
|
|
plus_price, plus_seats = price, seats
|
|
if direction == "inbound":
|
|
service = {
|
|
"depart_destination": dep,
|
|
"arrive_st_pancras": arr,
|
|
"destination": destination,
|
|
"train_number": train_number,
|
|
"price": std_price,
|
|
"seats": std_seats,
|
|
"plus_price": plus_price,
|
|
"plus_seats": plus_seats,
|
|
}
|
|
key = dep
|
|
arrive_key = "arrive_st_pancras"
|
|
else:
|
|
service = {
|
|
"depart_st_pancras": dep,
|
|
"arrive_destination": arr,
|
|
"destination": destination,
|
|
"train_number": train_number,
|
|
"price": std_price,
|
|
"seats": std_seats,
|
|
"plus_price": plus_price,
|
|
"plus_seats": plus_seats,
|
|
}
|
|
key = dep
|
|
arrive_key = "arrive_destination"
|
|
if key not in best or arr < best[key][arrive_key]:
|
|
best[key] = service
|
|
sort_key = "depart_destination" if direction == "inbound" else "depart_st_pancras"
|
|
return sorted(best.values(), key=lambda s: s[sort_key])
|
|
|
|
|
|
def _parse_graphql(data: dict[str, Any], destination: str) -> list[dict[str, Any]]:
|
|
journeys = data["data"]["journeySearch"]["outbound"]["journeys"]
|
|
return _parse_journeys(journeys, destination, "outbound")
|
|
|
|
|
|
def _parse_graphql_leg(
|
|
data: dict[str, Any], destination: str, leg: str, direction: str
|
|
) -> list[dict[str, Any]]:
|
|
journeys = data["data"]["journeySearch"][leg]["journeys"]
|
|
return _parse_journeys(journeys, destination, direction)
|
|
|
|
|
|
def _payload(
|
|
origin: str, destination_id: str, outbound: str, inbound: str | None = None
|
|
) -> dict[str, Any]:
|
|
variables: dict[str, Any] = {
|
|
"origin": origin,
|
|
"destination": destination_id,
|
|
"outbound": outbound,
|
|
"inbound": inbound,
|
|
"currency": "GBP",
|
|
"adult": 1,
|
|
"filteredClassesOfService": [_STANDARD, _STANDARD_PLUS],
|
|
}
|
|
return {
|
|
"operationName": "NewBookingSearch",
|
|
"variables": variables,
|
|
"query": _GQL_QUERY,
|
|
}
|
|
|
|
|
|
def _headers() -> dict[str, str]:
|
|
return {
|
|
"User-Agent": DEFAULT_UA,
|
|
"Content-Type": "application/json",
|
|
"Accept": "*/*",
|
|
"Accept-Language": "en-GB",
|
|
"Referer": "https://www.eurostar.com/",
|
|
"x-platform": "web",
|
|
"x-market-code": "uk",
|
|
"x-source-url": "search-app/",
|
|
"cid": _generate_cid(),
|
|
}
|
|
|
|
|
|
def fetch(
|
|
destination: str, travel_date: str, direction: str = "outbound"
|
|
) -> list[dict[str, Any]]:
|
|
"""
|
|
Return all Eurostar services for destination on travel_date.
|
|
|
|
Each dict contains timetable info (depart_st_pancras, arrive_destination,
|
|
train_number) plus pricing (price, seats) from a single GraphQL call.
|
|
"""
|
|
dest_id = DESTINATION_STATION_IDS[destination]
|
|
if direction == "inbound":
|
|
origin, destination_id = dest_id, ST_PANCRAS_STATION_ID
|
|
else:
|
|
origin, destination_id = ST_PANCRAS_STATION_ID, dest_id
|
|
resp = requests.post(
|
|
_GATEWAY_URL,
|
|
json=_payload(origin, destination_id, travel_date),
|
|
headers=_headers(),
|
|
timeout=20,
|
|
)
|
|
resp.raise_for_status()
|
|
leg_direction = "inbound" if direction == "inbound" else "outbound"
|
|
return _parse_graphql_leg(resp.json(), destination, "outbound", leg_direction)
|
|
|
|
|
|
def fetch_return(
|
|
destination: str, outbound_date: str, return_date: str
|
|
) -> dict[str, list[dict[str, Any]]]:
|
|
dest_id = DESTINATION_STATION_IDS[destination]
|
|
resp = requests.post(
|
|
_GATEWAY_URL,
|
|
json=_payload(ST_PANCRAS_STATION_ID, dest_id, outbound_date, return_date),
|
|
headers=_headers(),
|
|
timeout=20,
|
|
)
|
|
resp.raise_for_status()
|
|
data = resp.json()
|
|
return {
|
|
"outbound": _parse_graphql_leg(data, destination, "outbound", "outbound"),
|
|
"inbound": _parse_graphql_leg(data, destination, "inbound", "inbound"),
|
|
}
|