boarding-pass/aerodatabox.py

"""Fetch scheduled + actual gate departure/arrival times, plus actual
takeoff/landing (runway) times, from AeroDataBox (RapidAPI) - the same
provider AirTrail's own "Search" button uses internally when you fill in
a flight manually.

Confirmed directly against AirTrail's source (johanohly/AirTrail):
- src/lib/server/utils/flight-lookup/aerodatabox.ts (endpoint, headers,
  response shape, and the actualTime -> revisedTime -> scheduledTime
  fallback order for "best known actual time")
- src/lib/zod/flight.ts (target field names: departure/departureScheduled/
  arrival/arrivalScheduled expect full-seconds ISO-8601 datetimes)

Why this exists as well as flightaware.py: FlightAware's public history
page (scraped there) only gives us a GPS track - no gate/schedule times.
BCBP boarding passes don't carry a time-of-day at all. AeroDataBox is the
only piece that can fill in actual scheduled/gate-actual times without
requiring a logged-in AirTrail session (its own lookup is cookie-authed,
not reachable with our API key).

VERIFIED (2026-09-12, real key, Free Tier): a flight 4 days in the past
(VA559, 2026-09-08) returned full data - scheduledTime/revisedTime/
runwayTime on both legs. Free Tier quota is 400 API units / period, 1
unit per lookup here, so the daily one-flight job has plenty of headroom.
A missing/empty response should still fail soft (see lookup_gate_times'
return of None) rather than block the FlightAware track or the
review/approve flow - just in case an individual flight falls outside
whatever window AeroDataBox actually indexed.

Bonus found in that same real response: AeroDataBox also returns
departure.runwayTime / arrival.runwayTime - actual takeoff/landing off
the runway, not just gate times. AirTrail's own aerodatabox.ts never
reads this field at all (confirmed against its source above - it types
departure/arrival with only scheduledTime/revisedTime/actualTime, no
runwayTime), which is why AirTrail's own Search button leaves
Takeoff/Landing blank even after a successful search. We use it here for
takeoffActual/landingActual - preferred in scheduler.py over the
FlightAware-track-derived guess when present, since it's an authoritative
field rather than an inferred one (the track only tells us when the
transponder started/stopped squawking, which is close to but not exactly
wheels-up/touchdown).
"""

import json
import re
import urllib.error
import urllib.request
from datetime import date, datetime

import config

BASE_URL = "https://aerodatabox.p.rapidapi.com"

# Fields this module can fill in build_save_payload(), keyed the same way
# AirTrail's /api/flight/save expects them.
GATE_FIELDS = (
    "departureScheduled",
    "departure",
    "arrivalScheduled",
    "arrival",
    "takeoffActual",
    "landingActual",
)


class AeroDataBoxError(Exception):
    """Network/config/HTTP error - safe to retry on a later run, same
    treatment as FlightAwareTransientError."""


def _sanitize(flight_number: str) -> str:
    return re.sub(r"[\s-]", "", flight_number).upper()


def _parse_local(local: str | None) -> str | None:
    """AeroDataBox 'local' strings look like '2026-06-22 13:45+10:00' -
    already carrying their own UTC offset, no separate airport-timezone
    lookup needed. Reformats to the seconds-precision ISO-8601 string
    AirTrail's zod schema (z.string().datetime({offset: true})) requires,
    e.g. '2026-06-22T13:45:00+10:00'."""
    if not local:
        return None
    try:
        dt = datetime.strptime(local, "%Y-%m-%d %H:%M%z")
    except ValueError:
        return None
    return dt.isoformat(timespec="seconds")


def lookup_gate_times(flight_number: str, flight_date: date) -> dict | None:
    """Returns a dict with some/all of GATE_FIELDS set (None for anything
    AeroDataBox didn't have), or None if the lookup failed outright or
    matched nothing. Never raises for "no data" - only for genuine
    network/config failures (AeroDataBoxError), mirroring flightaware.py's
    distinction between "definitively not found" and "try again later"."""
    if not config.AERODATABOX_API_KEY:
        return None

    cleaned = _sanitize(flight_number)
    url = (
        f"{BASE_URL}/flights/number/{cleaned}/{flight_date.isoformat()}"
        "?dateLocalRole=Both&withAircraftImage=false&withLocation=false"
    )
    # A default urllib request (no User-Agent) gets HTTP 403 from AeroDataBox's
    # Cloudflare-fronted gateway even with a perfectly valid, subscribed key -
    # confirmed live, side by side, same key/URL: bare urllib -> 403, same
    # request + this UA -> 200 with real data. See config.BROWSER_USER_AGENT.
    req = urllib.request.Request(
        url,
        headers={
            "x-rapidapi-key": config.AERODATABOX_API_KEY,
            "User-Agent": config.BROWSER_USER_AGENT,
        },
    )
    try:
        with urllib.request.urlopen(req, timeout=20) as resp:
            raw = resp.read()
    except urllib.error.HTTPError as e:
        if e.code == 204:
            return None
        raise AeroDataBoxError(f"HTTP {e.code} from AeroDataBox") from e
    except urllib.error.URLError as e:
        raise AeroDataBoxError(f"error reaching AeroDataBox: {e}") from e

    try:
        data = json.loads(raw.decode("utf-8"))
    except (json.JSONDecodeError, UnicodeDecodeError) as e:
        raise AeroDataBoxError(f"bad AeroDataBox response: {e}") from e

    if not isinstance(data, list) or not data:
        return None

    # BCBP gives no time-of-day, so - same limitation as the FlightAware
    # step - we can't disambiguate multiple same-day legs of the same
    # flight number from the boarding pass alone. Take the first result,
    # same as AirTrail's own lookup falls back to when nothing else to
    # filter on (from/to aren't sent here, unlike AirTrail's UI search,
    # since AeroDataBox's API doesn't support filtering by them server-side
    # anyway - see the PR that added client-side from/to filtering upstream).
    flight = data[0]
    dep = flight.get("departure") or {}
    arr = flight.get("arrival") or {}

    return {
        "departureScheduled": _parse_local((dep.get("scheduledTime") or {}).get("local")),
        "departure": _parse_local(
            (dep.get("actualTime") or {}).get("local")
            or (dep.get("revisedTime") or {}).get("local")
            or (dep.get("scheduledTime") or {}).get("local")
        ),
        "arrivalScheduled": _parse_local((arr.get("scheduledTime") or {}).get("local")),
        "arrival": _parse_local(
            (arr.get("actualTime") or {}).get("local")
            or (arr.get("revisedTime") or {}).get("local")
            or (arr.get("scheduledTime") or {}).get("local")
        ),
        "takeoffActual": _parse_local((dep.get("runwayTime") or {}).get("local")),
        "landingActual": _parse_local((arr.get("runwayTime") or {}).get("local")),
    }