Files
TREK/shared/src/reservation/reservation.schema.ts
T
Maurice 724c88c201 style: format server and shared with prettier
A mechanical prettier --write over server/src, server/tests, server/scripts
and shared/src with the workspaces' .prettierrc; nothing but layout and
import order changes, and every file keeps the same set of imports. shared
is formatted throughout and its format baseline is empty. The server
baseline keeps the 35 files that Prettier would grow past their lint:size
limit (they need a split first, which is not a formatting change), and the
generated entity files stay as their generator writes them.
2026-10-10 21:23:37 +02:00

397 lines
18 KiB
TypeScript

import { RESERVATION_STATUSES, RESERVATION_TYPE_KEYS } from './reservation-types';
import { z } from 'zod';
/**
* Reservation + accommodation API contract — single source of truth for the
* /api/trips/:tripId/reservations and /api/trips/:tripId/accommodations endpoints.
*
* Trip-scoped. Reservations use the 'reservation_edit' permission; accommodations
* use 'day_edit' (they live in the day/accommodation service). The legacy routes
* (server/src/routes/reservations.ts + the accommodations sub-router in
* routes/days.ts) carry several side effects — auto-creating/updating/deleting a
* linked budget item, accommodation broadcasts and booking notifications — which
* the Nest service reproduces 1:1. Reservation bodies are wide and provider-ish,
* so the create/update payloads stay mostly open with `title` pinned.
*/
const open = z.record(z.string(), z.unknown());
/**
* A reservation's `type` as the API returns it: one of the catalog's types
* (`RESERVATION_TYPES`), or whatever free text an older row or an import
* stored, which is passed through as it is.
*/
export const reservationTypeSchema = z.enum(RESERVATION_TYPE_KEYS).or(z.string());
/** A reservation's `status` as the API returns it: a known status, or a stored value outside them. */
export const reservationStatusSchema = z.enum(RESERVATION_STATUSES).or(z.string());
// Inside reservationSchema the two fields stay z.string(), described by the
// catalog. Zod 4 types a union field as optional for a consumer compiled
// without strictNullChecks (the client), so the union there would loosen the
// inferred Reservation type; the runtime answer is the same, any string.
const reservationTypeField = z.string().describe('One of RESERVATION_TYPES, or free text an older row stored');
const reservationStatusField = z.string().describe('One of RESERVATION_STATUSES, or a value an older row stored');
/**
* A reservation endpoint (flight/train leg terminal) — row of the
* reservation_endpoints table (server/src/services/reservationService.ts).
*/
export const reservationEndpointSchema = z.object({
id: z.number().optional(),
reservation_id: z.number().optional(),
role: z.enum(['from', 'to', 'stop']),
sequence: z.number(),
name: z.string(),
code: z.string().nullable(),
lat: z.number(),
lng: z.number(),
timezone: z.string().nullable(),
local_time: z.string().nullable(),
local_date: z.string().nullable(),
});
export type ReservationEndpoint = z.infer<typeof reservationEndpointSchema>;
/**
* A traveler assigned to a reservation — a trip member or a named guest (both are
* `users` rows; a guest carries is_guest=1). Mirrors the budget member read shape
* without the cost-split fields; joined + avatar-resolved in reservationService.
*/
export const reservationTravelerSchema = z.object({
user_id: z.number(),
username: z.string(),
avatar_url: z.string().nullable().optional(),
avatar: z.string().nullable().optional(),
is_guest: z.number().nullable().optional(),
});
export type ReservationTraveler = z.infer<typeof reservationTravelerSchema>;
/**
* Endpoints as accepted on a WRITE (create/update body `endpoints` array). This
* pins the STRUCTURE (role must be a known value, name/code are strings) so a
* plugin can't crash the service with a stray type, but stays permissive on the
* rest to match the service exactly: it silently drops rows whose lat/lng are
* missing and defaults sequence from the array index. So lat/lng/sequence are
* nullable+optional here — a coord-less endpoint validates and is then dropped
* downstream, byte-for-byte the pre-existing REST/importer behaviour (no breaking
* change), while `role: 'banana'` or a non-string name is still rejected up front.
*/
export const reservationEndpointsInputSchema = z.array(
z.object({
role: z.enum(['from', 'to', 'stop']),
name: z.string().min(1),
lat: z.number().nullable().optional(),
lng: z.number().nullable().optional(),
sequence: z.number().nullable().optional(),
code: z.string().nullable().optional(),
timezone: z.string().nullable().optional(),
local_time: z.string().nullable().optional(),
local_date: z.string().nullable().optional(),
}),
);
export type ReservationEndpointsInput = z.infer<typeof reservationEndpointsInputSchema>;
/**
* ONE SEGMENT of a multi-leg transport booking, as stored in
* `reservations.metadata.legs` (#1914).
*
* The model: the ordered `reservation_endpoints` rows are the geometry (which
* airports/stations, in which order), while the per-segment detail (each leg's
* own day + local time and its airline/train identity) lives in the legs. That
* split is why a stopover needs legs at all: its endpoint carries only the
* ONWARD departure time, so the arrival at the stop exists nowhere else.
*
* The shape mirrors 1:1 what the planner form (client TransportModal) and the
* importers (AirTrail, KItinerary) already write, including the `null`-for-unset
* convention on the day/time fields. One schema covers both kinds of booking:
* flights fill airline/flight_number, trains train_number/platform.
*
* `day_positions` is deliberately NOT part of the input: the day planner owns it
* per leg, so a writer preserves it rather than sets it.
*/
export const transportLegInputSchema = z.object({
/** Flights: IATA code of the departure airport. Trains: departure station label. */
from: z.string().min(1).nullable().optional(),
/** Flights: IATA code of the arrival airport. Trains: arrival station label. */
to: z.string().min(1).nullable().optional(),
airline: z.string().max(100).nullable().optional(),
flight_number: z.string().max(20).nullable().optional(),
train_number: z.string().max(20).nullable().optional(),
platform: z.string().max(20).nullable().optional(),
seat: z.string().max(20).nullable().optional(),
/**
* This segment's OWN booking reference (#1943). Airlines and railways often
* issue one per flight instead of one per booking, so the reservation-level
* `confirmation_number` cannot cover the chain. Unset means the booking's own
* reference is the one that counts for this segment; the column stays the
* reference for the booking as a whole and is never derived from a leg.
*/
confirmation_number: z.string().max(100).nullable().optional(),
dep_day_id: z.number().int().positive().nullable().optional(),
/** Local departure time of this segment, 'HH:mm'. */
dep_time: z.string().max(10).nullable().optional(),
arr_day_id: z.number().int().positive().nullable().optional(),
/** Local arrival time of this segment, 'HH:mm'. */
arr_time: z.string().max(10).nullable().optional(),
});
export type TransportLegInput = z.infer<typeof transportLegInputSchema>;
/**
* The legs of one booking. A leg list only makes sense with at least one
* stopover: a direct booking keeps the flat metadata it always had, and every
* reader (client flightLegs/dayMerge, calendar ICS, the AirTrail dedupe) treats
* `legs.length > 1` as the multi-segment marker.
*/
export const transportLegsInputSchema = z.array(transportLegInputSchema);
export type TransportLegsInput = z.infer<typeof transportLegsInputSchema>;
/**
* Reservation entity as returned by the reservation list endpoint
* (server/src/services/reservationService.ts -> listReservations). Columns of
* the `reservations` table plus the joined day_number / place_name / linked
* accommodation fields and the computed `day_positions` + `endpoints`.
* `accommodation_id` is stored as TEXT in the DB.
*/
export const reservationSchema = z.object({
id: z.number(),
trip_id: z.number(),
day_id: z.number().nullable().optional(),
end_day_id: z.number().nullable().optional(),
place_id: z.number().nullable().optional(),
assignment_id: z.number().nullable().optional(),
title: z.string(),
reservation_time: z.string().nullable().optional(),
reservation_end_time: z.string().nullable().optional(),
location: z.string().nullable().optional(),
confirmation_number: z.string().nullable().optional(),
notes: z.string().nullable().optional(),
url: z.string().nullable().optional(),
status: reservationStatusField,
type: reservationTypeField,
accommodation_id: z.union([z.number(), z.string()]).nullable().optional(),
metadata: z.string().nullable().optional(),
needs_review: z.number().optional(),
// 'live' for anything a person put on the trip, 'staged' while an automated
// ingest waits for confirmation. Only the anonymous exports filter on it.
ingest_state: z.string().optional(),
day_plan_position: z.number().nullable().optional(),
created_at: z.string().optional(),
// AirTrail (or future provider) linkage — drives the "synced" badge (#214).
external_source: z.string().nullable().optional(),
external_id: z.string().nullable().optional(),
external_owner_user_id: z.number().nullable().optional(),
external_synced_at: z.string().nullable().optional(),
sync_enabled: z.number().nullable().optional(),
// joined / computed in listReservations
day_number: z.number().nullable().optional(),
place_name: z.string().nullable().optional(),
accommodation_place_id: z.number().nullable().optional(),
accommodation_name: z.string().nullable().optional(),
accommodation_start_day_id: z.number().nullable().optional(),
accommodation_end_day_id: z.number().nullable().optional(),
day_positions: z.record(z.string(), z.number()).nullable().optional(),
endpoints: z.array(reservationEndpointSchema).optional(),
// Trip members / named guests this booking is for (#1517). Joined in listReservations.
travelers: z.array(reservationTravelerSchema).optional(),
});
export type Reservation = z.infer<typeof reservationSchema>;
/**
* Accommodation entity as returned by listAccommodations / getAccommodationWithPlace
* (server/src/services/dayService.ts). Columns of the day_accommodations table
* plus the joined place fields and (on list) the linked reservation_title.
*/
export const accommodationSchema = z.object({
id: z.number(),
trip_id: z.number(),
place_id: z.number().nullable().optional(),
start_day_id: z.number(),
end_day_id: z.number(),
check_in: z.string().nullable().optional(),
check_in_end: z.string().nullable().optional(),
check_out: z.string().nullable().optional(),
confirmation: z.string().nullable().optional(),
notes: z.string().nullable().optional(),
created_at: z.string().optional(),
// joined in listAccommodations / getAccommodationWithPlace
place_name: z.string().nullable().optional(),
place_address: z.string().nullable().optional(),
place_image: z.string().nullable().optional(),
place_lat: z.number().nullable().optional(),
place_lng: z.number().nullable().optional(),
reservation_title: z.string().nullable().optional(),
});
export type Accommodation = z.infer<typeof accommodationSchema>;
/**
* A booking link, as it is rendered into an href by the reservations panel. It
* deliberately stays free-form — people paste bare hosts ("www.hotel.com") and
* provider deep links, and rows written before this check exist — so the contract
* only rejects the schemes that execute in this origin instead of opening a page.
* Same class of bug as placeWebsiteSchema, one step narrower so nobody's stored
* link becomes unsavable.
*/
export const reservationUrlSchema = z.string().refine(
// Browsers strip control characters and whitespace before they resolve the
// scheme, so a tab spliced into 'javascript:' still runs. Everything at
// or below U+0020 goes, which is the same set a browser drops.
(v) =>
!/^(javascript|data|vbscript):/i.test(
Array.from(v)
.filter((c) => c > ' ')
.join(''),
),
{ message: 'must not be a javascript:, data: or vbscript: URL' },
);
/** Reservation create: title is required; the many optional fields stay open. */
export const reservationCreateRequestSchema = open.and(
z.object({
title: z.string().min(1),
url: reservationUrlSchema.nullable().optional(),
}),
);
export type ReservationCreateRequest = z.infer<typeof reservationCreateRequestSchema>;
export const reservationUpdateRequestSchema = open.and(
z.object({
url: reservationUrlSchema.nullable().optional(),
}),
);
export type ReservationUpdateRequest = z.infer<typeof reservationUpdateRequestSchema>;
/** Assign trip members/guests to a reservation (mirrors budget's PUT :id/members). */
export const reservationTravelersRequestSchema = z.object({
user_ids: z.array(z.number()),
});
export type ReservationTravelersRequest = z.infer<typeof reservationTravelersRequestSchema>;
export const reservationPositionsRequestSchema = z.object({
// day_plan_position is optional on the wire: the legacy route never
// validated position items, and an absent value binds NULL (clears the
// stored position) — pinned by the server integration suite (RESV-006).
positions: z.array(z.object({ id: z.number(), day_plan_position: z.number().optional() })),
day_id: z.union([z.number(), z.string()]).nullable().optional(),
});
export type ReservationPositionsRequest = z.infer<typeof reservationPositionsRequestSchema>;
export const accommodationCreateRequestSchema = z.object({
place_id: z.union([z.number(), z.string()]),
start_day_id: z.union([z.number(), z.string()]),
end_day_id: z.union([z.number(), z.string()]),
check_in: z.string().nullable().optional(),
check_in_end: z.string().nullable().optional(),
check_out: z.string().nullable().optional(),
confirmation: z.string().nullable().optional(),
notes: z.string().nullable().optional(),
});
export type AccommodationCreateRequest = z.infer<typeof accommodationCreateRequestSchema>;
/**
* REST body variant of the create contract: the three refs are required by the
* endpoint but optional on the wire — their absence answers with the bespoke
* controller 400 ('place_id, start_day_id, and end_day_id are required'),
* which the validation pipe must not pre-empt. The plugin RPC host keeps the
* strict schema above (missing refs are BAD_PARAMS there).
*/
export const accommodationCreateBodySchema = accommodationCreateRequestSchema.partial({
place_id: true,
start_day_id: true,
end_day_id: true,
});
export type AccommodationCreateBody = z.infer<typeof accommodationCreateBodySchema>;
export const accommodationUpdateRequestSchema = open;
export type AccommodationUpdateRequest = z.infer<typeof accommodationUpdateRequestSchema>;
// ---------------------------------------------------------------------------
// Booking import (KItinerary)
// ---------------------------------------------------------------------------
const bookingImportEndpointSchema = z.object({
role: z.enum(['from', 'to', 'stop']),
sequence: z.number(),
name: z.string(),
code: z.string().nullable(),
// Nullable: the mapper emits named endpoints without coords; confirm() geocodes
// them, and only the coord'd ones are persisted.
lat: z.number().nullable(),
lng: z.number().nullable(),
timezone: z.string().nullable(),
local_time: z.string().nullable(),
local_date: z.string().nullable(),
});
const bookingImportVenueSchema = z.object({
name: z.string(),
lat: z.number().optional(),
lng: z.number().optional(),
address: z.string().optional(),
website: z.string().optional(),
phone: z.string().optional(),
});
const bookingImportAccommodationSchema = z.object({
check_in: z.string().optional(),
check_out: z.string().optional(),
confirmation: z.string().optional(),
});
export const bookingImportPreviewItemSchema = z.object({
type: z.string(),
/**
* The extractor could not read a type and filled one in, so the import UI
* should offer the form the user was importing into rather than trusting the
* placeholder (#2076).
*/
type_guessed: z.boolean().optional(),
title: z.string().min(1),
reservation_time: z.string().nullable().optional(),
reservation_end_time: z.string().nullable().optional(),
confirmation_number: z.string().nullable().optional(),
location: z.string().nullable().optional(),
metadata: z.record(z.string(), z.unknown()).optional(),
endpoints: z.array(bookingImportEndpointSchema).optional(),
needs_review: z.boolean().optional(),
_venue: bookingImportVenueSchema.optional(),
_accommodation: bookingImportAccommodationSchema.optional(),
source: z.object({ fileName: z.string(), index: z.number() }),
});
export type BookingImportPreviewItem = z.infer<typeof bookingImportPreviewItemSchema>;
/**
* How the preview endpoint should treat the LLM fallback:
* - `no-ai` — kitinerary only (default; existing behaviour)
* - `fallback-on-empty` — run the LLM for files kitinerary returns nothing for
* - `force-ai` — run the LLM on every submitted file (skip kitinerary)
*/
export const bookingImportModeSchema = z.enum(['no-ai', 'fallback-on-empty', 'force-ai']);
export type BookingImportMode = z.infer<typeof bookingImportModeSchema>;
/** Per-file AI report so the preview UI can offer "Try AI parsing" only where it applies. */
export const bookingImportFileReportSchema = z.object({
fileName: z.string(),
aiAvailable: z.boolean(),
aiUsed: z.boolean(),
});
export type BookingImportFileReport = z.infer<typeof bookingImportFileReportSchema>;
export const bookingImportPreviewResponseSchema = z.object({
items: z.array(bookingImportPreviewItemSchema),
warnings: z.array(z.string()),
// Optional so existing/no-AI responses stay byte-compatible.
files: z.array(bookingImportFileReportSchema).optional(),
});
export type BookingImportPreviewResponse = z.infer<typeof bookingImportPreviewResponseSchema>;
export const bookingImportConfirmRequestSchema = z.object({
items: z.array(bookingImportPreviewItemSchema).min(1),
});
export type BookingImportConfirmRequest = z.infer<typeof bookingImportConfirmRequestSchema>;
export const bookingImportConfirmResponseSchema = z.object({
created: z.array(reservationSchema),
});
export type BookingImportConfirmResponse = z.infer<typeof bookingImportConfirmResponseSchema>;