TypeScript Examples

Complete, runnable TypeScript examples live in examples/typescript/. Each file mirrors its Python counterpart so you can compare the two APIs.

Running

cd examples/typescript
bun install                 # or: npm install

bun run basic_one_way_search.ts
# ...or use the package scripts:
bun run basic        # one-way
bun run round-trip   # round trip
bun run dates        # cheapest dates in a range
bun run multi-city   # 3-leg itinerary
bun run advanced     # alliances + exclusions + locale
bun run errors       # client tuning + typed error handling

On Node, run through a TypeScript loader: npx tsx basic_one_way_search.ts.

Files

File	What it shows
basic_one_way_search.ts	Minimal one-way search and result printing
round_trip_search.ts	Round trip; itineraries returned as [outbound, return]
date_range_search.ts	Cheapest dates across a flexible window
multi_city_search.ts	Three-leg multi-city itinerary
advanced_filters_search.ts	Alliances, airline exclusions, layover bounds, currency/locale
error_handling_with_retries.ts	Custom Client config and typed Search*Error handling

Multi-city

Not available through the current transport

Google loads multi-city results client-side through the RPC it gated in 2026-08, so the search page carries no rows to read. search(...) throws SearchUnsupportedError for TripType.MULTI_CITY rather than returning the first leg's one-way board, which would decode cleanly into wrong results. Search each leg separately for now. The example below is kept for when that changes.

import {
  Airport,
  type FlightResult,
  FlightSearchFilters,
  FlightSegment,
  SearchFlights,
  TripType,
} from "fli-js";

const inDays = (n: number) => new Date(Date.now() + n * 86_400_000).toISOString().slice(0, 10);

const leg = (from: Airport, to: Airport, date: string) =>
  new FlightSegment({
    departure_airport: [[[from, 0]]],
    arrival_airport: [[[to, 0]]],
    travel_date: date,
  });

const filters = new FlightSearchFilters({
  trip_type: TripType.MULTI_CITY,
  passenger_info: { adults: 1, children: 0, infants_in_seat: 0, infants_on_lap: 0 },
  flight_segments: [
    leg(Airport.JFK, Airport.LHR, inDays(30)),
    leg(Airport.LHR, Airport.CDG, inDays(34)),
    leg(Airport.CDG, Airport.JFK, inDays(38)),
  ],
});

const itineraries = (await new SearchFlights().search(filters, { topN: 3 })) as
  | FlightResult[][]
  | null;

for (const legs of itineraries ?? []) {
  const total = legs.reduce((sum, l) => sum + (l.price ?? 0), 0);
  console.log(`Itinerary total: $${total}`);
}

Error handling

import {
  SearchClientError,
  SearchConnectionError,
  SearchHTTPError,
  SearchParseError,
  SearchTimeoutError,
  SearchUnsupportedError,
} from "fli-js";

try {
  await new SearchFlights().search(filters);
} catch (err) {
  if (err instanceof SearchTimeoutError) {
    // raise the client timeout and retry
  } else if (err instanceof SearchConnectionError) {
    // network/proxy problem
  } else if (err instanceof SearchHTTPError) {
    // non-2xx from Google
  } else if (err instanceof SearchParseError) {
    // the page loaded but carried no readable results — a consent or
    // block page, or a change in Google's page shape
  } else if (err instanceof SearchUnsupportedError) {
    // the current transport cannot serve this query (multi-city)
  } else {
    throw err;
  }
}

All of the above extend SearchClientError, so catch (err) { if (err instanceof SearchClientError) … } covers "the search failed" in one branch. A search that simply found nothing returns null instead of throwing.

See the Python examples for the equivalent scripts.