# Changelog of Cruise Data Feed: Lines, Itineraries, Ports & Prices (`vulnv/cruise-data-feed`) Actor - **URL**: https://apify.com/vulnv/cruise-data-feed/changelog.md - **Full Actor documentation**: https://apify.com/vulnv/cruise-data-feed.md ## Changelog ### 1.4 — continued Shipped in place, per the one-version-overwritten-in-place decision recorded in 187c5e7. Deploy with `python apify/deploy_cruise_data_feed.py` — a plain `apify push` drops the Actor's API key secret. - **A multi-line request now returns every line you asked for.** The Actor issues one request per cruise line, and it used to page the first line until `maxResults` ran out — so a 3-line, 40-result run came back 40 rows from line one and nothing from lines two and three. The budget is now shared: each round gives every line an equal slice of what is left, so the same run returns 14 / 13 / 13. A line with fewer sailings than its share hands the remainder back to the others instead of losing it, so you still receive the full `maxResults` wherever the catalogue can supply it. - **Two new filters for data quality, both off by default.** `onlyWithBookingUrl` keeps only sailings carrying a direct booking link (26.0% of upcoming sailings); `scrapedSince` keeps only sailings last confirmed against their source on or after a date you choose, matched against each record's own `last_seen_at` (26.8% were last read before 2026-06-28). Both are applied by the feed itself, so neither shortens your results: `maxResults` still means the number of records you receive. Until now `minPrice` was the only input that happened to exclude stale rows, which was a coincidence rather than a contract. - **Every run reports what it returned.** The log now states how many stored records carry a booking link and how many were confirmed in the last 30 days, plus the per-cruise-line split, so a thin result is visible rather than inferred. - **The cruise-line dropdown offers all 70 bookable brands and stays that way.** It had drifted to 67, so Atlas Ocean Voyages, Aurora Expeditions, Paul Gauguin, Scenic Ocean and Tauck could not be selected at all, and two renamed entries ("Atlas Ocean Voyages Cruises", "Quark Expeditions Cruises") pointed at names the catalogue no longer uses. The list is now regenerated daily from the live catalogue by CI rather than by hand. Ship owners and browse categories are still deliberately excluded — nobody books under "Scylla Cruises" — but the field now says so and names the retail brand that carries their sailings. - **`sort` still defaults to `departure_date`,** and the README now says so and shows why: imminent departures are not staler than the catalogue average (25.0% vs 26.8%). Departure dates are heavily tied, and the feed now breaks those ties by freshness, so a short run leads with the most-recently-confirmed sailings rather than with whatever was inserted first — the first 25 rows went from 0% carrying a booking link and 96% stale to 40% and 0%. The README's new coverage table gives the real per-field numbers. - `cabinCategory` and `includePast` are passed through to the feed. ### 1.4 — includes a breaking output change Two fields the Actor added on top of the feed are **removed from every record**: `portCount` and `itineraryPortsText`. - **`portCount` was wrong, not merely redundant.** It was `len(itinerary)` — a count of itinerary ROWS — while the `port_count` sitting on the same record is the distinct ports the ship actually calls at, with days at sea and package flight/land/hotel days excluded. Two different numbers under two near-identical names: across 86,126 served records they agreed on only 4,122 of the 55,643 where `port_count` is populated. A sample record carried `port_count: 3` and `portCount: 5` at once, and `portCount` also reported a confident `0` for sailings that publish no itinerary at all, where `port_count` correctly says `null`. Anyone who summed, filtered or charted the wrong one got a wrong answer with nothing to warn them. **Use `port_count`.** If you specifically want the number of itinerary rows — sea days and travel days included — it is `len(itinerary)`, and it is yours to compute rather than ours to name. - **`itineraryPortsText`** was `" → ".join(...)` over `itinerary[].port`, which is in every record already. Rebuild it in one line if you display it. `price_amount`, `price_currency`, `price_per_night`, `fares[]` and every other field are unchanged. The dataset views were updated to match: **Overview** and **Itineraries** now show `port_count` ("Ports of Call") in place of the removed fields. #### Also in 1.4 - **Picking a cruise line now returns the whole line.** Each source spelled a line its own way, and the dropdown could only offer one spelling — so choosing Carnival matched just the sailings stored as "Carnival Cruise Line Cruises", 610 of 11,806, with no indication the rest existed. Line names are now canonicalized in the upstream feed and the dropdown lists the canonical names, so a line is one option covering all of its sailings. 18 of the 51 options were affected; 19 are renamed (e.g. "Royal Caribbean Cruises" -> "Royal Caribbean", "MSC Explora Journeys Cruises" -> "Explora Journeys"). Sub-brands stay separate: Viking's river/ocean/expedition fleets, Celebrity vs Celebrity River, and Explora vs MSC are each still their own option. ### 1.3 - Removed the "API base URL" input. The upstream data source is fixed; callers can no longer point the Actor at a different host (operators may still override it at deploy time via the `CRUISEFEED_API_BASE` env var). The optional `apiKey` input is unchanged. - Surfaced the enriched feed fields in the dataset views: Overview now shows taxes & fees, onboard credit and sold-out; Itineraries shows sea days. Records already carried the full set (`ship_code`, `destination_code`, `sea_days`, `port_count`, `taxes_and_fees`, `taxes_currency`, `obc_amount`, `sold_out`, `booking_url`, `detail_url`, `image_url`) via pass-through. - Every record is keyed by an opaque `id` (data providers are abstracted away); the `sources` input and per-source de-duplication were removed. ### 1.1 - Every record now includes the full per-cabin-class `fares` breakdown (interior/oceanview/balcony/suite) with availability, alongside the lead-in price. - Reads the versioned upstream API (`/v1`) and authenticates with `Authorization: Bearer`. ### 1.0 - Initial release. Download normalized cruise inventory from 60+ cruise lines (ships, sailing dates, durations, ports, day-by-day itineraries and pricing) via a single normalized cruise data API. Filter by line, source, region, port, ship, dates, duration, price and trip type.