This changelog highlights releases that materially changed Airbnb Reviews Scraper for users. Dates use UTC; small internal refactors and routine maintenance are intentionally omitted.
Reliability
- Numeric listing IDs and remote URL lists (
requestsFromUrl) no longer crash the run. Long IDs sent as JSON numbers are rejected with a clear message, because they would otherwise lose digits and name another listing.
- A removed or unknown listing is reported as not found right away instead of retrying and failing with a misleading error. A run with no usable listing finishes with an empty dataset, no charge, a clear status and
RUN_DIAGNOSTICS in the key-value store.
- If Airbnb fails partway through a listing, the reviews already collected are saved and charged instead of being discarded. Reviews are saved page by page, so a run that stops before its timeout or at the Maximum charge per run keeps what it fetched and stops fetching.
- The run-start event is now charged with the first saved review, so a run that saves no reviews costs nothing.
Reliability
- Airbnb request failures now use bounded retries and fresh proxy sessions instead of silently producing successful empty results. Multi-listing runs preserve successful reviews plus uncharged error records, while runs where every listing fails now finish with a failed status.
Changes
- Review scraping now records per-review billing events reliably for every saved review item.
Features
- Added public task examples for common Airbnb review scraping workflows.
Features
- Initial release for extracting Airbnb listing reviews from listing URLs or IDs, including review text, ratings, reviewer details, host replies, dates, and media when available.
- Added review sorting, topic filters, full-text search, and optional AI sentiment or translation fields for review analysis workflows.
