# Changelog of Realtor.ca Scraper (`haketa/realtor-ca-scraper`) Actor

- **URL**: https://apify.com/haketa/realtor-ca-scraper/changelog.md
- **Full Actor documentation**: https://apify.com/haketa/realtor-ca-scraper.md

## Changelog

### 1.1.0 — 2026-11-02

- **Cloudflare resilience (transport swap).** Realtor.ca migrated its anti-bot from Imperva Incapsula to **Cloudflare bot-management**. Cloudflare 403-blocks the previous Chrome / `got-scraping` TLS fingerprint on both `www.realtor.ca` and the `api2.realtor.ca` API host — including the actor's HTML modes. Swapped the HTTP transport to **`impit` with a genuine Firefox fingerprint**, which clears Cloudflare on the `www.realtor.ca` host and restores the `search-area`, `agent-profile`, and `office-profile` modes (verified HTTP 200 with the embedded `PropertySearch_Post` payload intact, where the Chrome fingerprint returned the "you have been blocked" page).
- **Removed Chrome-only client-hint headers** (`sec-ch-ua`, `sec-ch-ua-mobile`) and the hardcoded Chrome `User-Agent`. The `impit` client now supplies a self-consistent Firefox UA / TLS / Accept-Encoding; sending Chrome hints alongside a Firefox fingerprint was an instant Cloudflare bot tell.
- **Build hardened.** Dropped `--no-optional` from the Dockerfile `npm install` (it would strip `impit`'s platform-specific native binary and break the actor at import time) and switched to `npm install --omit=dev`.
- **Verification.** Fresh live run against 12 Canadian metros on CA residential proxy. The `api2.realtor.ca` API host is currently behind a hard Cloudflare challenge (0/10 fresh residential IPs cleared it this run), so the API-driven cities/map-bbox/detail flow returned 0 records and exits cleanly with a re-run status message. The `www.realtor.ca` HTML surface (area pages, agent & office profiles) was re-verified as reachable (HTTP 200) with the new Firefox fingerprint.
- **Docs.** README corrected throughout from "Imperva Incapsula" to **Cloudflare**, engine updated to `impit` (Firefox) + Cheerio, and pagination / FAQ sections now give honest re-run guidance for Cloudflare-challenged API runs. No input or output field was renamed; the output schema is unchanged.

### 1.0.0 — 2026-05

- Initial public release.
- Five scrape modes: `search-area` (Realtor.ca area-page links), `map-bbox` (map bounding box), `single-listing` (MLS numbers), `agent-profile`, and `office-profile`.
- City-dropdown input (27 Canadian markets across all provinces) that auto-builds map bounding boxes — no URLs or codes required.
- Deep normalized output schema: price (CAD), Canadian `X + Y` bed/bath decoding, ownership type, annual taxes + tax year, monthly condo fees, heating fuel/type, cooling, basement, parking, lot size, year built, GPS coordinates, photos, public remarks, EN/FR language detection, agent contact, and brokerage/franchise data.
- Bounding-box grid auto-split to stay under Realtor.ca's 200-results-per-query cap; de-duplication on MLS number / listing ID.
- Sticky-session Canadian residential proxy with fresh-IP rotation on blocks; soft-deadline shutdown so runs always finish cleanly.
- Pay-per-event pricing.
