# Changelog of Bilibili Scraper — Video Search & Video Details (No Login) (`cemented_jest/bilibili-scraper`) Actor

- **URL**: https://apify.com/cemented_jest/bilibili-scraper/changelog.md
- **Full Actor documentation**: https://apify.com/cemented_jest/bilibili-scraper.md

## Changelog

### 1.0 - 2026-11-04

First release (not yet published; see `docs/store-listing.md` for the publish steps).

- `search` mode: keywords in, videos out. Plain-HTTP search API first (`wbi/search/type`, no signature, no cookies), server-rendered search page second,
  Playwright browser third. Orders: `relevance`, `pubdate`, `click`, `dm`, `stow`. Up to 1,000 per keyword (the site exposes about 1,000 results per keyword).
- `video` mode: BV ids, av ids and video URLs in, full metadata out (including coin and share counts and the complete tag list).
  Plain-HTTP video document first, Playwright browser second.
- Output: 24 documented fields per item, exact types, `null` when unavailable; items missing a critical field are never pushed or charged.
- Pay per event: `video-result` (search) and `video-detail` (video).
- `RUN_SUMMARY` record with per-keyword window information (`numResults`, `numPages`, `windowLimited`), the acquisition path used for every page, raw-key hashes for
  drift detection and the list of failures with reasons.
- Block handling: HTTP 412 / 403 / 429, JSON -352 / -403 / -412 / -799 and the soft block (HTTP 200, `code 0`, no results, only a risk-control voucher) are treated as blocks: retry on a fresh proxy session, then the next path.
  A half-open circuit breaker skips a path after 3 failed pages and probes it again after 60 s; the browser phase stops after 5 consecutive failures.
- Default proxy: Apify residential, US exit (datacenter IPs are refused by video pages with HTTP 412).
- Not in 1.0: comments (anonymous access is capped by the site), uploader pages, danmaku, media downloads.