# Changelog of YouTube Comments Scraper — Author, Likes, Replies & Text (`haketa/youtube-comments-scraper`) Actor

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

## Changelog

### 2026-11-01

- **Fixed a blocking regression where runs on bare datacenter IPs returned 0 comments.** YouTube was serving a degraded watch page (video metadata present, but the comment continuation token stripped) to single un‑rotated IPs, so every video fell through to a `"no comments"` row. Apify Proxy (datacenter) is now **on by default** and the actor also defaults to proxy rotation when `proxyConfiguration` is omitted via the API. Verified: the same input went from **0 → 600 comments**.
- **Fixed `likeCountText` showing a blank space for zero‑like comments.** YouTube returns a literal `" "` for comments with no likes; it is now normalized to `"0"` to match the documented field contract.
- **Fresh verification run:** 600 unique comments across 3 popular videos in 17s, 0 duplicates on `commentId`, and 100% coverage on all core fields (`commentId`, `author`, `videoId`, `likeCount`, `replyCount`, `publishedTime`; `text` 99.8% — the single blank is a genuine emoji‑only comment). No value bugs, all numeric fields clean.
- **README refreshed to the gold standard:** output example and field table regrounded in the fresh sample, proxy‑by‑default behavior documented across Input/Pagination/Tips/FAQ, and benchmark numbers updated.
