Bilibili Scraper

Scrape Bilibili (哔哩哔哩 / B站) by keyword search, creator uploads, video comments, and danmaku (bullet chat). Returns clean, structured data for each record: titles, play / like / danmaku / favorite / comment counts, author info, cover image, video URL, comment text, and danmaku content with timing.

Unofficial. This Actor is not affiliated with, authorized, or endorsed by Bilibili (Shanghai Kuanyu / 上海宽娱数码科技). It is an independent tool that retrieves publicly available data via a third-party API. Use it in compliance with Bilibili's terms and all applicable laws; you are responsible for how you use the retrieved data.

What it does

Four modes run in order, all sharing one maxItems budget:

• Keyword search — give one or more keywords; the Actor paginates Bilibili comprehensive search results for each. Sort by comprehensive / latest / most played / most danmaku / most favorited.
• Creator uploads — give one or more creator UIDs (the numeric mid) to scrape the videos they have uploaded.
• Video comments — give one or more video BV IDs to scrape their top-level comments.
• Danmaku — give one or more video cids to scrape the bullet-chat overlay messages.

Input

Field	Type	Default	Description
searchKeywords	string[]	["编程"]	Keywords to search. Each is paginated independently. Chinese keywords return the most results.
searchOrder	enum	comprehensive	comprehensive / pubdate (latest) / click (most played) / dm (most danmaku) / stow (most favorited).
userIds	string[]	[]	Creator UIDs (numeric mid) whose uploaded videos to scrape.
bvIds	string[]	[]	Video BV IDs whose top-level comments to scrape (e.g. BV1rpWjevEip).
cids	string[]	[]	Video segment cids whose danmaku to scrape.
maxItems	integer	50	Max total items across all targets.

Example input

{
  "searchKeywords": ["编程", "python"],
  "searchOrder": "click",
  "userIds": ["3546597933714079"],
  "bvIds": ["BV1rpWjevEip"],
  "maxItems": 200
}

Output

One dataset item per record. Videos:

{
  "id": "BV1rpWjevEip",
  "bvid": "BV1rpWjevEip",
  "aid": 113006243481679,
  "type": "video",
  "title": "目前B站最全最细的Python零基础全套教程",
  "url": "https://www.bilibili.com/video/BV1rpWjevEip",
  "coverUrl": "https://i2.hdslb.com/bfs/archive/...jpg",
  "duration": "2398:14",
  "playCount": 16634964,
  "danmakuCount": 111398,
  "likeCount": 410099,
  "favoriteCount": 653700,
  "author": { "uid": 3546597933714079, "name": "Python官方课程" },
  "categoryName": "计算机技术",
  "pubdate": 1724338758,
  "source": "search:编程"
}

Comments (type: "comment") carry content, likeCount, replyCount, author, ctime. Danmaku (type: "danmaku") carry content, videoTime (seconds into the video), sendTime, color, mode, cid.

Notes

• Data is sourced live; Bilibili occasionally rate-limits, so the Actor retries transient blocks with exponential backoff.
• IDs are de-duplicated within a run.
• A video's cid (needed for danmaku) is part of its video detail / parts data.