Bilibili Scraper — Search, Videos, Creator Uploads, Danmaku, Transcripts

Scrape public data from Bilibili (B站), China's largest video community, without login or cookies. Seven modes:

Mode	Input	Output
search	keywords	videos matching each keyword (up to 1,000 per keyword, Bilibili's own cap)
video_detail	video URLs / BV IDs / av IDs	full stats: views, likes, coins, favorites, shares, comments, danmaku count, tags
user_videos	creator IDs (mid) or space URLs	the complete upload list of each creator, newest first
danmaku	video URLs / BV IDs	full danmaku (bullet comment) text with video timestamp, mode, color, send time; all parts of multi-part videos
popular	—	the current popular feed
ranking	optional category ID	the top 100 ranking
transcript	video URLs / BV IDs	speech-to-text transcript in Simplified Chinese with per-segment timestamps, for every part of the video

Why this Actor

• Complete creator uploads. Paginates the creator's own upload list instead of searching by name, so you get every public video, not just what search surfaces.
• Raw danmaku text, timestamped to the second of the video: ready for sentiment analysis, highlight detection, or LLM pipelines.
• Flat, typed records (ISO dates, seconds, numbers) that drop straight into spreadsheets, BI tools or AI agents.
• No login, no cookies. Public data only.

Output example (video)

{
  "bvid": "BV1GJ411x7h7",
  "url": "https://www.bilibili.com/video/BV1GJ411x7h7",
  "title": "【官方 MV】Never Gonna Give You Up - Rick Astley",
  "durationSec": 213,
  "publishedAt": "2019-12-31T23:43:23.000Z",
  "ownerMid": 486906719,
  "ownerName": "索尼音乐中国",
  "views": 107488601, "likes": 2958870, "coins": 1285522, "favorites": 1556417,
  "shares": 500693, "comments": 238211, "danmakuCount": 151037,
  "tags": ["Rick Astley", "MV", "欧美音乐"],
  "source": "video_detail"
}

Output example (danmaku)

{ "bvid": "BV1GJ411x7h7", "part": 1, "timeSec": 12.5, "text": "前方高能", "mode": "scroll",
  "color": "#ffffff", "sentAt": "2021-02-19T11:32:56.000Z" }

Transcripts

Bilibili only shows subtitles to logged-in users, and most videos have none. This mode downloads the public audio track and runs speech recognition (Whisper large-v3 turbo on Groq), then converts the output to Simplified Chinese with full-width punctuation.

• Accuracy is high for clear narration (documentaries, tutorials, talks) and lower for strong accents, fast banter or loud background music. Proper nouns and brand names are the most common errors.
• Billed per started minute of audio. Parts longer than Max minutes per video part are skipped and not billed.
• Bring your own Groq API key for a lower per-minute price.

{ "bvid": "BV1pGnn67EbB", "part": 1, "durationSec": 2682, "billedMinutes": 45, "language": "zh",
  "text": "Sabrina是一位在OnlyFans上大约前百分之五的创作者。……",
  "segments": [{ "start": 58.0, "end": 66.1, "text": "Sabrina是一位……" }] }

How to use

1. Pick a Mode and fill the matching input: keywords for search; video URLs or BV IDs for video_detail, danmaku and transcript; creator IDs or space URLs for user_videos.
2. Set Max videos per keyword / creator / list and click Start.
3. Open the Output tab and export to CSV, Excel or JSON, or read the dataset through the API.

Call it from code with the Apify API client:

from apify_client import ApifyClient

client = ApifyClient("<YOUR_APIFY_TOKEN>")
run = client.actor("dimpled_academy_qp4/bilibili-scraper").call(run_input={
    "mode": "user_videos", "userIds": ["https://space.bilibili.com/946974"], "maxResults": 50,
})
for video in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(video["publishedAt"][:10], video["views"], video["title"])

Recipes: find creators in a niche (search, then sort by views); track a competitor's channel weekly (user_videos plus an Apify schedule); find highlight moments (danmaku, count per 30-second window); feed videos to an LLM (transcript).

中文快速上手

本工具无需登录即可采集 B 站公开数据：关键词搜索、视频详情、UP 主全部投稿、弹幕、热门、排行榜，以及语音转文字（输出简体中文文字稿）。

使用方法：在 Mode 中选择模式并填写对应输入，即关键词、视频链接或 BV 号、UP 主 ID 或空间链接；设定数量上限后点击 Start；运行完成后，在 Output 标签页导出 CSV、Excel 或 JSON。BV 号位于视频链接中，UP 主 ID 即空间链接 space.bilibili.com/ 后的数字。

费用按成功写入的记录计算，失败与跳过的条目不收费，详见下方 Pricing。

Notes and limits

• search and user_videos results come from list endpoints that lack coins, shares and full tags. Turn on Fetch full details to fill them (2 extra requests per video, slower).
• Video category names are no longer returned by Bilibili's detail endpoint; categoryId is provided.
• Some videos are restricted to mainland China (for example, licensed music videos). Outside China Bilibili answers "not found" (-404) for them; they are skipped with a warning.
• Full comment threads and official subtitles require a logged-in account on Bilibili and are not scraped; use the transcript mode to get the spoken text instead.
• Danmaku: Bilibili serves a capped pool per video part (often a few thousand items). If the pool is larger than Max danmaku per video, an even sample across the whole video timeline is returned.
• Bilibili throttles by IP. The Actor retries with a fresh visitor identity and, if you enable a proxy, a fresh proxy session. If a page still fails, it is skipped and listed in the key-value store (FAILED_PAGES_<mid>) instead of failing the run.
• Personal data: the Actor outputs creator names and IDs (public channel info) and never outputs danmaku sender identifiers. You are responsible for using the data lawfully.

Pricing

Pay per result. You pay only for records written to the dataset; skipped or failed items are free.

Event	Free / Bronze	Silver	Gold and above
Video record (any mode except danmaku)	$0.004	$0.0035	$0.003
Danmaku record	$0.0001	$0.0001	$0.00008
Transcript, per started audio minute	$0.005	$0.0045	$0.004
Transcript with your own Groq key, per minute	$0.002	$0.002	$0.0015

Plus Apify's standard Actor start fee ($0.00005 per run). Set Max cost per run to cap spending; the Actor stops cleanly when the limit is reached.