Bilibili Scraper | 哔哩哔哩数据采集

Turn Bilibili activity into usable data

Find videos, research creators, read audience comments, and measure danmaku activity from Bilibili (哔哩哔哩). Enter a keyword, BV ID, or creator MID and run the Actor. Results go to an Apify dataset that you can inspect or export. No Bilibili account, browser, proxy configuration, or third-party API key is required.

Use it to follow a creator's uploads, discover videos about a topic, compare engagement, or collect comments for audience research. The Actor also supports repeat runs that return only newly seen search or popular videos.

What can you collect?

Mode	Give it	Get back	Useful for
search	A keyword and optional related terms	Video titles, links, creators, dates, engagement counts	Content discovery and campaign research
video_detail	BV IDs or video URLs	Detailed video records, with optional comments and danmaku profiles	Video and campaign analysis
video_comments	BV IDs or video URLs	Comments and optional thread replies	Audience feedback research
user_videos	Creator MIDs or profile URLs	A creator profile and discovered uploads	Creator monitoring and benchmarking
user_profile	Creator MIDs or profile URLs	Followers, following, bio, level, verification, video count	Creator shortlists
popular	An optional category	Popular videos and their engagement counts	Trend spotting

includeComments and includeDanmaku can enrich video results. Each row has a type value (video, user, comment, or danmaku) so you can separate the records after export. maxResults limits primary video or profile results; maxComments limits comments per video.

Examples you can use now

Copy an input into the Actor's JSON input editor, or choose the same fields in the form. These examples use real Bilibili identifiers; results and engagement counts change over time.

Discover AI tutorial videos across related searches

Find videos mentioning Chinese AI tutorial terms. The Actor deduplicates BV IDs found by more than one search term.

{
  "mode": "search",
  "searchQuery": "人工智能教程",
  "searchAliases": ["AI教程", "机器学习入门"],
  "sortOrder": "pubdate",
  "maxResults": 50
}

For larger historical research jobs, set maxResults above 1,000. The Actor splits search into publication-date windows to get past Bilibili's roughly 1,000-result limit per window. Very busy single days can still be capped.

Follow a creator's uploads

Collect a profile and up to 30 discovered videos from creator MID 546195 (老番茄). Each upload includes discoverySource, showing whether it came from the archive, creator-space listing, or search fallback.

{
  "mode": "user_videos",
  "userIds": ["546195"],
  "maxResults": 30
}

If you only need creator facts for a shortlist, switch to user_profile and pass several MIDs in userIds.

Compare video engagement and danmaku activity

Get video metrics, up to 20 comments, and a danmaku activity profile for a real Bilibili video. The danmaku row includes the fetched count, unique sender count, rate, peak minute, and a small timeline sample.

{
  "mode": "video_detail",
  "videoUrls": ["https://www.bilibili.com/video/BV1VCHY6BEcn"],
  "maxResults": 1,
  "includeComments": true,
  "maxComments": 20,
  "includeDanmaku": true
}

Read a video's comment threads

Use the dedicated comments mode when you want feedback rather than a video record. includeReplies brings in thread replies within the per-video comment limit.

{
  "mode": "video_comments",
  "videoUrls": ["BV1VCHY6BEcn"],
  "maxComments": 100,
  "sortComments": "time",
  "includeReplies": true
}

Watch technology trends without repeating old results

Run this input regularly to collect newly seen popular technology videos. deltaMode keeps the seen BV IDs in a named Apify key-value store for this Actor, so later runs with the same settings return new matches.

{
  "mode": "popular",
  "category": "tech",
  "maxResults": 30,
  "deltaMode": true,
  "deltaStateKey": "tech-trends"
}

What does the output look like?

Row type	Selected fields	What it tells you
video	bvid, title, url, authorName, authorMid, publishDate, viewCount, likeCount, danmakuCount	Video identity, creator, timing, and engagement
user	mid, name, profileUrl, followers, following, videoCount, totalLikes, sign	Creator profile and reach
comment	commentId, text, authorName, likeCount, createdAt, videoBvid, rootRpid	Audience feedback and thread context
danmaku	videoBvid, danmakuFetched, danmakuUniqueSenders, danmakuPeakMinute, danmakuSample	Aggregate activity across the fetched danmaku pool

For example, a successful run on 3 October 2026 returned these selected fields for the video above. Counts are a snapshot, not live values:

[
  {
    "type": "video",
    "bvid": "BV1VCHY6BEcn",
    "title": "我记得课本里的新疆，不是这样的啊？？",
    "viewCount": 483992,
    "likeCount": 25285
  },
  {
    "type": "danmaku",
    "videoBvid": "BV1VCHY6BEcn",
    "danmakuFetched": 1068,
    "danmakuUniqueSenders": 766,
    "danmakuPeakMinute": 0
  }
]

Open the run's Output tab to inspect the dataset, or export it as JSON, CSV, or Excel. The dataset also has an output schema for integrations and a table view for quick inspection.

How much does it cost?

Charge	Price	When it applies
Bilibili record	$0.018 per dataset row ($18 per 1,000)	Each video, creator profile, comment, or danmaku profile row saved
Actor start	$0.00005	When a run starts at the default 1 GB memory
Platform usage	Included	No separate compute or proxy charge from this Actor

For example, 100 video rows cost $1.80 plus the start charge. Adding one danmaku profile adds one row at the same $0.018 rate. You only pay the row charge for records the Actor saves to the dataset.

How does the Actor handle Bilibili limits?

Situation	Behavior
A search window reaches roughly 1,000 results	For deep searches, split it into smaller publication-date windows and deduplicate BV IDs.
A creator upload endpoint blocks pagination	Try the signed creator-space listing, then exact-MID search results. discoverySource identifies each recovered video's path.
A video has many comments	Stop at maxComments per video; optionally include thread replies.
A video has a large danmaku history	Return a profile of the fetched pool. Danmaku enrichment is limited to 200 videos per run.
A repeated search or popular run sees old videos	With deltaMode, omit BV IDs recorded by earlier runs with the same settings.

Creator upload listings can be restricted by Bilibili. A requested count is a maximum, and a run may return fewer uploads than requested. Search fallback is exact-MID filtered, but it is not a guarantee of complete upload history. The run log reports listing failures and shortfalls. The danmaku row summarizes the fetched pool; it does not contain every individual danmaku message.

Questions

Do I need a Bilibili login or API key? No. The Actor reads public responses over HTTP.

Can I monitor more than one topic? Yes. Create a separate saved task for each topic or category. For repeat runs, use a distinct deltaStateKey where you want separate monitoring histories.

Can I use the results in another tool? Yes. Export the dataset or read it through Apify's dataset API. Every row includes type and scrapedAt for downstream filtering.

Why did a run return fewer records than requested? Bilibili may return fewer matching results, cap a search window, or restrict a creator listing. Check the run log for coverage notes and endpoint errors; delivered rows remain in the dataset.