Bilibili Scraper - Videos, Comments, Danmaku & Creator Uploads

What does Bilibili Scraper do?

Bilibili Scraper extracts public data from Bilibili (哔哩哔哩, B站), China's largest video community, and turns keywords, video links, creator pages and the popular lists into structured rows: each video with views, likes, coins, favorites, shares, comments, danmaku count, tags, category, duration and publish date; comments and replies; danmaku (bullet comments) with the second they appear; creator (UP主) profiles; live rooms and the hot-search board. Built for China-market researchers, brand and KOL teams, agencies, AI-training and data pipelines that need Bilibili numbers without an account.

Bilibili Scraper - B站 Scraper - 哔哩哔哩 Scraper - Bilibili Comments Scraper - Bilibili Danmaku Scraper - Bilibili API

• Ten modes in one actor: keyword search, videos by link, comments of links, danmaku of links, a creator's uploads, creator profiles, creator search, popular / ranking / must-watch lists, hot search and live-room search. type tells the rows apart; each kind has its own price and you pay only for the kinds you switch on.
• Past Bilibili's 1,000-results-per-search cap: when a keyword has more, the run walks it by date window (measured: 1,200 unique videos for "原神" in 24 s).
• A creator's whole upload list: 935 videos back to 2015 for 影视飓风 (Bilibili's own counter shows 938), 99 per request, newest first.
• Comments as deep as you ask: 20 per page, hot or newest first, with replies, pinned comments and whether the uploader liked or replied.
• Danmaku from two sources merged: the video's danmaku pool plus its 6-minute timeline segments, deduplicated: 9,560 danmaku for one 45-minute video where the pool alone gives 7,200. Packed as one item per video by default, or one row per danmaku.
• Danmaku reaction profile: unique senders, danmaku per minute, the peak minute and an evenly spread sample, in one row per video.
• No login, cookies or API key. Paste keywords, links or creator ids and run.
• Monitor mode for schedules: each run returns only videos, comments, danmaku, creators, live rooms and topics no earlier run delivered.
• Export anywhere: JSON, CSV, Excel, XML; schedules, webhooks, Make, Zapier, n8n, Google Sheets; Apify API and MCP, with compact output for AI agents.

What data can you get from Bilibili?

🎬 Video id, link, title (`bvid`, `url`)	👁️ Views (`viewCount`)	👍 Likes, 🪙 coins, ⭐ favorites, ↗️ shares
💬 Comment count (`replyCount`)	🔤 Danmaku count (`danmakuCount`)	⏱️ Duration, publish date
🏷️ Tags, category (`category`, `categoryId`)	📺 Parts (P1, P2…), resolution, original / repost	🏆 Honors and highest site rank (`honors`, `hisRank`)
📈 Engagement rate, like rate	👥 Watching now (`onlineCount`)	🔗 Related videos, stream links
🗨️ Comments + replies: text, likes, time, level	💥 Danmaku: text, second, colour, mode, sent time	📊 Danmaku profile: senders, per minute, peak
🧑‍🎨 Creator: followers, videos, total likes, level	✅ Verification, VIP, live room	📧 Email / WeChat / QQ from the public bio
🔴 Live rooms: title, watching, area, streamer	🔥 Hot search: keyword, heat score	🙂 Sentiment of comments and titles (optional)

On a 100-video search run with full details, views, likes, coins, favorites, shares, tags, category, parts and resolution were filled on 100% of rows.

How to scrape Bilibili with Bilibili Scraper

1. Create a free Apify account — $5 of monthly credit, no card needed.
2. Open Bilibili Scraper and click Try for free.
3. Pick a mode under What to scrape, then type a keyword, paste video links or paste creator ids.
4. Switch on any extras (comments, danmaku, full details) and set Max results.
5. Click Start, then open Storage and Export as JSON, CSV or Excel — or let a schedule run it for you.

⬇️ Input

Parameter	Type	Description
🔍 What to scrape
mode	Select	search · video_detail · video_comments · video_danmaku · user_videos · user_profile · user_search · popular · hot_search · live_search. Default search
searchQuery / searchQueries	Text / list	Keywords, Chinese or English, e.g. 原神, 机械键盘, AI tutorial
searchAliases	List	Other spellings searched with each query and merged into it, e.g. 耐克 for Nike
videoUrls	List	bilibili.com/video/BV…, BV1… ids, av… ids, b23.tv share links; ?p=2 picks a part
userIds	List	Creator ids (the number in space.bilibili.com/<mid>) or those links
popularSource, category	Select	Popular feed, ranking (by 15 categories or all) or the must-watch list
🎯 Filters (applied by Bilibili's search)
sortOrder	Select	Relevance, most views, newest, most danmaku, most favorites, most comments
durationFilter	Select	Any, under 10 min, 10-30, 30-60, over 60
pubtimeBegin, pubtimeEnd	Text	Published window, e.g. 2026-09-01; also filters uploads and popular rows
searchCategoryId	Integer	Bilibili category id, e.g. 4 = games, 188 = tech
sortComments	Select	Hot, newest, or most liked
✂️ Refine results (before anything is charged)
minViews, minLikes	Integer	Skip videos under these counts
minDurationSec, maxDurationSec	Integer	Length bounds in seconds
publishedWithinDays	Integer	Only videos from the last N days
includeKeywords, excludeKeywords	List	Words that must / must not appear in the title, description or tags
minFollowers, onlyWithContacts	Integer / Boolean	Creator rows: follower floor; only creators whose bio lists an email, WeChat, QQ or phone
➕ Extra data
includeDetails	Boolean	Full page data for search, uploads and popular rows (coins, shares, tags, parts, honors). Free
includeRelated, includeCreatorProfile, includeOnlineCount, includeStreamUrls	Boolean	Related videos, uploader stats, watching-now, DASH stream links. Free
includeComments, maxComments, includeReplies, maxRepliesPerComment		Comments and replies as their own rows (charged per comment)
includeDanmaku, danmakuOutput, maxDanmakuPerVideo		Danmaku, packed per video (default) or one row each (charged per danmaku)
danmakuAllParts, danmakuStartSec, danmakuEndSec		Every part of a multi-part video; a time window inside the video
includeDanmakuProfile, danmakuSampleSize		One reaction-profile row per video (charged per profile)
sentimentAnalysis, anonymizeUsers	Boolean	Lexicon sentiment on comments and titles; hashed commenter names
🔁 Monitoring
monitorMode, monitorStoreName	Boolean / Text	Only rows no earlier run with the same store delivered. Pair with a schedule
followUpComments	Boolean	With monitor + comments: also new comments under videos delivered before (up to 200)
📦 Output
maxResults	Integer	Default 100; 0 = no limit. Videos (or creators, live rooms, topics) per run
maxResultsPerSearch	Integer	Cap per keyword or creator
compact	Boolean	Core fields only, for AI agents and MCP
⚙️ Advanced
proxyConfiguration, maxConcurrency		Requests go direct first; the proxy is used only when Bilibili refuses the container's IP

Input JSON

{
  "mode": "search",
  "searchQuery": "机械键盘",
  "sortOrder": "pubdate",
  "maxResults": 100,
  "includeDetails": true,
  "includeComments": true,
  "maxComments": 20
}

Pick the mode before anything else

Each mode reads its own box: search, user_search and live_search read the keywords; the three video_* modes read videoUrls; user_videos and user_profile read userIds. Do: {"mode": "user_videos", "userIds": ["946974"]} · Don't: {"mode": "search", "userIds": ["946974"]} (search has no keyword, so the sample search runs).

Use a creator id, not a name

userIds takes the number from the creator's space link (space.bilibili.com/546195 → 546195). To find creators by name, use user_search with the name as the keyword.

Want only coins and shares? Switch on full details

Search results do not carry coins or shares (Bilibili's search API leaves them out), so those fields are empty on plain search rows. includeDetails: true reads each video's page data and fills them, at the same price per video.

⬆️ Output

Every record becomes one dataset item. These tables are real rows from platform runs:

A video row (shortened):

{
  "type": "video",
  "bvid": "BV1EJpg6SEVV",
  "url": "https://www.bilibili.com/video/BV1EJpg6SEVV",
  "title": "【2026年10月键盘推荐】自费万元！机械+磁轴键盘全配列实拍实测…",
  "duration": 1037,
  "viewCount": 584578,
  "likeCount": 11340,
  "coinCount": 4297,
  "favoriteCount": 5697,
  "shareCount": 229,
  "replyCount": 498,
  "danmakuCount": 177,
  "authorName": "BR店长",
  "authorMid": 96153951,
  "publishDate": "2026-10-12T02:00:00.000Z",
  "category": "数码",
  "tags": ["键盘", "评测", "机械键盘", "客制化键盘"],
  "engagementRate": 0.037738,
  "source": "search"
}

A packed danmaku item holds a video's danmaku in danmaku, one object per danmaku: danmakuId, text, videoTime (second), videoTimeFormatted, mode, fontSize, color, sentAt, pool, senderHash, weight. With danmakuOutput: "rows" each of those is its own type: "danmaku" row.

Fields and how often they were filled

Rows	Field fill on platform runs
Search, no details (20 rows)	bvid, title, views, likes, favorites, comments, danmaku, duration, publish date, author, category, tags 100%; coins, shares, cid empty (not in search results)
Search + full details (100 rows)	the above plus coins, shares, cid, parts, resolution, original/repost 100%; description 63%; honors 3%
Creator uploads (935 rows)	views, likes, coins, favorites, shares, comments, danmaku, duration, description, cid 99-100%; tags only with full details
Comments + replies (386 rows)	text, likes, time, level, VIP, uploader-liked 100%; gender 44%; bio 58%; emotes 28%; mentions 23%; pictures 7%; replies carry rootCommentId
Danmaku (9,560 in a pack)	id, text, second, mode, colour, font size, sent time, sender hash, weight 100%
Creator search (5 rows)	name, followers, videos, level, verification, recent videos 100%; bio 80%; live room 60%
Live rooms / hot search	room, title, watching, area, streamer 100% / keyword, heat score 100%

How can I use Bilibili data?

• 🎯 Brand and KOL teams — find creators by keyword, read their followers, total likes and the business contact in their bio, then track their uploads.
• 📊 Market researchers — follow a category's ranking or a product keyword week over week (views, coins, favorites as purchase-intent signals).
• 💬 Consumer-insight teams — pull comments and danmaku to see what viewers say at each second of a launch video.
• 🎮 Game and anime studios — measure reaction to a trailer with the danmaku profile: peak minute, danmaku per minute, unique senders.
• 🤖 AI and NLP teams — build Chinese text corpora from comments and danmaku, with timestamps.
• 📈 Agencies — schedule monitorMode runs and get only new uploads and comments each day.

How much does it cost to scrape Bilibili?

Bilibili Scraper is pay-per-event (the Pricing tab always shows the live rates). There is no start fee.

Event	Price	When it is charged
video	$0.0009 per video ($0.90 per 1,000)	Each video row delivered after your filters
comment	$0.00044 per comment ($0.44 per 1,000)	Each comment or reply row
danmaku	$0.000018 per danmaku ($0.018 per 1,000)	Each danmaku, packed or as a row
creator	$0.0009 per creator	Each creator profile row
danmaku-profile	$0.0045 per profile	Each video part profiled
hot-search	$0.00045 per topic	Each hot-search row
live-room	$0.0009 per live room	Each live-room row

Worked examples:

• The sample input (search "原神", 20 videos) = 20 × $0.0009 = $0.018
• 1,000 videos from a keyword = $0.90
• 1 video + 100 comments = $0.0009 + 100 × $0.00044 = $0.045
• All danmaku of a 45-minute video (about 9,500) = $0.17; its reaction profile = $0.0045
• A creator's 935 uploads + their profile = 935 × $0.0009 + $0.0009 = $0.84
• The hot-search board (50 topics) = $0.0225

Apify platform usage is included in these prices. Measured on the platform at the default 256 MB: 20 videos in 3 s, 100 videos with full details in 17 s, 1,000 with full details in 123 s, 935 uploads in 15 s, 1,000 comments in 29 s.

Integrations, API, scheduling & MCP

• Schedules: run Bilibili Scraper hourly or daily from the Schedules tab; with monitorMode each run returns only what is new.
• Webhooks: fire a URL when a run finishes — push rows to your CRM, Slack or a sheet.
• No-code: Make, Zapier, n8n, Google Sheets, Airtable, Slack via Apify integrations.
• API: apify-client for Node.js or Python, and the Apify API reference.
• MCP / AI agents: connect Claude, Cursor or any MCP client through the MCP tab; switch on compact to keep answers small.

More TopApi scrapers

🎵 Douyin Scraper - 抖音 videos, creators, comments	🎬 TikTok Scraper - TikTok posts by profile, hashtag or URL	🔗 TikTok Video Scraper - TikTok videos by URL
▶️ YouTube Shorts Scraper - Shorts by channel or URL	📸 Instagram Scraper - posts and reels	🧭 All TopApi actors

Limits & empty runs

• What Bilibili shows a signed-out visitor is the ceiling. Subtitles, the commenter's IP location, follower and following lists and the AI summary are shown to logged-in viewers only, so they are not in the rows.
• Search: Bilibili returns at most about 1,000 videos per keyword; more comes from date windows, and the run summary says how many windows ran.
• Zero counts are Bilibili's: for some keywords many results show 0 views (23 of 50 for "机械键盘"); the row carries what Bilibili reports, and engagementRate stays empty when views are 0. Use minViews to skip them.
• Danmaku: Bilibili keeps a capped pool per video (7,200 of 31,800 on the test video); the two sources merged gave 9,560. Above maxDanmakuPerVideo an even spread across the timeline is returned.
• Uploads: the list a visitor gets held 935 of the 938 videos Bilibili counts for the test creator.
• Refusals: when Bilibili refuses the container's IP the request moves to the Apify proxy; if Bilibili asks for a human verification the run stops that keyword and says so. Nothing is charged for rows not delivered.
• Empty dataset: your filters removed everything, the video is deleted or private (not_found in the run summary), or monitor mode found nothing new. Lines that are not Bilibili links get a free error row.

FAQ

How many videos can Bilibili Scraper get for one keyword?

About 1,000 per search on Bilibili, and past that by date window: a test run returned 1,200 unique videos for "原神" in 24 seconds. Set maxResults to the number you need.

Can Bilibili Scraper get every video of a Bilibili creator?

Yes, with user_videos: the creator's upload list, newest first, 99 per request (935 videos in 15 seconds in testing, against the 938 Bilibili shows).

Can Bilibili Scraper get Bilibili comments and replies?

Yes. includeComments (or the video_comments mode) returns 20 comments per page, hot or newest first, as deep as maxComments allows; includeReplies adds each thread's replies. A test run read 1,000 comments of one video in 29 seconds.

What is danmaku and can Bilibili Scraper export it?

Danmaku (弹幕) are the comments that fly across a Bilibili video at a given second. Bilibili Scraper returns them with the second, colour, display mode and send time, packed into one item per video (the default) or one row each, plus an optional reaction profile.

Can I run Bilibili Scraper on a schedule?

Yes. Save the input as a task, open Schedules, pick a cadence and switch on monitorMode; a run that finds nothing new delivers no rows and charges nothing.

Can I use Bilibili Scraper with the Apify API?

Yes — start runs and read datasets with apify-client for Node.js or Python.

Can I use Bilibili Scraper through an MCP server?

Yes. Open the MCP tab, connect your client and ask for Bilibili videos, comments or creators in plain language.

Do I need a Bilibili account, cookies or a proxy?

No. Bilibili Scraper reads what Bilibili shows a signed-out visitor; no account, cookie or QR login is involved.

Why did my Bilibili Scraper run return no rows?

The video is deleted or private, your filters removed every row (the run summary lists which filter), monitor mode found nothing new, or the input was not a Bilibili link (each such line gets a free error row). Nothing is charged for those.

Is it legal to scrape Bilibili?

Bilibili Scraper collects publicly available data only, nothing behind a login. You are responsible for complying with Bilibili's terms of service, creators' copyright and data-protection law (GDPR, CCPA, PIPL); creator profiles and comments are personal data.

Your feedback

Found a bug or missing a field? Open an issue on the Issues tab — we answer within two working days.