Bilibili Scraper

Returns: one row per Bilibili video with bvid, title, authorName, publishedAt, views, likes and sourceUrl. Input: {"mode":"search","searchTerms":["python tutorial"],"maxItems":10,"includeComments":false,"includeDanmaku":false} Best for: Bilibili creator discovery, video engagement research and search trend monitoring.

{
  "bvid": "BV1rpWjevEip",
  "title": "【全748集】目前B站最全最细的Python零基础全套教程，2026最新版，包含所有干货！七天就能从小白到大神！少走99%的弯路！存下吧！很难找全的！",
  "authorName": "Python官方课程",
  "publishedAt": "2024-08-22T14:59:18.000Z",
  "views": 20754385,
  "likes": 571611,
  "sourceUrl": "https://api.bilibili.com/x/web-interface/search/type?search_type=video&keyword=python%20tutorial&page=1&page_size=50"
}

What does Bilibili Scraper do?

This Bilibili scraper collects public video data through the public Bilibili API from Bilibili (bilibili.com), the large Chinese video platform. You can search videos by keyword, fetch specific videos by URL or BV id, list every upload of a creator, or read Bilibili's popular (trending) list. For each video you get the title, description, creator name and id, publish date, duration, view count, like count, danmaku count, comment count and cover image URL.

Two optional extras go deeper. Turn on comments to attach the top comments of each video, and turn on danmaku to attach the time-synced bullet comments that scroll over the video. Two more optional switches add engagement details (favorites, coins, shares, part count) and the uploader's profile (followers, following, level, bio, video count, total likes). A further optional step uses a third-party AI processor to label each video with a content category.

No login or account is needed. The Actor reads the same public endpoints the Bilibili website uses, with polite request pacing and automatic retries on a fresh proxy session when Bilibili answers with its risk-control responses. This Actor is not affiliated with or endorsed by Bilibili.

Typical uses:

• Track what Chinese creators publish on a topic and how many views each video gets.
• Build a dataset of video titles and descriptions for language or market research.
• Monitor a creator's uploads over time.
• Collect audience reactions from comments and danmaku for sentiment analysis. Tip: Watch the popular list for emerging trends.

What data can you get?

Every record is one video. Fields that a given mode cannot provide are returned as null.

Field	Type	Description
id	string	Stable id, equal to the BV id
bvid	string	Bilibili BV video id
aid	number	Legacy numeric video id
title	string	Video title (search highlight tags removed)
description	string	Description written by the uploader
authorName	string	Creator display name
authorMid	number	Creator numeric id
publishedAt	string	Publish time, ISO 8601 UTC
durationSeconds	number	Video length in seconds
views	number	Play count
likes	number	Like count (in creatorUploads mode it is read from the video detail endpoint)
danmakuCount	number	Number of danmaku on the video
commentCount	number	Number of comments
coverUrl	string	Cover image URL
tags	array	Video tags (from search results, or the tag endpoint in other modes)
url	string	Public video URL
sourceUrl	string	Endpoint the record came from
comments	array	Top comments, only with includeComments
favorites	number	Favorite count, only with includeDetails
coins	number	Coins thrown at the video, only with includeDetails
shares	number	Share count, only with includeDetails
pageCount	number	Number of parts in the video, only with includeDetails
authorFollowers	number	Creator follower count, only with includeCreatorProfile
authorFollowing	number	Accounts the creator follows, only with includeCreatorProfile
authorLevel	number	Creator account level, only with includeCreatorProfile
authorVerifiedTitle	string	Creator verification title if any, only with includeCreatorProfile
authorBio	string	Creator signature text, only with includeCreatorProfile
authorVideoCount	number	Creator upload count, only with includeCreatorProfile
authorTotalLikes	number	Creator total likes, only with includeCreatorProfile
danmaku	array	Danmaku texts with video time, only with includeDanmaku
contentCategory	string	AI label, only with enrichWithAI

Each comment holds rpid, authorName, authorMid, message, likes, replyCount and publishedAt. Each danmaku holds text, timeSeconds and sentAt.

How to use Bilibili Scraper

1. Open the Actor and pick a mode.
2. For search, enter one or more keywords. For video, paste video URLs or BV ids. For creatorUploads, paste creator page URLs such as https://space.bilibili.com/1458132. For popular, no target is needed.
3. Set Max videos to the number of results you want.
4. Optionally enable comments, danmaku or AI category.
5. Keep the proxy on residential. Bilibili blocks datacenter addresses after only a few requests.
6. Click Start and download the dataset as JSON, CSV, Excel or HTML, or read it through the API.

Search returns up to about 1000 results per keyword, which is a limit of Bilibili's own search. Results are de-duplicated by BV id within a run, and the Actor never returns more than Max videos.

Choosing a mode

Search is the best starting point when you know a topic but not the creators. Each keyword is paged until Max videos is reached or Bilibili runs out of results. With several keywords, they are processed in order and a video that matches more than one keyword is returned once.

Video is for a known list of videos. It returns the most accurate counters because it reads the video detail endpoint. Full www.bilibili.com and m.bilibili.com video URLs, bare BV IDs and legacy av IDs are accepted. For example, use ["BV1rpWjevEip"] or ["av113006243481679"] in startUrls with mode="video". Query strings and video part parameters on full URLs are accepted; the output remains one row for the video.

Creator uploads walks the upload list of each creator newest first, thirty videos per page. The list gives views, comment count and danmaku count; the Actor reads likes from the video detail endpoint. Supply a creator page URL or numeric creator ID in startUrls, for example ["1458132"].

Popular returns the current trending list in ranked order. It needs no input other than Max videos.

Tips for reliable runs

Tip: Keep Max videos modest while testing, then raise it. Requests are spaced by one to two seconds on purpose, so large runs take time.

• Comments and danmaku each add requests per video. For a quick list of videos, leave them off. Tip: If a run reports that requests failed, check that the proxy setting still uses a residential group and run again.

Input

Field	Type	Description
mode	string	search, video, creatorUploads or popular. Default search
searchTerms	array	Keywords for search mode
startUrls	array	Video URLs or BV ids (video mode), creator URLs or ids (creatorUploads mode)
maxItems	integer	Maximum videos returned, 1 to 10000, default 50
includeComments	boolean	Attach top comments, default false
commentsPerVideo	integer	Comments per video, default 20
includeDanmaku	boolean	Attach danmaku, default false
includeDetails	boolean	Add favorites, coins, shares and part count, default false
includeCreatorProfile	boolean	Add creator followers, level, bio and totals, default false
danmakuPerVideo	integer	Danmaku per video, default 100
enrichWithAI	boolean	Add an AI content category, default false
proxyConfiguration	object	Apify Proxy settings, residential by default

Example input:

{
  "mode": "search",
  "searchTerms": ["python tutorial"],
  "maxItems": 10,
  "includeComments": false,
  "includeDanmaku": false,
  "proxyConfiguration": { "useApifyProxy": true, "apifyProxyGroups": ["RESIDENTIAL"] }
}

Example for a single video with comments and danmaku:

{
  "mode": "video",
  "startUrls": ["https://www.bilibili.com/video/BV1rpWjevEip/"],
  "includeComments": true,
  "commentsPerVideo": 20,
  "includeDanmaku": true,
  "danmakuPerVideo": 50
}

If an input is invalid, for example a URL that is not a Bilibili video, the run fails right away with a message that says what to fix. A valid target that has no videos or remains blocked ends successfully with zero items. The final status names the relevant input to change: broaden searchTerms, check video or creator IDs in startUrls, or retry the popular list later. Check residential proxy availability if requests are blocked. Each target is handled separately, so a failed target does not prevent trying the next one. Targets that have not produced a result have a 70-second request budget before the Actor moves on.

Output

A shortened real record from a local search run on October 11, 2026 (see the dataset for complete data). Counters and titles can change:

{
  "id": "BV1rpWjevEip",
  "bvid": "BV1rpWjevEip",
  "aid": 113006243481679,
  "title": "【全748集】目前B站最全最细的Python零基础全套教程，2026最新版，包含所有干货！七天就能从小白到大神！少走99%的弯路！存下吧！很难找全的！",
  "authorName": "Python官方课程",
  "authorMid": 3546597933714079,
  "publishedAt": "2024-08-22T14:59:18.000Z",
  "durationSeconds": 143894,
  "views": 20664837,
  "likes": 568771,
  "danmakuCount": 132638,
  "commentCount": 359083,
  "coverUrl": "https://i2.hdslb.com/bfs/archive/a979056b1a32012cdd00d48fbc3732d253e30620.jpg",
  "url": "https://www.bilibili.com/video/BV1rpWjevEip/",
  "sourceUrl": "https://api.bilibili.com/x/web-interface/search/type?search_type=video&keyword=python%20tutorial&page=1&page_size=50"
}

Titles, descriptions, comments and danmaku are returned in the original language, mostly Chinese. Nothing is translated.

How much does it cost?

The Actor is charged per result, meaning per video returned. An extra event applies only to videos that were successfully labelled when AI category is on. Open the Pricing tab of the Actor for the current rates. Comments and danmaku do not add separate charges, but they add requests and therefore run time.

Use with AI agents (MCP)

Connect an MCP client to the Apify MCP server and expose this Actor as a tool:

{
  "mcpServers": {
    "apify": {
      "url": "https://mcp.apify.com?tools=axiomworks/bilibili-scraper"
    }
  }
}

Example prompts:

• "Find the 20 most viewed Bilibili videos about machine learning and list title, creator and views."
• "Get the latest 30 uploads of https://space.bilibili.com/1458132."
• "Fetch the top 20 comments and 50 danmaku for this Bilibili video."

Integrations and API

Python:

from apify_client import ApifyClient

client = ApifyClient("<APIFY_TOKEN>")
run = client.actor("axiomworks/bilibili-scraper").call(run_input={
    "mode": "search", "searchTerms": ["python tutorial"], "maxItems": 10,
})
for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(item["title"], item["views"])

JavaScript:

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: '<APIFY_TOKEN>' });
const run = await client.actor('axiomworks/bilibili-scraper').call({
    mode: 'search', searchTerms: ['python tutorial'], maxItems: 10,
});
const { items } = await client.dataset(run.defaultDatasetId).listItems();
console.dir(items);

cURL:

curl -X POST "https://api.apify.com/v2/acts/axiomworks~bilibili-scraper/run-sync-get-dataset-items" \
  -H "Authorization: Bearer $APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"mode":"search","searchTerms":["python tutorial"],"maxItems":10}'

You can also connect the dataset to Make, Zapier, Google Sheets or webhooks through Apify integrations.

FAQ

Do I need a Bilibili account? No. Only public data that loads without logging in is collected.

Why is a residential proxy the default? Bilibili answers datacenter addresses with risk-control errors after a few requests. Residential traffic is far more reliable, and the Actor retries on a fresh session when it is blocked.

Why does a search return fewer videos than I asked for? Bilibili search exposes at most about 1000 results per keyword, and a creator or keyword may simply have fewer videos.

How does creatorUploads mode get likes? The upload list has no like counts, so the Actor makes one extra detail request per video to read them; runs in this mode are somewhat slower.

Why are comments fewer than the comment count? Only the top comments are collected, up to commentsPerVideo, and Bilibili limits how deep anonymous paging goes.

What is danmaku? Short comments tied to a moment in the video that scroll across the screen. Each one has its text and the video time in seconds.

How does the AI category work? When enabled, the title and description are sent to a third-party AI processor that picks one of tutorial, entertainment, news, gaming, music or other. If the processor is unavailable the video is still returned without a category and no enrichment charge is made.

Can I run it on a schedule? Yes, use Apify schedules to monitor a creator or keyword regularly.

Is it legal to scrape Bilibili?

The Actor collects only information that is publicly visible without logging in, and it does not bypass logins or CAPTCHAs. Whether your use of the data is lawful depends on your jurisdiction, the purpose and how you handle personal data such as creator names and comments. Bilibili's terms of service may restrict automated access, and you are responsible for complying with them and with data protection law such as GDPR. Collect only what you need, do not republish copyrighted content, and seek legal advice for commercial use.

Feedback

If a field is missing, a mode misbehaves or you need another data point, open an issue from the Actor page on Apify. Include the input you used and the run link so the problem can be reproduced quickly.