Divya Raj

Posted on Nov 2

How to Scrape Bilibili Comments, Danmaku and UP主 Data in Python (2026, No Login)

#python #ai #productivity #webscraping

Bilibili (哔哩哔哩, B站) is China's biggest video platform for anime, gaming, tech and lifestyle. Its comment sections and danmaku (弹幕, the bullet comments that fly across the video) are a goldmine for market research, sentiment analysis and influencer discovery.

The problem: Bilibili has no official public API for this data. Its web endpoints now need a w_rid / wts signature (WBI signing), they throw silent risk-control errors (-352, -412) and they rate-limit you. Most tutorials from before 2023 use old unsigned endpoints and simply don't work anymore.

In this post I'll show two ways to get the data without a Bilibili account:

No code: a ready-made Bilibili scraper on Apify, with a 1-click export to Excel.
Python: the same open-source library, run locally.

👉 Try it: Bilibili Scraper on Apify · Source on GitHub

What you can scrape

Data	What you get
Comments	Hot or newest first, with full nested reply threads. 23 fields per comment: user, likes, pinned, VIP, verified, "liked by uploader", IP location, reply count…
Danmaku (bullet comments)	Every bullet comment as its own row, with its playback time, colour, mode and send time
Videos	Views, likes, coins, favourites, shares, danmaku and reply counts, tags
UP主 (creators)	Name, bio, level, verified title, VIP, followers, following, video count, total likes
Creator uploads	Every video a creator uploaded, sorted by date, views or favourites
Search & trending	Keyword search (sort by views, date, danmaku…), the popular feed and the top-100 ranking
Subtitles / transcripts	Including AI captions. This is the one thing that needs a login cookie

Option 1: No code (Apify)

Open Bilibili Scraper and click Try for free. A free Apify account includes monthly credit.
Pick What to scrape: videos, comments, danmaku, subtitles or users.
Give it a source: paste video links, space.bilibili.com/... creator links or search links, or just type keywords such as 原神.
Click Start, then Export to Excel, CSV or JSON (or send it straight to Google Sheets).

Example: the comments, with replies, on the top 20 search results for a keyword:

{
  "mode": "comments",
  "searchQueries": ["原神"],
  "maxVideosPerSource": 20,
  "maxCommentsPerVideo": 200,
  "includeReplies": true
}

Scraping UP主 (creator) profiles

Set mode to users and pass creator IDs (the number in space.bilibili.com/<UID>):

{
  "mode": "users",
  "userIds": ["2", "546195"]
}

Each row has name, sign (bio), level, verified, vip, followers, following, videos, likes and the profile url. To get a creator's videos instead, use mode: "videos" with the same userIds.

Pricing: pay per result, no subscription. $0.005 per comment, video or creator profile and $0.001 per danmaku, so 1,000 comments cost about $5.

Option 2: Python

The scraper is open source (MIT):

git clone https://github.com/Dunkmap/Bilibili-Scraper
cd Bilibili-Scraper
pip install -e .

From the command line:

bili comments BV1xx411c7mD --replies -o comments.csv
bili danmaku BV1xx411c7mD -o danmaku.csv
bili user 2                                   # UP主 profile
bili uploads 2 --order click --limit 100 -o uploads.csv
bili search "原神" --order click --limit 200 -o search.csv

Or as a library:

import asyncio
from bilibili_scraper import Scraper

async def main():
    async with Scraper(rate=4) as s:
        video = await s.video("BV1xx411c7mD")
        async for c in s.comments(video["bvid"], limit=500, replies=True):
            print(c["uname"], c["content"])

        creator = await s.user(video["owner_mid"])
        print(creator["name"], creator["followers"])

        async for v in s.user_videos(video["owner_mid"], order="click", limit=20):
            print(v["title"], v["views"])

asyncio.run(main())

How it handles w_rid signing and risk control

WBI signing: fetches img_key / sub_key, generates w_rid and wts, caches the keys and refreshes them when Bilibili rejects a signature (-352).
Silent risk control: Bilibili sometimes answers code: 0 with only a v_voucher and no data. The scraper detects this and retries instead of treating it as an empty page.
Device cookies: generates a buvid itself, so no account is needed.
Rate limits: jittered delays and backoff retries on -412. The cloud version also switches to a proxy when it gets blocked.

Extras

Sentiment analysis: label every comment and danmaku positive, neutral or negative, using your own free Groq API key.
Delta mode: scheduled runs return only items that are new since the last run, which is handy for brand monitoring.
Danmaku-only version: if you only need bullet comments, the Bilibili Danmaku Scraper adds a keyword filter and a free per-video "highlight" timeline.

FAQ

Is there an official Bilibili API? No public one for comments or danmaku. This project uses the same web endpoints bilibili.com uses and handles their signing for you.

Can I read Bilibili comments without an account? Yes. Comments, danmaku, videos, search, creators and trending all work logged out. Only subtitles need a cookie.

How do I generate w_rid? You don't have to. The scraper signs every request itself.

Can I export Bilibili comments to Excel? Yes. On Apify, click Export → Excel. Locally, use -o comments.csv.

Is this legal? It reads public data the same way the bilibili.com website does. Respect Bilibili's terms, keep the request rate reasonable, and don't republish personal data.

If this helped, try the Bilibili Scraper or ⭐ the GitHub repo. Questions and feature requests are welcome in the comments!

DEV Community