How to scrape Bilibili videos, comments, and creators from a BV id

Which Bilibili URLs and ids does the scraper accept?

Bilibili identifies videos by BV ids, the 12-character strings that start with BV in every video URL, and the actor extracts them from any bilibili.com/video/ link. Short links of the form b23.tv/xxx are resolved to their canonical BV URL before scraping, so links copied from the Bilibili app work as pasted. Creators are addressed by their numeric uid through space.bilibili.com URLs, and live rooms by the numeric room id in live.bilibili.com URLs.

You never have to set the platform field by hand. The default auto setting recognises bilibili.com and b23.tv hosts and routes the URL to the Bilibili pipeline, and trending mode with platform left on auto defaults to Bilibili's popular feed.

An invalid or deleted BV id produces a 404 error row rather than a failed run, and error rows are not billed.

• Video: https://www.bilibili.com/video/BV1GJ411x7h7 (BV id extracted automatically)
• Short link: https://b23.tv/xxxx, resolved to the canonical BV URL first
• Creator: https://space.bilibili.com/{uid}, used by user_posts and user_profile
• Live room: https://live.bilibili.com/{room_id}, used by live_info
• Popular feed: no URL at all, just mode trending with platform bilibili
• Bulk: a urls array processed concurrently, with 500 links per run the recommended ceiling

What fields come back for a Bilibili video?

A video_detail run returns one flat JSON row per BV id in the same unified schema used for TikTok and Douyin, so downstream code does not branch on platform. Bilibili's stat object is mapped onto the shared engagement fields: view becomes play_count, like becomes like_count, reply becomes comment_count, share becomes share_count, and favorite becomes collect_count.

Descriptive fields cover title, description, publish timestamp as ISO created_at, duration_sec, cover_url, and the creator's numeric mid plus display name. Hashtags are extracted from the description text. Two shared fields are always null on Bilibili, music_title and music_author, because the Bilibili video response carries no background audio object for the actor to map.

The raw field carries the complete platform response, so anything Bilibili returns beyond the unified fields, such as multi-part page lists, stays available for your own parsing. The video_url_nowm field is filled by a separate play URL request when downloadVideos is on, and Bilibili rotates those tokens, so download promptly or fetch through your own client.

• item_id is the BV id; url is the canonical bilibili.com/video/BV... link
• play_count, like_count, comment_count, share_count, collect_count from Bilibili's stat object
• title, description, created_at, duration_sec, cover_url, hashtags
• author_id (numeric mid), author_name, author_username
• video_url_nowm when a play URL is issued; tokens rotate, so some downloads fail
• raw: the untouched platform response for fields the unified schema does not cover

How do you scrape Bilibili comments and creator data?

Set mode to video_comments and pass a video URL to pull the reply thread. Each comment row carries comment_id (Bilibili's rpid), reply_to_id (the parent rpid, null for top-level comments), comment_text, the commenter's mid and username, a like_count, and an ISO created_at. Pagination continues until maxItems is reached; the README example pulls the top 200 comments on one video with maxItems set to 200.

For creators, user_posts walks a space URL page by page and returns one row per upload with title, description, cover, publish date, play count, and comment count. user_profile returns the creator's mid, display name, bio, avatar, and a verified flag derived from Bilibili's official account type.

Bilibili's list endpoints return thinner records than the single-video endpoint. If you need like, share, and favorite counts for a creator's whole catalogue, collect the BV ids from user_posts, then run video_detail over them in a second bulk run.

• video_comments: rpid, parent rpid, text, commenter mid, like count, timestamp
• user_posts: title, description, cover_url, created_at, play_count, comment_count per upload
• user_profile: mid, name, bio, avatar_url, verified (official account type)
• live_info: room id, live_status, live_title, viewer_count, host name, cover
• Comment and profile rows are billed as their own pay-per-event items, listed on the store page
• user_likes is not available on Bilibili and returns a clear error row

What can you do with Bilibili data for creator analytics and China market research?

Bilibili reported 371 million monthly active users and 116.5 million average daily active users for the quarter ended June 30, 2026, in results released on August 27, 2026. It is the primary home of anime, gaming, and educational long-form video in China. Datapika gives you the numbers behind those audiences without a Chinese phone number or a Bilibili account.

Analytics teams benchmark anime and gaming channels by running user_posts on a set of space URLs on a schedule, then tracking play_count and comment_count over time. Education creators and course sellers use video_comments to read how learners react to a series. Brands entering China pull the popular feed daily to see which formats and topics are gaining traction before committing to a campaign.

Bilibili is also the better choice among the actor's platforms for time-sensitive monitoring, because Douyin's public creator feed can lag by around 6 days for some accounts while Bilibili's endpoints return current data.

• Creator benchmarking: schedule user_posts across competing anime, gaming, or education channels
• Audience research: video_comments on flagship uploads, then sentiment analysis on comment_text
• Trend spotting: trending mode pulls Bilibili's popular feed as full video records
• Live commerce and event monitoring: live_info for viewer_count and live_status on a room
• Cross-platform comparison: the same schema lets you line Bilibili up against TikTok and Douyin rows

How much does scraping Bilibili cost and how does it run at scale?

Video rows cost $0.001 each as of August 2026 on the Apify Store, which works out to $1.00 per 1,000 videos from video_detail, user_posts, live_info, or trending. Comments and profiles are separate pay-per-event items with their own rates on the store's pricing tab. Error rows for deleted videos, bad BV ids, or anti-bot blocks are never billed, and you can cap a run with the ACTOR_MAX_TOTAL_CHARGE_USD setting so a large batch stops at your budget. The actor had recorded 1,340 runs on the store as of August 2026.

maxItems defaults to 100 and accepts values from 1 to 5,000 per input URL. Bilibili requests run through an HTTP client pool rather than a browser, so they are fully concurrent, and a batch of 100 URLs typically finishes in 2 to 3 minutes. Keep runs at 500 URLs or fewer and split bigger jobs for parallelism.

You can run the actor from the Apify console, call it over the Apify REST API, or expose it to an AI agent through the Apify MCP server. Residential proxy is on by default.

• $0.001 per video, so 1,000 Bilibili videos cost $1.00 (Apify Store, August 2026)
• Failed URLs return error rows and are not charged
• maxItems from 1 to 5,000 per URL, default 100
• About 100 URLs in 2 to 3 minutes; recommended maximum 500 URLs per run
• Console, REST API, and MCP access; export as JSON, CSV, or Excel from the dataset

Questions, answered

Keep reading