Douban Scraper: reviews, discussions, ratings, movies and books

Scrape Douban (豆瓣), China's largest reviews and interest community, without an account. Long-form reviews with their full text, group discussions with the whole post and its replies, short comments with ratings, and complete subject records for movies, TV shows and books, from search, from ready-made lists or from any URL.

Built for Chinese-language corpora, sentiment analysis, entertainment research and market tracking, where what matters is the text people wrote, not only the score.

What you need	What this Actor does	Typical price here
Long-form reviews with full text	reviews per subject, opened one by one	US$ 10.00 / 1,000
Group discussions with the full post	topics from any group, newest first	US$ 10.00 / 1,000
Replies inside a discussion	50 per request	US$ 4.00 / 1,000
Short comments with ratings	20 per request, newest first	US$ 4.00 / 1,000
Full subject record	rating, votes, genres, cast or author, synopsis	US$ 10.00 / 1,000
Search and ready-made lists	Top 250, now in cinemas, hot TV, book lists	US$ 4.00 / 1,000

Rows that could not be delivered (removed subject, no reviews, a blocked page) come back with a stable errorCode and are never charged.

What you can ask for

You send	You get
urls: https://movie.douban.com/subject/1292052/, https://book.douban.com/subject/1084336/ or a bare id	the full subject record plus its long-form reviews (and short comments, if you ask)
searchQueries: matrix, 三体	subjects matching the term, in Chinese or English
collections: movie_top250, movie_showing, tv_hot, book_top250	the list, up to maxItems, rank included
groups: https://www.douban.com/group/732764/ or 732764	that group's topics, each with the full post text, plus replies if you ask
groups: a single discussion URL .../group/topic/<id>/	just that thread and its replies
searchQueries with searchType: group	group profiles: name, members, topic count

Available lists, with how many rows each one holds (checked 22/09/2026):

List	Rows	What it is
movie_top250	250	Douban's all-time Top 250 films
book_top250	250	the same for books
tv_hot	204	TV shows people are watching now
book_fiction, book_nonfiction	100 each	current fiction and non-fiction charts
show_hot	60	variety shows
tv_domestic	49	Chinese TV shows
movie_showing	20	now in cinemas
movie_weekly_best	10	the weekly critics' board

Ask for more than a list holds and you simply get the whole list.

Output

One row per thing, with type telling them apart: subject, review, comment, group, topic. Discussion rows carry groupId, groupName and topicId, so replies join back to their thread and their group.

{
  "type": "review",
  "id": "1000369",
  "subjectId": "1292052",
  "subjectType": "movie",
  "title": "肖申克的救赎",
  "reviewTitle": "十年·肖申克的救赎",
  "rating": 5,
  "text": "距离斯蒂芬·金和德拉邦特们缔造这部伟大的作品已经有十年了……",
  "textLength": 2284,
  "createdAt": "2005-05-12 20:44:13",
  "voteCount": 4200,
  "commentsCount": 1082,
  "authorName": "某人",
  "url": "https://movie.douban.com/review/1000369/",
  "scrapedAt": "2026-09-22T18:40:11.002Z"
}

A subject row carries title, originalTitle, year, ratingValue, ratingCount, genres, countries, durations, directors, actors for screen titles, and author, press, pubdate, pages for books, plus intro, commentCount, reviewCount and coverUrl.

Error codes

errorCode	Meaning
SUBJECT_NOT_FOUND	The subject was removed, is private or never existed
INVALID_SUBJECT	The line is not a Douban subject URL or id
NO_REVIEWS, NO_COMMENTS, NO_TOPICS	That subject or group has nothing public of that kind
GROUP_NOT_FOUND, TOPIC_NOT_FOUND	The group or discussion is gone, private or never existed
INVALID_GROUP, INVALID_TOPIC	The line is not a Douban group or discussion URL or id
NO_RESULTS	The search term returned nothing public
COLLECTION_NOT_FOUND	No list with that name, or it came back empty
BLOCKED	Douban refused every retry for that request. Run again or switch the proxy group.
ITEM_UNREADABLE	One row came in a shape this Actor cannot read. Only that row is lost, and it is free.
UNEXPECTED	Something unforeseen broke on that request. The run keeps everything already delivered, and nothing extra is charged.

Pricing

Pay per event, no monthly fee and no charge per run:

Event	Price
List item, short comment, reply or group profile	US$ 0.004 (US$ 4.00 per 1,000)
Full subject record	US$ 0.01 (US$ 10.00 per 1,000)
Long-form review or group topic	US$ 0.01 (US$ 10.00 per 1,000)

A default run (one movie, its full record and 10 reviews) costs US$ 0.11. One thousand reviews for a sentiment dataset cost US$ 10.00. The Top 250 list, as rows, costs US$ 1.00. Twenty discussions from a group, with 50 replies each, costs US$ 4.20.

How it works, and what it will not do

• No login, no account, no cookies of yours. Everything comes from Douban's public mobile API, the same one the mobile website calls.
• Douban rate-limits by IP, and the penalty lasts for minutes, so the Actor rotates sessions and IPs on every block and falls back to residential proxy. Keep the proxy on.
• Only public content, and no user tracking. Reviews and comments carry the text, the rating, the date and the display name. This Actor never returns user ids, profile links or avatars, and there is no mode to scrape a person's profile.
• Douban's desktop pages are behind a block page; this Actor does not try to break it, it uses the public mobile API instead.

FAQ

Do failed rows cost anything? No. Only rows with real content are charged, and duplicates inside a run are removed before charging.

Can I get the complete review text? Yes, fullReviewText is on by default and opens each review. Turn it off and you get the abstract Douban shows in the list, at the same price.

Chinese or English? Both work in search. The content comes as Douban stores it, which for most subjects is Chinese.

Which rows have the most text? Long-form reviews are the densest single rows, around a couple of thousand characters each. Group discussions bring the volume instead: the post itself plus dozens of replies, all of them text people wrote.

Why are there two prices for a subject? A row from a search or a list is cheap because it comes 20 or 100 at a time. Opening the subject for the full record costs one request each, so it is charged as a full record. includeSubjectDetails controls that, and it is off by default.