Integration Yard · Apify tools

Douban (豆瓣) Reviews Data in 2026: Ratings, 短评 and 影评 Without Login

Published October 21, 2026 · Updated October 21, 2026 · Integration Yard

TL;DR

  • Douban (豆瓣) is China's main rating and review site for movies, TV and books. Its public pages give you ratings, vote counts, short comments (短评) with star ratings, and long reviews (影评/书评), all without logging in.
  • In a live export on Oct 21, 2026, the Douban Top 250 movies came back as 250 rows. Ratings run from 8.4 to 9.7 (median 8.9), and 106 of the 250 score 9.0 or higher. The #1 film, The Shawshank Redemption, has 3.35 million ratings.
  • Short comments are a ready-made sentiment dataset: every row carries a 1–5 star rating next to the text. In the top 100 comments for To Live (活着), 27% gave 3 stars or fewer, while Shawshank got just 13%.
  • Attention is concentrated. In both titles, the top 10% of comments hold over 70% of all helpful votes.
  • Without login you get roughly the top 150–200 short comments and the top 40 long reviews per title, sorted most-helpful first. For most research, that's the part that matters.
  • Cost with our actor: $3 per 1,000 rows. Douban review data commonly sells for around $30 per 1,000 on the Apify Store.

Douban (豆瓣) is where Chinese audiences rate and argue about films, TV series and books. A Douban score is quoted in Chinese film marketing the way an IMDb or Rotten Tomatoes score is quoted in the West, and Douban's long reviews are some of the most thoughtful Chinese-language writing about culture on the open web.

Douban's public API was shut down years ago, so if you want this data for market research, recommendation models, sentiment analysis or academic work, you need another route. This guide shows what you can get without any login, what the data looks like, and how to schedule it. Every number comes from real runs on Oct 21, 2026.

Disclosure: we build and maintain the Douban Scraper used here.

Contents

What Douban data is available

Data What you get Typical use
Search (movies, TV, books) title, year, rating, vote count, link; optional full details (directors, cast, authors, publisher, genres, tags, intro) catalog matching, market sizing
Subject details full metadata and rating for any Douban movie or book URL enrichment, recommendation features
Short comments (短评) text, 1–5 stars, helpful votes, date, author sentiment, audience reaction
Long reviews (影评/书评) title, full text (optional), stars, helpful/unhelpful counts, reply count, spoiler flag deep opinion mining, LLM datasets
Charts Top 250, now in theaters, real-time hot movies, hot TV, fiction books and more trend tracking, release monitoring

No login, no cookies and no Douban account are needed. Results export as JSON, CSV or Excel.

The Top 250: a benchmark in one table

Douban's Top 250 is the reference list of what Chinese audiences consider the best films. One run returns all 250 with rank, rating, vote count and a summary line (year, country, genre, director, cast).

Douban Top 250 rating distribution
  • Ratings range from 8.4 to 9.7, with a median of 8.9. 106 films are rated 9.0 or higher.
  • The median film on the list has 748,000 ratings, and even the least-rated has 172,000. These are very stable scores.
  • By country of origin, the first one listed is the US (112 films), then Japan (33), the UK (20), Hong Kong (19), mainland China (18) and South Korea (11).
Douban Top 250 by decade

The 1990s, 2000s and 2010s dominate (56, 76 and 79 films). Only 4 films from the 2020s have made the list so far, which says as much about how long it takes to collect votes as about recent cinema.

Short comments (短评): sentiment with labels built in

Short comments are Douban's one-line reactions. The useful detail for analysts is that each comment carries the user's star rating, so you get labeled sentiment data without any manual tagging.

We exported the top 100 short comments for a movie (The Shawshank Redemption) and a book (To Live, 活着, by Yu Hua):

Star ratings on Douban short comments
Shawshank (movie) To Live (book)
5 stars 70% 43%
4 stars 17% 18%
3 stars or fewer 13% 34% (plus 5 unrated)
Most helpful comment's votes 26,520 17,595
Share of helpful votes held by top 10% 72% 74%

Two takeaways:

  1. Even classics get split reactions. To Live is one of China's most-read novels, yet a third of its most-helpful short comments give it 3 stars or fewer. That's the kind of nuance a single average rating hides.
  2. Sort by helpfulness. The top 10% of comments hold over 70% of the votes, and Douban already returns them most-helpful first, so the first 50–100 rows cover most of the audience's voice.

Long reviews (影评/书评): depth for research and LLMs

Long reviews are essays. With full text turned on, the top 20 reviews for Shawshank had a median length of 1,256 characters (the longest was over 4,000). For To Live, the median was 1,075 characters, and the longest was 5,189.

Each review row includes:

  • title and full text (or just the abstract if you only need a preview)
  • stars (the reviewer's rating)
  • usefulCount / uselessCount: community votes, a quality signal
  • commentsCount: how much discussion it started
  • spoiler flag, date and author

The top Shawshank review, written in 2005, has 18,956 "useful" votes and over 1,000 replies. For training or evaluating Chinese-language models, that's long-form, human-written, quality-ranked text with a rating label attached.

Charts: what's in theaters right now

Chart mode returns Douban's live lists. On Oct 21, 2026, the now in theaters list had 40 titles, and 29 of them already had a rating. The top entries ranged from 6.2 to 7.3, with vote counts from a few thousand to 185,690. Run it daily to see how a new release's score settles over its first weeks. That's useful for distributors, streaming buyers and anyone tracking the Chinese box office.

Other chart ids include movie_real_time_hotest, tv_hot and book_fiction.

What it costs

Cost per 1,000 Douban rows

Our Douban Scraper charges $3 per 1,000 rows (plus $0.001 per run). Every run has a hard spending cap you set.

Job Rows Cost
Full Top 250 250 ≈ $0.75
100 short comments for 10 titles 1,000 ≈ $3.00
Top 20 full-text reviews for 50 books 1,000 ≈ $3.00
Daily now-in-theaters chart for a month ≈ 1,200 ≈ $3.60

Code: Python and the Apify API

from apify_client import ApifyClient

client = ApifyClient("YOUR_APIFY_TOKEN")
run = client.actor("dheerajitbhu/douban-scraper").call(run_input={
    "mode": "reviews",
    "subjectUrls": ["https://book.douban.com/subject/4913064/"],
    "maxReviews": 20,
    "fullReviewText": True,
    "maxChargeUsd": 0.5,
})
for row in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(row["stars"], row["usefulCount"], row["title"])

Change mode to "comments" for short comments, "chart" with "charts": ["movie_top250"] for the Top 250, or "search" with "searchQueries": ["流浪地球"] to look titles up by name. To keep a dataset fresh, schedule the run daily in Apify.

Limits, ethics and data protection

  • Without login, Douban shows the top ~150–200 short comments and the top 40 long reviews per title. Those are the most-helpful ones, which is usually what you want. If you need every comment ever posted, this isn't the tool.
  • Only public data. The actor never logs in and never asks for your cookies.
  • Personal data: comments and reviews include public usernames. If you're under GDPR, PIPL or similar rules, keep only the fields you need. Aggregate or pseudonymize author fields for analysis.
  • Be polite: runs are paced, and you should collect what you need rather than everything.

FAQ

Is there an official Douban API in 2026? Not a public one. Douban closed its open API to new developers years ago. A no-code tool is the practical way to get the data now.

Do I need a Douban account or a Chinese phone number? No. Everything in this guide works without login.

Can I get the Douban Top 250 books too? Search and chart modes cover books. Use book search for specific titles, or book charts such as book_fiction.

How many reviews can I get per movie? Up to about 40 long reviews and 150–200 short comments per title, most-helpful first.

Does it work for TV series? Yes. Douban's movie search also returns TV series, and tv_hot is a chart.

What formats can I export? JSON, CSV, Excel or XML from the dataset. There's also a compact output for LLM pipelines.

Is it legal to scrape Douban? It collects publicly visible data only. Check your own jurisdiction and use case, and follow the data protection notes above.

Try it