Douban (豆瓣) Movie & TV Scraper — Ratings, Reviews, Comments & Hot Lists

Scrape Douban (豆瓣) — China's most trusted movie & TV community — into clean datasets: trending & top-rated lists, full detail (cast, crew, IMDb id, synopsis), Douban ratings, long-form reviews, and short comments with location — all with no login, no account, no cookies and no API key.

🔥 What is it — and when should you use it?

Douban is China's IMDb + Letterboxd: the authoritative source for 豆瓣评分 (Douban ratings) on films and shows. This Actor gives you four things in one, keyless:

1. Trending / top-rated discovery — 18 curated movie & TV lists (hot, newest, top-rated, hidden gems, by region & type).
2. Full title detail — rating & vote count, directors, full cast, genres, countries, languages, runtime, IMDb id, alternate titles, synopsis.
3. Long-form reviews — author, title, rating, date, snippet, review URL.
4. Short comments with location — author, rating, date, province/location, useful votes, text.

Point it at trending lists, at specific Douban subject IDs/URLs, or turn on Enrich trending to do everything in one zero-config run. Use it for film/TV research, recommendation engines, sentiment & audience analysis, and market tracking.

🤖 Use with AI agents

Pay-per-event and agent-friendly: run with zero input, or pass subject IDs. Each result row is one clean record tagged with its resultType. Point any MCP-enabled assistant (Claude, ChatGPT, Cursor) at it on the Apify MCP server, or call the Apify API. You pay per delivered result; empty runs are free.

📋 Overview

Source	movie.douban.com (discovery JSON + detail/review/comment pages)
Auth	None — no login, cookie, account or API key
Method	Pure HTTP over a rotating Apify proxy exit; anti-bot proof-of-work solved in-process (no browser)
Output	hot, detail, review, comment rows
Billing	Pay per delivered result (per type)

✨ Features

• 🔓 Keyless — no login, no account, no cookies. The Douban "sec" anti-bot proof-of-work is solved in-process; no browser, no CAPTCHA service.
• 🎯 Zero-config — run empty for all trending lists; add enrichTrending for full detail + reviews + comments on everything discovered.
• ⭐ Trusted ratings — the 豆瓣评分 and vote count on every rated title.
• 🎬 Rich detail — directors, full cast, genres, countries, languages, runtime, IMDb cross-id, aka titles, synopsis.
• 💬 Reviews & comments — long-form reviews and short comments with province/location.
• 🧱 Reliable — a sticky exit holds anti-bot clearance (one proof-of-work unlocks many pages); blocked exits rotate.
• 🧭 Honest scope — every field is null when Douban did not carry it; never a fake value.

🎬 Quick Start

Trending + full enrichment (everything), zero-config:

{ "categories": ["movie_hot", "movie_top"], "enrichTrending": true, "scrapeComments": true }

Deep-scrape specific titles:

{ "subjectIds": ["1292052", "https://movie.douban.com/subject/37516455/"], "scrapeReviews": true, "scrapeComments": true }

Just the trending lists (cheapest):

{ "categories": ["movie_hot", "tv_hot"], "maxResults": 500 }

📥 Input

Field	Type	Description
categories	array	Trending lists (movie 9 + TV 9). Empty + no IDs = all.
sort	string	recommend, time, rank.
maxResults	integer	Cap on trending titles. Default 500.
subjectIds	array	Douban subject IDs or URLs to deep-scrape.
enrichTrending	boolean	Deep-scrape every discovered trending title. Default false.
scrapeDetail	boolean	Fetch full detail per deep-scrape subject. Default true.
scrapeReviews	boolean	Fetch long-form reviews. Default false.
scrapeComments	boolean	Fetch short comments with location. Default false.
maxReviewsPerTitle / maxCommentsPerTitle	integer	Per-title caps.
maxSubjects	integer	Cap on deep-scraped subjects per run. Default 200.
proxyConfiguration	object	Apify Proxy; switch to residential if exits are blocked.

📤 Output

Each row carries a resultType. Examples:

detail

{
  "resultType": "detail", "id": "1292052", "type": "movie",
  "url": "https://movie.douban.com/subject/1292052/",
  "title": "肖申克的救赎 The Shawshank Redemption",
  "ratingScore": 9.7, "ratingVotes": 3353472,
  "directors": ["弗兰克·德拉邦特 Frank Darabont"],
  "actors": ["蒂姆·罗宾斯 Tim Robbins", "摩根·弗里曼 Morgan Freeman"],
  "genres": ["剧情", "犯罪"], "countries": ["美国"], "languages": ["英语"],
  "releaseDates": ["1994-09-10(多伦多电影节)", "1994-10-14(美国)"],
  "runtimeMinutes": 142, "imdbId": "tt0111161",
  "aka": ["月黑高飞(港)", "刺激1995(台)"],
  "cover": "https://img.../p.jpg", "datePublished": "1994-09-10",
  "synopsis": "一场谋杀案使银行家安迪…", "scrapedAt": "2026-10-28T..."
}

comment

{ "resultType": "comment", "subjectId": "37516455", "commentId": "32514679",
  "author": "如小果", "ratingStars": 5, "ratingText": "力荐",
  "time": "2024-02-27 21:43:23", "location": "北京", "votes": 265, "text": "…" }

review

{ "resultType": "review", "subjectId": "1292052", "reviewId": "1000369",
  "url": "https://movie.douban.com/review/1000369/", "author": "大头绿豆",
  "title": "十年·肖申克的救赎", "ratingStars": 5, "time": "2005-05-12", "snippet": "…" }

hot (trending list item): id, type, category, tag, rank, title, ratingScore, cover, url, isNew.

Download as JSON, CSV, Excel or HTML.

🧠 How it works

Trending lists come from Douban's public search_subjects JSON. Detail / review / comment pages are guarded by Douban's "sec" anti-bot: a request without clearance gets a page carrying a sha512 proof-of-work. The Actor solves that proof-of-work in-process (pure Node, ~20 ms), submits it for a clearance cookie, and then fetches pages normally — one solve unlocks many pages on the same sticky exit. No browser, no CAPTCHA service.

🧪 Data quality — measured (2026-10-28)

• Trending: a 1,000-title sweep returned 1,000/1,000 unique, every field 100% filled, every title rated (median ≈ 8.1).
• Detail: on real titles, rating, votes, directors, full cast, genres, countries, languages, runtime, IMDb id, aka and synopsis all parse (verified against live pages).
• Comments: author, rating, time, votes and text 100%; location fills on current titles (older pre-location comments are honest null).
• Reviews: author, title, rating, date, snippet and URL 100%.
• Anti-bot: two subjects + five pages needed only one proof-of-work solve (clearance amortized).

💼 Use cases

• Film/TV research — ratings, cast/crew, IMDb cross-ids across trending and specific titles.
• Recommendation engines — rated discovery lists + full metadata.
• Audience & sentiment — comments with location and long-form reviews.
• Market tracking — watch a genre/region's momentum and reception over time.

🔗 Integration

Schedule it, connect to Make / Zapier / n8n, or call the Apify API. Each run's per-type counts are written to the SOURCE_REPORT key-value record.

📊 Pricing

Pay per delivered result, by type (trending title, detail, review, comment). Empty runs are free. See the pricing panel for current rates.

❓ FAQ

• Does it need my Douban login or cookies? No. It never logs in and never asks you for an account, cookie or proxy of your own.
• How does it get past Douban's anti-bot? By solving Douban's own proof-of-work in-process — no browser, no third-party CAPTCHA service.
• Why is a field null? Douban did not carry it (e.g. an unrated new title, or an old comment with no location). It is never faked.

🐛 Troubleshooting

• A run returned little/nothing: an exit was rate-limited — rerun (exits rotate) or switch to a residential proxy.
• Deep scrape slow: lower maxCommentsPerTitle / maxReviewsPerTitle or maxSubjects.

🚧 What it does not do

It reads publicly served pages. It does not log in, does not scrape private lists, and does not fabricate missing fields.

⚖️ Is it legal to scrape Douban?

The Actor reads only publicly served, non-personal data (titles, public ratings, public reviews/comments). Use the data in line with Douban's terms and applicable law; do not store personal data without a lawful basis.

🤝 Support & contact

Questions or a custom need? Open an issue on the Actor's Issues tab.

Changelog

See CHANGELOG.md.