Baidu Search (百度搜索) Scraper - SERP, Ranks & Snippets

From Zen Studio, creators of Baidu Maps and the #1-ranked 1688 scraper on Apify, alongside Taobao and Weibo. Established tools for Chinese search and marketplace research.

Why choose this actor? | 为什么选择本采集器

Five keywords. 250 results. About 21 seconds. Collect Chinese and English search results in a single export, with up to 50 keywords per run.
五个关键词，250 条结果，约 21 秒。每次运行支持最多 50 个中英文关键词。

Real links and usable search rankings. Get the destination URL, full search snippet, rank, domain, publisher and available dates. Duplicate results are removed within each keyword.
返回真实网址、完整摘要、排名、域名、来源及可用日期；每个关键词的结果独立去重。

Get more than the result list. Enable search-page extras for Baidu’s written answers, local businesses with published phone numbers and ratings, news and trending searches, at no extra charge.
开启附加内容，可获取百度答案、本地商家及其公开电话和评分、新闻与热搜，不额外收费。

Quick start | 快速开始

Try two keywords with up to 25 results each and search-page extras enabled.
先试两个关键词，每个最多 25 条结果，并开启搜索页附加内容。

{
  "keywords": [
    "新能源汽车 政策",
    "china ev subsidy"
  ],
  "maxResultsPerKeyword": 25,
  "includeSerpFeatures": true
}

Open actor input / 打开输入 to enter your keywords or set a date range. Export JSON, CSV or Excel.

Snippets are search excerpts, not full articles. Search-page extras vary by query and appear in a separate row for each keyword.
摘要不是文章全文。附加内容因关键词而异，每个关键词单独输出一行。

Zen Studio · China Data Suite
Baidu Search You are here	Baidu Maps Places & contacts	1688 Wholesale products	Taobao Product details	Weibo Posts & discussions

Copy to your AI assistant

zen-studio/baidu-search-scraper on Apify. Collects Baidu (百度) web search results as one row per result: title, snippet, resolved destination url, baiduUrl, domain, sourceName, sourceLogo, sourceAuthority, thumbnail, images[{url,width,height}], position, page, dateText, publishedAt, fetchedAt, resultType. Call ApifyClient("TOKEN").actor("zen-studio/baidu-search-scraper").call(run_input={"keywords":["人工智能"],"maxResultsPerKeyword":25}), then client.dataset(run["defaultDatasetId"]).list_items().items. Inputs: keywords (required, max 50, searched verbatim, supports site: operators), maxResultsPerKeyword (default 25, max 500), dateFrom/dateTo (YYYY-MM-DD, China time, filters Baidu's own date for the page), includeSerpFeatures (default false; adds ONE extra row per keyword with resultType="serpFeatures" carrying answerBox, relatedSearches, topSearches, relatedVideos, knowledgePanel, localResults, latestNews; no extra charge). Filter resultType=="web" for search results only. Baidu itself runs out at roughly 100-200 unique results per query, so a higher limit ends early with stopReason "exhausted". publishedAt is null when Baidu's date text has no clear meaning; dateText is null on about a quarter of results. Charged per result ($3.99/1k free tier down to $2.99/1k) plus $0.01 per run; duplicates within a keyword are removed before charging. Per-keyword outcome (status, results, pagesFetched, stopReason, warnings) is in the SUMMARY record of the run's key-value store. Full spec: GET https://api.apify.com/v2/acts/zen-studio~baidu-search-scraper/builds/default (Bearer TOKEN) → inputSchema, actorDefinition.storages.dataset, readme. Token: https://console.apify.com/account/integrations

What you get | 返回内容

One row per result:

Field	Meaning
query	The query that produced this result, exactly as you typed it
title	Result title, HTML formatting removed
snippet	The search excerpt Baidu shows, complete and unshortened. It is an excerpt, not the article text
url	The destination the result points to
baiduUrl	The original Baidu result link
domain	Destination domain, without www.
sourceName	Publisher or site name as Baidu displays it (e.g. 人民网, www.runoob.com/python/)
sourceLogo	Publisher's logo image, when Baidu shows one (about a third of results)
sourceAuthority	Baidu's verified-publisher level for the source, 1 to 4, when it shows a badge (about one result in thirteen)
thumbnail, images	Result images, when the result carries any (about one in six). Each image has url, width, height; thumbnail is the first image's URL
position	1-based rank across all pages collected for this keyword
page	Which result page this row came from
dateText	The date text Baidu displays (2024年3月21日, 3天前), or null
publishedAt	dateText normalised to ISO 8601 China time, or null when its meaning is not clear
fetchedAt	When the row was collected (UTC)
resultType	web

Missing values are null, never invented. Ads are excluded.

{
  "query": "人工智能",
  "title": "为完善全球人工智能治理贡献中国智慧--理论-中国共产党新闻网",
  "snippet": "中国提出《全球人工智能治理倡议》，强调发展人工智能应符合和平、发展、公平、正义…",
  "url": "http://theory.people.com.cn/n1/2026/0914/c40531-40798069.html",
  "baiduUrl": "http://www.baidu.com/link?url=Vn0Y1u5kfihqpXe6SM68S3c52nlAiCOeMQXAvH_BrR8…",
  "domain": "theory.people.com.cn",
  "sourceName": "人民网",
  "sourceLogo": "https://t14.baidu.com/it/u=3311898808,3993627011&fm=195…",
  "sourceAuthority": 2,
  "thumbnail": null,
  "images": [],
  "position": 1,
  "page": 1,
  "dateText": "4天前",
  "publishedAt": "2026-09-13T00:00:00+08:00",
  "fetchedAt": "2026-09-17T12:52:41Z",
  "resultType": "web"
}

Search page extras (optional)

Switch on Include search page extras and each keyword also gets one row with resultType: "serpFeatures":

Field	What it is
answerBox	Baidu's own written answer at the top of the page (百看), title plus full text
relatedSearches	相关搜索 / 大家还在搜
topSearches	百度热搜, the trending list
relatedVideos	相关视频, with source, duration and date
knowledgePanel	百度百科 entity card
latestNews	最新相关信息, the recent-news strip on live topics: title, snippet, destination URL, publisher, date
localResults	On local-intent queries, the map pack: place name, address, phone, rating, review count

No extra charge: it all comes from the page the keyword already fetched.

---

Input | 输入参数

Input	Required	Default	Description
keywords	✅	none	Search queries, one per line, up to 50 per run. Chinese and English both work. Searched exactly as typed, with no translation and no rewriting. Operators such as site:zhihu.com work
maxResultsPerKeyword		25	Unique results per keyword, across pages (max 500)
dateFrom		none	Only results dated on or after this day (YYYY-MM-DD, China time)
dateTo		none	Only results dated on or before this day (YYYY-MM-DD, China time)
includeSerpFeatures		false	Add the extras row described above

{
  "keywords": ["人工智能", "electric vehicles china"],
  "maxResultsPerKeyword": 100,
  "dateFrom": "2026-01-01",
  "dateTo": "2026-06-30"
}

---

Run summary

Every run writes a machine-readable SUMMARY record to the default key-value store, so a pipeline can tell a genuinely empty search apart from a keyword that did not finish:

{
  "queries": [
    {"query": "上海 咖啡馆", "status": "ok", "results": 25, "pagesFetched": 1,
     "stopReason": "limit_reached", "warnings": []},
    {"query": "潜水艇维修 拉萨", "status": "empty", "results": 0, "pagesFetched": 1,
     "stopReason": "no_results", "warnings": []}
  ],
  "totalResults": 25,
  "failedQueries": []
}

status is ok, empty, partial or failed. stopReason is limit_reached, exhausted, no_results, page_limit_reached, source_unavailable, budget_reached or time_limit. The same keys are written whether the run succeeds or fails. One keyword failing never stops the others, and results are written as they are collected.

---

Good to know | 注意事项

• The snippet is a search excerpt, the same text a searcher sees under the title. It is not the article body, and the actor does not open the linked pages to fetch one.
• Dates are Baidu's date for the page, which is usually the publication date. For pages Baidu re-crawls it can be Baidu's own record instead. About a quarter of web results carry no date at all, and those come back as null. The date filter uses the same signal, so a filtered run can still include a few undated results.
• Baidu runs out of results. Most queries have roughly 100 to 200 unique results in total; asking for more ends the keyword early with stopReason: "exhausted". That is the source's ceiling, not a limit of this actor.
• Ads are removed from the results.
• Up to 50 keywords per run. Three keywords are collected at a time, so 50 finishes well inside the run's time limit even on a slow day. For bigger batches, start more runs.
• Same URL under two different keywords is kept twice, once per keyword. Duplicates are removed within a keyword.

---

Pricing

Pay per result, not per page and not per minute. No monthly fee.

Plan	Search result charges
Free / Starter	$3.99 per 1,000 results
Scale	$3.49 per 1,000 results
Business and above	$2.99 per 1,000 results

100 results ≈ $0.40 in result charges, and a result already collected for the same keyword is never charged twice. Other Baidu search actors bill per page of about 10 results, which works out at $4 to $6 per 1,000.

---

Related actors | 相关采集器

Zen Studio China Search   •  Search, maps, marketplaces and social, from inside China
Baidu Search ➤ You are here	Baidu Maps Places, contacts, reviews	Taobao Products, prices, SKUs	Weibo Posts and engagement