Naver Search Scraper

Scrape Naver Search — South Korea's #1 search engine. Returns organic results with title, URL, domain, snippet and source name from the web / blog / cafe / news tabs. Korean SERP, no login required. Uses curl_cffi (chrome131 impersonation) with lazy Apify AUTO proxy escalation.

What this actor does

• Four tabs: webSearch (default), blogSearch, cafeSearch, newsSearch — server-rendered organic blocks
• Pagination via Naver's start parameter, early-stop on empty pages, cross-page URL dedupe
• Filters: result keyword (title/snippet/domain/source), domain allow-list (naver.com, blog.naver.com, …)
• Block detection: 403/429/451/503 or content markers missing → lazy-engage Apify AUTO datacenter proxy + backoff
• Empty fields are omitted; every record carries sourceUrl, scrapedAt, recordType

Output per result

• title, url, domain, sourceName, snippet
• resultUrl (= url)
• Blog/cafe records additionally carry date (post date / relative date)
• News records additionally carry date, naverNewsUrl (n.news.naver.com mirror), pressUrl (media.naver.com press profile)
• serpUrl (the SERP page it came from), recordType: "searchResult" | "error", scrapedAt

Input

Field	Type	Default	Description
mode	select	webSearch	webSearch / blogSearch / cafeSearch / newsSearch
searchQuery	string	비비바	Query (Korean or any language)
containsKeyword	string	–	Title/snippet/domain substring filter
domain	string	–	Domain allow-list (namu.wiki, blog.naver.com)
maxItems	int	20	Hard cap (1–500)

Example: web search for a Seoul restaurant

{
  "mode": "webSearch",
  "searchQuery": "서울 맛집",
  "maxItems": 30
}

Example: blog results only, filtered by domain

{
  "mode": "blogSearch",
  "searchQuery": "비비바",
  "domain": "blog.naver.com",
  "maxItems": 10
}

Example: latest news articles

{
  "mode": "newsSearch",
  "searchQuery": "서울 맛집",
  "maxItems": 20
}

Anti-bot notes

Plain curl gets rate-limited; curl_cffi chrome131 + browser UA returns 200 with real organic results (verified locally: 15 results/page for 비비바, apple). The image and kin tabs are JS-rendered (results not present in the SSR HTML) and are therefore not exposed — documented rather than faked.

Data source

search.naver.com SERP HTML (SSR organic blocks). No login required.

Limitations

• The image and kin tabs are JS-only and not supported.
• Naver currently serves the same merged result set for the blog and cafe tabs (where=blog and where=cafe both return blog+cafe posts — Naver unified these tabs in the current UI). blogSearch and cafeSearch outputs may therefore be identical; both modes are kept as they mirror Naver's tab API.
• Blog/cafe organic blocks carry the post URL on the Keep button (data-url); for the small subset of blocks without it (editor-recommended posts) the URL is resolved through Naver's ader.naver.com click-tracker (307 redirect).
• Snippets are extracted from the SSR block text with boilerplate tokens removed; occasional nav residue is possible.
• Invalid queries produce typed recordType: "error" records (no silent drops).