Google News Scraper — Real Article URLs + Full Text

Scrape Google News by keyword, topic, or top headlines — and get the real publisher URL for every article (decoded from Google's redirect links), with optional full article text. Built for news monitoring, competitor/brand intelligence, and AI/RAG pipelines.

No API key, no login.

Why this scraper

• Real article URLs, not Google redirects — most Google News scrapers hand you an opaque news.google.com/rss/articles/CB… link. This actor decodes it to the actual publisher URL (e.g. https://www.bloomberg.com/news/…), so your links are usable everywhere.
• Full article text (optional) — extract readable body text, author, and image for each article. Best-effort by design: paywalled/bot-walled pages keep their metadata, note the reason, and are not charged the content fee.
• Built for monitoring — deduplicate across scheduled runs via a named store, so a daily run returns only what's new. Already-seen articles are skipped without wasting requests.
• Search, topics & headlines — keyword queries (with Google News operators), the 8 topic sections, and Top Headlines, in any language/country edition, with time-range filters.

Input

{
    "queries": ["\"electric vehicles\"", "openai"],
    "topics": ["technology", "business"],
    "includeTopHeadlines": false,
    "language": "en-US",
    "country": "US",
    "timeRange": "week",
    "maxArticlesPerQuery": 100,
    "resolveUrls": true,
    "extractArticleText": true,
    "dedupByUrl": true
}

Field	Default	Description
queries	—	Search terms (support "exact", intitle:, site:)
topics	[]	world, nation, business, technology, entertainment, sports, science, health
includeTopHeadlines	false	Also fetch the Top Headlines feed
language / country	en-US / US	News edition
timeRange	any	any, hour, day, week, month, year (search queries)
maxArticlesPerQuery	100	Cap per feed (~100 available)
resolveUrls	true	Decode Google links to real publisher URLs
extractArticleText	false	Also fetch and extract article body text
dedupByUrl	true	Skip articles seen in previous runs
dedupStoreName	google-news-dedup	Named store per watchlist

Provide queries, topics, or enable includeTopHeadlines.

Output

{
    "title": "OpenAI completes $7 billion employee tender - Bloomberg",
    "source": "Bloomberg",
    "url": "https://www.bloomberg.com/news/articles/2026-08-10/openai-buys-back-7-billion",
    "googleNewsUrl": "https://news.google.com/rss/articles/CBMi…",
    "publishedAt": "2026-08-10T14:03:00.000Z",
    "query": "search:openai",
    "language": "en-US",
    "country": "US",
    "snippet": "OpenAI has completed a share sale…",
    "articleText": "…full readable body when extraction is enabled…",
    "wordCount": 812,
    "author": "Jane Doe",
    "image": "https://…/cover.jpg",
    "scrapedAt": "2026-08-11T23:10:00.000Z"
}

When extractArticleText is off, articleText/wordCount/author are omitted. When a page can't be extracted, articleText is null and extractionError explains why (that row is not charged the content fee).

Pricing

Pay per event:

• Article — $0.003 each ($3.00 / 1,000): one news article with metadata and the resolved real URL.
• Article content — $0.002 each: charged only when full text is successfully extracted (extractArticleText on). Paywalled/blocked pages are free.
• Actor start — $0.00005 per GB.

Platform usage (compute, proxy, storage) is included. Set Maximum total charge to cap spend.

Run	You pay
100 headlines, URLs resolved, no text	$0.30
100 articles with full text (70% extracted)	~$0.44
Daily monitor, ~20 new articles/day with text	~$0.10/day

Use cases

• Brand & PR monitoring — schedule keyword runs; get only new coverage with real links.
• Competitor intelligence — track competitor mentions across sources.
• Market research — topic feeds by country/language for trend analysis.
• AI / RAG — feed resolved URLs and clean article text into LLM pipelines.
• Newsletters & aggregators — assemble fresh, deduplicated story lists.

Limitations

• Google News returns up to ~100 items per feed; use multiple queries/topics or countries for breadth.
• Full-text extraction is best-effort — some publishers paywall or block automated fetches. Those rows still include the resolved URL and metadata, and aren't charged the content fee.
• Cross-run dedup uses a named store; use a distinct dedupStoreName per separate watchlist so their histories don't mix.

FAQ

Do I get the real article link? Yes — that's the point. Keep resolveUrls on (default).

Why isn't every article's text extracted? Paywalls and anti-bot protection. You're never charged the content fee for a failed extraction.

How do I run a daily monitor? Put your query in queries, keep dedupByUrl on, and schedule the actor. Each run returns only new articles.