News Article Extractor

Turn any news article URL into clean structured data — headline, full text, authors, publish date, and images, with the ads, navigation, and cookie banners stripped out. Point it at a news site and it finds the articles for you.

What does News Article Extractor do?

Give it article links, or give it a news site and let it discover the articles itself. For each article you get the headline, the full body text with boilerplate removed, the authors, the publication date, the lead image, and the page's own metadata — description, keywords, tags, language, and canonical URL.

It works on any news site rather than a fixed list, so you are not limited to a set of supported publishers.

Why use News Article Extractor?

• Media monitoring — track coverage of a brand, person, or topic across many outlets.
• AI and RAG pipelines — clean article text is far better training and retrieval input than raw HTML.
• Content research — study how competitors cover a subject, and how often.
• Archiving — keep a durable, readable copy of reporting that may change or disappear.
• Newsletters and aggregation — pull the day's stories from a set of sources on a schedule.

Running it on Apify adds scheduling, an API, integrations (Sheets, Slack, Zapier, S3), residential proxy rotation, and run monitoring.

How to use News Article Extractor

1. Paste article links into Article URLs, or news site homepages into News sites — or both.
2. Set Max articles.
3. Click Start.

Results appear in the Output tab as they are extracted, and download as JSON, CSV, Excel, or XML.

Input

Field	Description
article_urls	Direct links to articles, one per line.
site_urls	Homepage or section URLs. The Actor finds article links on each page and extracts them.
max_articles	Total cap across every site and URL given.
language	Two-letter hint such as en or ar. Leave empty to detect it — set it only when detection gets a site wrong.
include_article_html	Also return the cleaned HTML of the article body, keeping links and formatting.
max_concurrency	How many pages to fetch at once. Lower it if a publisher rate-limits you.
proxy_country	Comma-separated country codes for the residential proxy rotation.

Output

{
  "url": "https://techcrunch.com/2026/08/20/example-story/",
  "canonical_url": "https://techcrunch.com/2026/08/20/example-story/",
  "title": "Apollo Atomics wants to make nuclear power cheap",
  "text": "A nuclear start-up has raised forty million dollars...",
  "word_count": 498,
  "authors": ["Jane Roe"],
  "published_at": "2026-08-20T10:30:00+00:00",
  "language": "en",
  "site_name": "TechCrunch",
  "description": "The company plans to complete regulatory review this year.",
  "keywords": ["nuclear", "energy"],
  "tags": ["energy"],
  "top_image": "https://techcrunch.com/wp-content/uploads/lead.jpg",
  "images": ["https://techcrunch.com/wp-content/uploads/inline.jpg"],
  "videos": []
}

You can download the dataset in various formats such as JSON, HTML, CSV, or Excel.

Data fields

Field	Description
url, canonical_url	The link scraped, and the publisher's own canonical version of it.
title, text, word_count	Headline and clean body text.
authors, published_at	Byline and publication date, as an ISO timestamp where the page declares one.
language, site_name	Detected language and the publication's name.
description, keywords, tags	The page's own metadata.
top_image, images, videos	Lead image, other images in the article, and embedded video URLs.
article_html	Cleaned HTML of the body, when you ask for it.
error	Present only when a page could not be read, with the reason.

How much does it cost to extract articles?

Cost tracks the number of articles you extract, so Max articles is the main lever on what a run costs.

Tips

• Section pages beat homepages. Pointing at example.com/technology/ gives more focused results than the homepage.
• Lower the concurrency to 1 or 2 if a publisher starts refusing you. Five works well across mixed sources.
• Schedule it against the same sites daily and the dataset becomes a rolling archive.
• Check the error field. Rows that could not be read are still written, so nothing disappears silently.

FAQ and support

Which news sites does it support? Any of them. There is no supported-site list to check against.

Why did an article come back with no text? Some articles are behind a paywall, and a few sites load their text in a way that cannot be read. Those rows come back with an error explaining why, rather than disappearing, so you can see exactly which links need another approach.

Why is the publish date sometimes empty? Not every publisher declares one in a machine-readable way. When it is missing from the page, it is missing from the output — the Actor does not guess.

Is scraping news articles legal? Article text is usually copyrighted. Extracting it for analysis, monitoring, or internal research is common practice, but republishing it is a different matter. You are responsible for how you use the output, including copyright and each publisher's terms. Consult a lawyer if you are unsure.

Found a bug or want a field that is missing? Open an issue on the Actor's Issues tab.