Article Extractor: News & Blog Text, Author, Date to JSON

Turn any news or blog article into clean structured data: title, authors, publish and update date, publisher, section, tags, main image, word count, reading time, paywall flag, and the article text as plain text and Markdown, without menus, ads, related links or cookie banners.

Give it a news site or a section instead of an article and it finds the newest articles itself, through the site's RSS feed when it has one.

• Articles in, JSON out: one row per article with every field a news reader, a media monitor or an AI pipeline needs
• Sites and sections in, newest articles out: https://techcrunch.com or https://www.bbc.com/news/technology
• Dates and authors from the site's own structured data, then its meta tags, then the address
• Markdown with headings, lists, links and tables, ready for ChatGPT, Claude or a vector database
• Daily monitoring: keep only articles from the last day or week
• No browser, no API key

What does this article extractor do?

https://techcrunch.com
 → feed found, 15 newest articles
 "Google brings agentic AI to Gemini, starting with businesses"
 Sarah Perez · TechCrunch · AI · 8 Oct 2026 · 605 words · tags: gemini, Google

1. Paste article addresses, or home and section pages of news sites and blogs.
2. Articles are extracted directly. Home and section pages are expanded into their newest articles: through the RSS or Atom feed when the site has one, otherwise through the article links on the page.
3. You get one row per article and a CSV of the metadata.

flowchart LR
    A[Article addresses] --> D[Extract text,<br/>author, date, tags]
    B[News sites<br/>and sections] --> C[RSS feed or<br/>article links]
    C --> D
    D --> E[One row per article<br/>text + Markdown]

Who is it for?

• Media monitoring and PR: follow what tech, business or local news sites publish every day
• AI and RAG pipelines: clean article text and Markdown for summarising, classifying or embedding
• Researchers and analysts building news datasets with reliable dates and authors
• Newsletter and content curators collecting the day's stories from many sources at once
• SEO and content teams studying competitors' articles: length, sections, tags and publishing pace

What data do you get?

Articles (dataset, one row per article)

Field	What it tells you
title, description, url, canonicalUrl, image	The article
authors, publisher	Who wrote and published it
publishedAt, modifiedAt	When, as ISO dates
section, tags, articleType	The site's own section and tags; NewsArticle, BlogPosting, ...
text, markdown	The article body, without page furniture
wordCount, readingTimeMinutes, language	Length and language
isPaywalled, linksInText	Paywall flag; how many links the text has
foundVia, sourcePage	Listed by you, found in a feed, or found as a link

ARTICLES_CSV has the metadata of every article as flat columns. RUN_SUMMARY lists which feed each site used and why any page was skipped.

Example: 9 October 2026

14 inputs, 127 articles in 5 minutes 56 seconds: 12 news sites and sections, one essay and one single article.

Source	Found through	Articles
techcrunch.com	RSS feed	15
arstechnica.com	RSS feed	15
wired.com	RSS feed	15
blog.cloudflare.com	RSS feed	15
cnbc.com/technology	article links	15
theguardian.com/technology	article links	15
spiegel.de/netzwelt	article links	15
theverge.com	RSS feed	10
bbc.com/news/technology	article links	6
openai.com/news	article links	5
paulgraham.com essay	listed	1

• 120 of 127 articles came with a publish date and 106 with named authors (the rest are signed by the publication or by nobody). 15 are marked as paywalled (The Verge and Wired show a part to visitors).
• Skipped, not charged: Reuters (HTTP 401), Le Monde (HTTP 402) and The New York Times (HTTP 403) refuse automated visitors; 8 links on section pages were videos, series, topic or help pages; 1 Wired story had only 67 public words behind its paywall.

Pricing

$0.001 per article. Pages that are not articles, paywalled stubs and pages that cannot be opened are free.

• 1,000 articles: $1
• The run above: 127 articles, $0.13

Good to know

• Paywalls: articles are read as a logged-out visitor sees them. Sites that show the full text to search engines through structured data give the full text; others give what is public, flagged with isPaywalled.
• Sites that refuse automated visitors (some large newspapers) are listed in RUN_SUMMARY with the reason.
• Sections without a feed are read through the article links on the page; galleries, videos and series pages are recognised and skipped.
• Daily schedule: set Only articles from the last (days) to 1 and schedule the run every morning.
• Personal data: only the public byline of each article. Commenters are never collected.
• No login, no API key, no proxies needed.

Input

Field	Meaning
Articles or news sites	Article addresses, or home and section pages, one per line
Maximum articles per site or section	1 to 1,000 (default 20)
Only articles from the last (days)	0 = any date
Plain text, Markdown	Both on by default
Images in Markdown	Off by default
Minimum words	80 by default; shorter pages are skipped
Articles at once	1 to 20 (default 5)

Integrations

Export to JSON, CSV, Excel or Google Sheets, or connect through the Apify API, webhooks, Make, Zapier and n8n. Send the Markdown to an LLM for summaries, or to Pinecone, Qdrant or Weaviate for search.