Smart Article Extractor 📰

Turn any article URL into clean title, author, date, text and Markdown — no API key, no login, no config.

Point it at any news story, blog post, or documentation page and get back a tidy, structured object ready for RAG pipelines, LLM ingestion, news monitoring, and datasets. Built on the battle-tested trafilatura engine with a readability fallback, so it works across thousands of sites out of the box.

• ✅ No key required — just paste URLs and run
• ✅ Clean text + Markdown — perfect for feeding LLMs
• ✅ Rich metadata — author, publish date (ISO), description, site name, tags, main image, language
• ✅ Word count + reading time — built in
• ✅ Robust — Chrome-124 fetch with redirect follow, urllib fallback, and a BeautifulSoup readability fallback when needed
• ✅ Cheap — around $1 per 1,000 articles

---

What it does

For every article URL you provide, the actor:

1. Fetches the page HTML (curl_cffi with a real Chrome fingerprint, following redirects; urllib fallback).
2. Extracts the main article content with trafilatura — stripping nav, ads, comments, and boilerplate.
3. Falls back to a readability heuristic (largest <article>/<main>/<p>-dense block) if needed.
4. Returns one clean structured item per article.

Use cases

• RAG / LLM ingestion — feed clean Markdown into vector stores and prompts without HTML noise.
• News monitoring — track coverage across outlets and pull the full body text.
• Content archiving — snapshot articles as durable, structured records.
• Datasets — build training/eval corpora with consistent fields.

Input

Field	Type	Default	Description
mode	select	extract	Operation mode (currently extract).
urls	array	—	Article URLs to extract.
includeMarkdown	boolean	true	Also return the body as clean Markdown.
maxItems	integer	100	Max articles per run (1–1000).

Example:

{
  "mode": "extract",
  "urls": [
    "https://www.bbc.com/news/articles/cewrx0ldlkko",
    "https://apnews.com/article/some-story"
  ],
  "includeMarkdown": true,
  "maxItems": 100
}

Output

Each dataset item (all fields nullable):

{
  "url": "https://www.bbc.com/news/articles/cewrx0ldlkko",
  "final_url": "https://www.bbc.com/news/articles/cewrx0ldlkko",
  "title": "BBC visits Iran school where strike killed 120 children",
  "author": "Nawal Al-Maghafi",
  "date": "2026-07-23T00:00:00",
  "description": "Nearly five months on, grief is still raw...",
  "text": "Clean article body text...",
  "markdown": "# Heading\n\nClean article body in Markdown...",
  "language": "en",
  "sitename": "BBC News",
  "tags": ["iran", "middle east"],
  "image": "https://ichef.bbci.co.uk/news/1024/.../image.png",
  "word_count": 1772,
  "reading_time_min": 9,
  "source": "article-extractor",
  "scraped_at": "2026-07-23T13:02:39.924934+00:00"
}

text and markdown are capped at 50,000 characters. URLs that fail or contain no readable article are skipped with a warning.

Pricing

Pay-as-you-go on Apify platform usage — roughly $1 per 1,000 articles. No subscription, no key.

Related actors

• Website Content Crawler — crawl an entire site and extract content.
• HTML to Markdown Converter — convert raw HTML to Markdown.
• RSS Feed Finder — discover feeds for any site.
• PDF Text Extractor — pull clean text from PDFs.

Keywords

article extractor, web scraper, clean text, markdown, RAG, LLM, AI content feed, news scraper, readability, trafilatura, boilerplate removal, content extraction, article parser, text extraction, dataset builder, no API key.

---

Built by ninhothedev.