Article Text Extractor — Clean Text & Metadata from Any URL

Turn any article, blog post, or news page URL into clean, readable text (or Markdown) plus metadata — title, author, publish date, site name, language — with the navigation, ads, and boilerplate stripped away. Plain HTTP requests only, no browser: fast even on long lists of URLs.

What it does

For each URL, this Actor sends a normal HTTP GET request (the same request any browser would send), then runs the page through a content-extraction engine that identifies the actual article body and separates it from menus, sidebars, ads, and other boilerplate. You get the article's clean text (or Markdown) and its declared metadata as one dataset row.

What you get (output fields)

One dataset row per URL:

• url — the URL that was requested.
• finalUrl — URL after following redirects (null on error).
• title — article title.
• author — article author, when declared.
• publishedDate — publish date as declared by the page.
• siteName — name of the publishing site, when declared.
• language — detected/declared language code of the article.
• description — short description/excerpt, when declared (e.g. meta description).
• text — extracted article body, in the format you chose (plain text or Markdown). null on error.
• wordCount — word count of the extracted text. null on error.
• error — set only if the page could not be fetched or no article text could be found; null on success.

If a page cannot be read

A URL that is completely unreachable (DNS failure, connection refused, timeout), that returns an HTTP error status (404, 500, ...), or that has no extractable article text (e.g. a homepage, a login page, a page that is mostly navigation) becomes a row with error set. The rest of your list keeps running, the run does not fail, and you are not charged for that URL.

How to use

1. URLs — paste the article/blog/news page URLs you want extracted.
2. Output format — text for plain text (default) or markdown to keep headings, bold/italic, and links.
3. Include comments — when enabled, reader comments below the article (if the page has them) are appended to the extracted text. Off by default, for a clean article-only result.
4. Run the Actor. Each URL becomes one dataset row.

Input example (JSON)

{
  "urls": ["https://en.wikipedia.org/wiki/Web_scraping", "https://www.python.org/doc/essays/blurb/"],
  "outputFormat": "text",
  "includeComments": false
}

Output example (JSON)

{
  "url": "https://en.wikipedia.org/wiki/Web_scraping",
  "finalUrl": "https://en.wikipedia.org/wiki/Web_scraping",
  "title": "Web scraping - Wikipedia",
  "author": null,
  "publishedDate": "2005-09-17",
  "siteName": "Wikimedia Foundation, Inc.",
  "language": null,
  "description": null,
  "text": "Web scraping\nWeb scraping, web harvesting, or web data extraction is data scraping used for extracting data from websites...",
  "wordCount": 3841,
  "error": null
}

Pricing

Pay only for pages actually extracted — nothing charged for pages that could not be fetched or had no article text. Pricing model: pay-per-event.

Event	When it's charged	Price
article-extracted	a page was fetched and its article text successfully extracted	0.002 USD

Not charged for errors: unreachable URLs, HTTP error pages, and pages with no extractable article text never trigger article-extracted.

Limitations

• HTTP-only: it does not run a browser, so pages that need JavaScript to render their content will yield little or no text.
• Extraction is heuristic: pages that are not article-shaped (homepages, search results, login walls) will often return an error row rather than a false positive.
• publishedDate reflects whatever date the page itself declares; not every page declares one.
• No login, no CAPTCHA solving, no bypass of any site protection.
• Internal/private/loopback addresses and non-http(s) URLs (e.g. file://) are refused before any request is made.

FAQ

Does this use a headless browser like Playwright/Puppeteer?

No, by design — it uses plain HTTP requests only, which keeps runs fast and cheap. Pages that only render their article via client-side JavaScript are not supported.

Am I charged if a page has no article text?

No. You are only charged for pages where article text was actually found and extracted.

Can I get Markdown instead of plain text?

Yes, set "Output format" to markdown to keep headings, bold/italic, and links.

What happens with duplicate URLs?

They are de-duplicated before the run starts; each URL is fetched once.

Can this get past paywalls or login-protected pages?

No. This Actor makes a plain, unauthenticated HTTP request — it does not log in, solve a paywall, or bypass any access control.

Does it render JavaScript-heavy pages?

No. Only the initial HTML response is read; pages that build their article content client-side with JavaScript (single-page apps) can come back with a short or empty text.

Can I use this through the Apify API or an MCP server?

Yes. Like any Apify Actor, you can run it and read results through the standard Apify API, or through the Apify MCP server if you use Claude, Cursor, or another MCP-enabled client.

Export

Results can be downloaded from the Apify dataset as JSON, CSV, or Excel, or accessed via the Apify API.