Website Content Extractor

How to extract text from a url in three steps

Paste the address, run it, take the section you need. That is the whole workflow, and it replaces opening developer tools and hunting through markup.

The page is fetched once and parsed into groups. Headings sit together, paragraphs sit together, images and links each get their own list, and every meta tag is reported as written. You copy one group rather than taking everything and deleting the rest.

Relative links are resolved against the final URL after any redirects, not the address you typed. That sounds pedantic until you extract from a site that redirects between www and non-www — resolve against the wrong one and every link you collected quietly 404s.

One URL per run. We do not crawl a whole site, and that is a deliberate limit rather than a missing feature. Bulk crawling is what turns a convenience tool into something other people abuse.

Turning a webpage to plain text without the navigation

Copying a page by hand drags the menu, the sidebar and the cookie banner along with the part you wanted. Selecting around them is fiddly and you always miss something.

Grouping the output by element type solves most of that. Take the paragraphs and you leave the navigation behind, because navigation is links, not prose.

Where it still struggles: sites that wrap body copy in the same markup as their furniture. A page built entirely from styled containers gives the parser very little to separate. In our experience news sites and documentation are clean, marketing pages are messy.

If you already have the markup on your clipboard, skip the fetch. Our HTML to text converter handles pasted source without touching a server at all.

Is a free website text extractor enough for research?

For reading and reference, comfortably. For anything automated, you will hit two walls, and we would rather name them than have you discover them.

JavaScript-rendered pages. We fetch the first HTML response. On a React, Vue or Angular site that response is close to an empty shell, because the content arrives later from script. Your browser runs those scripts; a fetch does not. The full explanation is in why some pages come back empty.

Sites that refuse non-browser requests. A 403 here usually means the site blocked us on reputation, headers or address. That is the site's decision, and no extraction tool routes around it reliably.

We are not adding headless browser rendering to fix the first one. Running a full browser per request is expensive and a far larger attack surface, so we explain the limit instead of hiding it.

When to extract article text online rather than copy by hand

Three cases where this is clearly faster.

Feeding a model. Raw HTML burns tokens on markup that carries no meaning. If the structure matters — headings, lists, tables — markdown beats plain text, which is what our URL to markdown converter produces. There is more on the trade-off in preparing pages for an LLM.

Auditing your own pages. Reading the extracted meta tags is faster than checking them in developer tools, and it shows you what a crawler receives rather than what your browser renders after scripts run.

Collecting references. Pulling body text from a dozen sources into one document takes minutes rather than an afternoon. To get only the links, extract all links from a webpage separates them into internal and external with their anchor text.

What you can do with the content, and what you cannot

Reading a public page is ordinary browsing. Extraction changes the speed, not the permission.

What does not change is copyright. Text does not become yours because it was easy to copy, and a site's terms may say more about automated collection than most people check. We cover the shape of that in whether web scraping is legal — context rather than legal advice.

Politeness is the practical half. One page at a time, at human speed, is unlikely to bother anyone. Our guide to scraping etiquette that keeps you unblocked covers what actually gets people blocked, which is rarely what they expect.

Publishing sites can help here too. The llms.txt proposal is one attempt at telling automated readers what is worth reading, though adoption is still unproven.

Getting the most from this website content extractor

Start with the section you need rather than the whole page. Export JSON whenever something downstream will parse the result, because labelled data beats a wall of text.

If a result comes back empty, check whether the site renders client-side before assuming anything is broken. That single check explains most of the confusion we hear about.

For the parts of a page that are not text, two neighbours help: capture the page as an image, or download every image on a page in one go.