Extract URLs from Text

Four kinds, and you can switch each one on or off:

• Full web addresses starting with http:// or https:// (ftp:// too), wherever they sit in the text.
• Addresses without a scheme, like www.example.com or example.com/pricing. The tool checks the ending against a list of real top-level domains, so "version 2.0" and "file.txt" don't get picked up by mistake.
• Links inside HTML tags, meaning the href of links, the src of images, scripts and iframes, and a few others like srcset and form actions. Paste a page's source and you'll get everything it points to.
• Email addresses, which are off by default because most people don't want them mixed in with their URLs.

It also copes with the messy stuff: a URL followed by a period or comma at the end of a sentence, one wrapped in parentheses or angle brackets, one that was HTML-escaped so its ampersands became &, and Markdown links where the address is in parentheses after the label.

The most common one we hear is pulling links out of a long email thread or a Slack export so they can be checked or saved somewhere sensible. After that it's SEO work: grabbing every URL from a competitor's sitemap or a Google search results page you've copied, collecting the outbound links from a draft before publishing so they can be checked, or lifting the sources from a research document into a spreadsheet. Writers use it to build a reference list from their notes. Support teams use it to pull the links a customer pasted into a ticket.

Once you have the list, the formats help. "Domains only" shows you who a piece of content links to at a glance. CSV drops straight into a spreadsheet with a count for each address. HTML and Markdown give you a ready-made list of clickable links for a post or a wiki page.

Links copied from newsletters, social media, and ads usually carry extra baggage after a question mark: ?utm_source=newsletter&utm_medium=email, ?fbclid=..., ?gclid=..., and so on. That's tracking information for someone else's analytics, and it means the same page can show up five times in your list with five different tails.

"Strip tracking parameters" removes only the known tracking ones (utm_ anything, fbclid, gclid, msclkid, mc_cid, mc_eid, ref, and a few more) and keeps parameters that actually change the page, like ?page=2 or ?product=123. "Strip all query strings" is the blunt version: it cuts everything after the question mark. Use the first one by default, and the second when you only care about which pages were linked, not which version of them.

With either one turned on, duplicates are merged after cleaning, so those five copies of the same page collapse into one line with a count of 5 if you choose the "how many times" format.

Those are relative links. The page's HTML says "go to /about/" and leaves it to the browser to add the domain. The tool can't guess which site the code came from, so fill in the "Base URL" box with the page's address and it will turn every relative link into a full one, the same way a browser would. Links that start with two slashes, like //cdn.example.com/script.js, are given https:// automatically.

Yes, with one catch. When you copy text out of Word, Google Docs, or a PDF and paste it here, you get the words you can see, and a hyperlink's address is usually hidden behind its label. "Read our guide" comes through as those three words, not as the URL underneath. Any address that's written out in full in the text is found without trouble.

To get the hidden ones too, you need the document as HTML:

• Google Docs: File, Download, "Web page (.html, zipped)". Open the HTML file in a text editor, copy everything, and paste it here with "Links in HTML tags" ticked.
• Word: File, Save As, "Web Page, Filtered", then the same as above. Or press Alt+F9 in the document to show every link's address in the text, copy that, and paste it.
• PDF: if the links are clickable in the PDF, open it in a browser, print it to HTML with a converter, or simply copy the text if the addresses are printed in full, as they are in most reports and academic papers.

Almost always because the text broke it. Email clients wrap long addresses onto a second line, and the tool, like a browser, treats the line break as the end of the URL. The same happens with a space someone typed in the middle of a link. Join the pieces back together in the box and the full address will appear.

A few other causes: an address ending in an unusual domain the tool doesn't recognize when there's no http:// in front of it (add the scheme and it will always be picked up), a link that's an image or a file when "Images and files" is unticked, or a URL that's been shortened with a service like bit.ly, which the tool will list as-is because it never follows links.

You can paste up to about five million characters, which is a few thousand pages of text, and the whole thing is processed in your browser in well under a second. Nothing is sent to us or to anyone else. That makes it fine for internal documents, client emails, and anything else you wouldn't want to upload to a random website. If you close the tab, it's gone; only your checkbox settings are remembered, and those live in your own browser.

The one thing it deliberately doesn't do is visit the links. It doesn't check whether they work, follow redirects, or fetch titles. If you need to know which of them are broken, run the list through our Redirect Chain Checker next.