URL Extractor

About the URL Extractor

Both full http(s):// links and bare www. addresses get pulled from your pasted text and deduplicated as you type. Copying or downloading normalizes any www. link to a proper https:// URL, and none of it touches a server, the matching runs entirely client-side.

What's shown on screen isn't always exactly what gets copied

Worth knowing: the list displayed in the results box shows each URL in its original matched form — a bare www.example.com stays exactly as typed. But clicking Copy All or Download as .txt reads from a separate, normalized version of the list where every bare www. address has already been rewritten with an https:// prefix. That's deliberate: the on-screen preview reflects exactly what was found in your text, while the copied or downloaded output is upgraded to fully working, clickable links, since a bare www. string isn't reliably clickable everywhere it might get pasted.

The matching pattern stops at quote marks and angle brackets on purpose

The regex's character class explicitly excludes whitespace, single and double quotes, and angle brackets from what counts as part of a URL. That matters most when pasting raw HTML page source: without those exclusions, a URL sitting inside an href="..." attribute could accidentally swallow the closing quote and the start of the next HTML tag as if they were part of the link itself. Stopping at those specific characters keeps extracted URLs clean even when the pasted text is markup rather than plain prose.

The same link in two different forms won't be recognized as a duplicate

Deduplication compares the lowercased, trailing-punctuation-stripped text of each match exactly as it was found — not its normalized form. That means www.example.com and http://www.example.com appearing in the same pasted text are treated as two distinct entries and both show up in the results, even though they'd both resolve to the same working address once normalized. It's an honest limitation of comparing raw matched text rather than fully resolved URLs.

A bare domain with no www. or http(s):// prefix won't be caught at all

The extractor's pattern only recognizes text starting with http://, https://, or www. — a domain typed as plain example.com, with neither prefix, doesn't match either branch of the regex and is silently skipped. That's a real scope boundary of pattern-based matching: distinguishing a genuine bare domain from an unrelated piece of text containing a period would require a much more complex (and far more false-positive-prone) pattern than this tool uses.

A specific set of trailing punctuation is trimmed off every match

After a URL is matched, any run of periods, commas, semicolons, colons, exclamation points, question marks, closing parentheses, or closing square brackets at the very end is stripped away with replace(/[.,;:!?)\]]+$/, ""). That covers the common case of a link sitting at the end of a sentence — "check this out: https://example.com." — where the sentence's own closing punctuation would otherwise get glued onto the extracted URL and break it if someone tried to actually visit the link.

Sibling tools on this site solve a similar problem with a different technique

This site's Phone Number Extractor uses regex lookbehind and word-boundary assertions to stop a match from bleeding into surrounding digits. The URL Extractor takes a different approach to a related problem — rather than boundary assertions, it relies on an explicit exclusion character class during matching (to keep quotes and brackets out entirely) combined with a separate trailing-punctuation cleanup pass afterward. Both techniques solve the same underlying issue — making sure a match doesn't accidentally include characters that belong to the surrounding text — just suited to the different shapes of what they're each extracting.