Extract every link
Paste text, an HTML source, a Markdown file or a server log and the output lists each link once, in the order it first appears. [docs](https://example.org/start) gives https://example.org/start, and <https://example.org> loses its angle brackets.
Where a link ends
A link written into a sentence picks up the sentence’s punctuation. The extractor trims trailing . , ; : ! ? ' " *, then looks at closing brackets. A ) or ] is kept only when the link already contains a matching opener, which keeps Wikipedia’s Mercury_(planet) intact and drops the bracket from (see https://example.com).
Hosts are checked
The host must be an IPv4 or IPv6 address, localhost, or a name ending in a real top-level domain. https://intranet/wiki and http://build.zzzz are skipped, and http://10.0.0.1:8080/admin is kept with its port.
Relative links
/about, ../img/logo.png and #section have no host and are not listed. Links that start with www. are listed as written, without adding a scheme.