Extract links from text

Which links the tool recognizes

It picks up addresses starting with http://, https:// or ftp://, and those written without a protocol that start with www. followed by a domain, like www.example.com: that is how people write them in texts, and they were the ones missing. A file name like report.pdf does not start with www and stays out.

Punctuation stuck to the end (the full stop that ends a sentence, commas, colons) is removed, and a closing parenthesis is removed only if the address did not open it: in (see https://site.com/page) it stays out, while a Wikipedia address ending in Rome_(city) stays whole. Quotation marks, the curly ones from Word documents too, close the link, while an apostrophe between two letters stays inside, as in pages about L'Aquila or Sant'Angelo.

Duplicates and domains: reading the counters

Duplicates are removed while keeping the order of first appearance, without looking at the capitals of the protocol and of the domain, which don't matter on the web: HTTPS://Site.COM/a and https://site.com/a are the same address. The rest is compared as written, so https://site.com/page and https://site.com/page/ remain two distinct addresses, and a link written with www. stays different from the same link written with http://, because for the first one the protocol is unknown.

The site counter groups links by domain, ignoring the www prefix: handy for seeing at a glance how many different sources an article cites, checking a newsletter's links before hitting send, or taking inventory of the references in a thesis.

The same text, five things to look for

Above the box there are five buttons, and the same text can be run through the other kinds too, with the same rules: Email extractor, Extract hashtags, Extract mentions and Extract phone numbers.

The same text with the same button gives the same list on all five pages.