Remove email and URL

[zzz]

I'd like to be able to detect email address in a text and replace it.

For example change this:

Some demo text here [e-mail] blah blah

to

Some demo text here <email removed> blah blah

and the same for URLs in the text. I guess I would do something like this, just not sure how to detect email and URL.

$replaceEmail = "<email removed>";
$cleanTXT = preg_replace($replaceEmail, $email2remove, $cleanTXT);

[bpat1434]

Simply put:

$pattern = '~\b[^@]+@[^.]+\.[^ ]+\b~iU';
$replace = '';

$cleanTXT = preg_replace($pattern, $replace, $cleanTXT);

Domains are more tricky as you have all kinds of derivations:

www.domain.tld
domain.tld
www.domain.tld.2tld
domain.tld.2tld

You'd have to have an entire listing of tlds as far as I know. Something like:

$pattern = '~\b(www\.)?[^.]+\.(com|us|de|net|info|org|name|tv|mobi|biz)\.[a-z]{2,4}~iU';
$replace = '';

$cleanTXT = preg_replace($pattern, $replace, $cleanTXT);

I'm not great on patterns, so they may be completely wrong. But that's basically what you'd use. Both look for entire words (i.e. wrapped in spaces or non-word characters). Hope it helps.

[Jack_McSlay]

in fact the part of the e-mail after the @ is also a domain, so the same rules apply to e-mail addresses. Also, for links you must also match the part that comes after the domain

(besides, .us and .de identify the country, not domain type - USA and Germany, respectively)

this might work better:

$regex = '@(http|ftp|https)://[a-z0-9\-]+(\.[a-z0-9\-]+)*(/[^ ]*)?@i';
$string = preg_replace ($regex,'<link removed>',$string);
$regex = '/[a-z0-9\-]+(\.[a-z0-9\-]+)*@[a-z0-9\-]+(\.[a-z0-9\-]+)*/i'; 
$string = preg_replace ($regex,'<e-mail removed>',$string);

this catches some invalid urls and e-mail adresses but should remove any of them

[bpat1434]

I think my email regex would have been fine. Especially since you have to have a domain, a dot, then a TLD, and then more TLD stuff if it's there. So I think that would have been okay.

The only problem I have with the link replacement is that not all links will start with (http|https|ftp|ftps|smb):// .... some will just start with www. or just the domain name.

Either way, whatever works works 🙂

[Jack_McSlay]

I see what you mean, but if you don't include the schema, the regex will easily mistake possible parts of a text such as "best.movie.ever" for a link. then again, someone could write www[dot]example[dot]com, so there's no 100% fail-proof way to catch links on a text