I have this:

[(http(s)?):\/\/(www\.)?a-zA-Z0-9@:%._\ ~#=]{2,256}\.[a-z]{2,6}\b([-a-zA-Z0-9@:%_\ .~#?&\/\/=]*)

It matches for:

www.example.com 
http://example.com.nz 
example.com
http://www.example.com?2rjl6
example.com/first/second
https://example.us.edi?34535/534534?dfg=g&fg

etc...

I want no match if any of the above URLs are enclosed in square brackets [ ] like this:

[www.example.com]
[http://example.com.nz]
etc...

The text is long and may or may not contain more than one URL, spaces, line breaks, and so on.

e.g.

Lorem ipsum dolor sit amet, consectetur [http://example.com.nz] llamcorper et lacus. Morbi sodales convallis lectus a efficitur: example.com/first/second vitae nisl placerat.

Fusce non ipsum a augue http://example.com.nz http://www.example.com?2rjl6 aculis augue. Nullam eu nulla lectus.

In this case there should be only 3 matches.

I tried adding:

(?![^\[]*\])

But it doesn't work as expected.

Can you help me with this or recommend another approach? Thanks.

CodePudding user response：

You can match from an opening till closing square bracket, and then make use of SKIP FAIL using php.

You might also shorten the pattern a bit. You have the whole first part in a character class [(http(s)?):\/\/(www\.)?a-zA-Z0-9@:%._\ ~#=] but you can move the square bracket to before the [a-z,

And as you can write a-zA-Z0-9_ as \w, you can shorten the character class a bit starting with [\w

If you choose a different delimiter than / like ~ you don't have to escape the backslash.

\[[^][]*](*SKIP)(*F)|(?:https?://)?(?:www\.)?[\w@:%. ~#=]{2,256}\.[a-z]{2,6}\b[\w-@:% .~#?&/=]*

Explanation

• \[[^][]*] Match from [...]
• (*SKIP)(*F) Skip the match
• | Or
• (?:https?://)? Optionally match the protocol
• (?:www\.)? Optionally match www.
• [\w@:%. ~#=]{2,256} Repeat 2-256 times any of the listed in the character class
• \.[a-z]{2,6}\b match a dot and 2-6 chars a-z followed by a word boundary
• [\w-@:% .~#?&/=]* Optionally match what is listed in the character class

Regex demo