0

Being noob in regex I require some support from community

Let say I have this string str

  1. www.anysite.com hello demo try this link

  2. anysite.com indeed demo link

  3. http://www.anysite.com another one

  4. www.anysite.com

  5. http://anysite.com

Consider 1-5 as whole string str here

I want to convert all 'anysite.com' into clickable html links, for which I am using:

str =  str.replace(/((http|https|ftp):\/\/[\w?=&.\/-;#~%-]+(?![\w\s?&.\/;#~%"=-]*>))/g, '<a href="$1" target="_blank">$1</a>');

This converts all space separated words starting with http/https/ftp into links as 

<a href="url" target="_blank">url</a>

So, line 3 and line 5 has been converted correctly. Now to convert all www.anysite.com into links I again used

str = str.replace(/(\b^(http|https|ftp)?(www\.)[-A-Z0-9+&@#\/%?=~_|!:,.;]*[-A-Z0-9+&@#\/%=~_|])/ig, '<a href="https://$1" target="_blank">$1</a>');

Though it only converts www.anysite.com into link if it is found at very beginning of str. So it convert line number 1 but not line number 4.

Note that I have used ^(http|https|ftp)?(www.) to find all www not starting with http/https/ftp, as for http they already have been converted

Also the link on line number 2, where it is neither started with http nor www rather it ends with .com, how the regex would be for that.

For reference you can try posting this whole string to you facebook timeline, it converts all five line into links. Check snapshot

enter image description here

Anant
  • 534
  • 11
  • 40
  • Possible duplicate of [What is a good regular expression to match a URL?](http://stackoverflow.com/questions/3809401/what-is-a-good-regular-expression-to-match-a-url) – nicael Jul 28 '16 at 16:13
  • You could have googled for a regex to match the URL, and the problem is solved (you do know already how to use replace, right?) – nicael Jul 28 '16 at 16:14
  • Please proofread your question title. –  Jul 28 '16 at 16:50
  • @nicael ok, this would work http://regexr.com/39i0i from your provide link's solution, though still there is an issue. My simple link www.anysite.com converts to www.anysite.com but clickable link redirects to mysite.com/www.anysite.com, I am working in Joomla, may be if it is unable to find http or https before a link, it prepends site url to it. That was the reason I had to put https:// in these cases. Any solution to it? – Anant Jul 28 '16 at 17:20
  • Then just first remove http:// from any links and then add it to all the links. – nicael Jul 29 '16 at 06:04
  • @nicael oh yes, thats better. Thank you :) I post solution RegEx in answer – Anant Jul 29 '16 at 06:37
  • @nicael it doesn't work with shortened urls perfectly. Like- http://s.gl/WYWp302Xnjk will only give url till s.ly – Anant Aug 05 '16 at 15:14

2 Answers2

0

^(http|https|ftp)?(www\.) does not mean "all www not starting with http/https/ftp" but rather "a string that starts with an optional http/https/ftp followed by www..

Indeed, ^ in this context isn't a negation but rather an anchor representing the start of the string. I suppose you used it this way because of its meaning when used in a character class ([^...]) ; it is rather tricky since its meaning change depending on the context it is found in.

You could just remove it and you should be fine, as I see no point of making sure the string does not start with http/https/ftp (you transformed those occurrences just before, there should be none left).

Edit : I mentioned lookbehind but forgot it's not available in JS...

If you wanted to make some kind of negation, the easiest way would be to use a negative lookbehind : 

(?<!http|https|ftp)www\.

This matches "www." only when it's not preceded by http, https nor ftp.

Graham
  • 7,431
  • 18
  • 59
  • 84
Aaron
  • 24,009
  • 2
  • 33
  • 57
  • It says 'invalid regexp expression', on checking with http://regexr.com/ it says 'Lookbehind is not supported in JavaScript – Anant Jul 28 '16 at 16:39
  • Woops, right... Doesn't just removing the `^` solve your problem? because the alternatives aren't great (either use multiple regex to check that it contains "www..." but does not contain "http[https|ftp"), or craft an horrible regex that match every 5 or less letters preceding "www..." that do not compose http, https or ftp) – Aaron Jul 28 '16 at 16:49
  • That's what I was trying but unable to make, though I am still trying it – Anant Jul 29 '16 at 06:05
0

Thanks for help, the final RegEx that helped me is:

//remove all http:// and https://
str = str.replace(/(http|https):\/\//ig, "");

//replace all string ending with .com or .in only into link
str = str.replace( /((www\.)?[-a-zA-Z0-9@:%._\+~#=]{2,256}\.(com|in))/ig, '<a href="//$1" target="_blank">$1</a>');

I used .com and .in for my specific requirement, else the solution on this http://regexr.com/39i0i will work

Though sill there is issue like- it doesn't convert shortened url into links perfectly. e.g http://s.ly/qhdfTyuiOP will give link till s.ly

Still any suggestions?

Anant
  • 534
  • 11
  • 40