Preg_replace regex in PHP gives unexpected empty result

Question

I am using a regex to replace all email addresses in a string with a nice <a> to make them clickable. This works perfect, except for the case when there are two words of a certain minimum length and a dash between them in front of the email address. Only then I get an empty string as result.

<?php

$search = '#(^|[ \n\r\t])(([a-z0-9\-_]+(\.?))+@([a-z0-9\-]+(\.?))+[a-z]{2,5})#si';
$replace = '\\1<a href="mailto:\\2">\\2</a>';

$string = "tttteeee-sssstttt [email protected]";
echo preg_replace($search, $replace, $string);
// Output: "" (empty)

$string = "te-st [email protected]";
echo preg_replace($search, $replace, $string);
// Output: "te-st <a href="mailto:[email protected]">[email protected]</a>" (as expected)

$string = "[email protected] tttteeee-sssstttt";
echo preg_replace($search, $replace, $string);
// Output: "<a href="mailto:[email protected]">[email protected]</a> tttteeee-sssstttt" (as expected)

?>

I have tried everything, but I really can't find the problem. A solution would be removing the first dash in the regex (before the @ sign), but that way email addresses with a dash before the @ wouldn't be highlighted.

Before my brain is going to parse your regex: you do know {1,} == + right? — Wrikken
– Wrikken, Commented Jul 17, 2011 at 22:50
Yup. I got this regex online, so it's not my own style. I'll fix it now, + is much clearer! — Jonathan
– Jonathan, Commented Jul 17, 2011 at 22:53

Wrikken · Accepted Answer · 2011-07-17 23:26:07Z

2

OK, minimum use case: #([a-z-]+\.?)+@#, which reaches the backtrack limit (use preg_last_error()), it cannot determine where to put things, as the \. is optional, determining whether to use the inside or the outside + is a lot of work. The default limit of pcre.backtrack_limit of 100000 does not work, setting it to 1000000 does.

To solve this, make it easier on the parser: the first (([a-z0-9\-_]+(\.?))+ should become: ([a-z0-9\-_]+(\.[a-z0-9\-_]+)*), which is a lot easier to solve internally. And as a bonus, instead of the accepted answer, this still doesn't allow consecutive dots.

answered Jul 17, 2011 at 23:26

Wrikken

70.8k8 gold badges99 silver badges136 bronze badges

Sign up to request clarification or add additional context in comments.

8 Comments

Paul Over a year ago

Thanks Wrikken. Good solution. I was surprised to note that (^|[ \r\n\t])(([A-Z0-9_\-]+\.?)+@([A-Z0-9_\-]+\.?)+\.[A-Z]{2,5})($|[ \r\n\t]) worked with eregi_replace which is deprecated. So the working solution in it's entirety is: #(^|\b)(([a-z0-9\-_]+(\.[a-z0-9\-_]+)*)@([a-z0-9\-_]+(\.[a-z0-9\-_]+)*)\.[A-Z]{2,5})($|\b)#i. @Jonathon, I think you should accept Wrikken's answer instead, as I had no idea why it wasn't working before and I think this answer is more helpful.

Wrikken Over a year ago

Nice addition of \b too, and actually, I think we can forgo the ^ & $ in that case, starting & ending it with just \b instead of (^|\b) & ($|\b). And .museum is 6 characters at the moment, as long as ICANN isn't yet handing our TLD's (they start next year with custom ones, I do not know whether they limit the lenght), having \.[A-Z]{2,6} should temporarily do the trick.

Wrikken Over a year ago

And just to be a bitch about it: how would you mail the good people at http://пример.испытание/?

Paul Over a year ago

Good point, and thanks for pointing that out. So we'd end up with $search = '#\b((([a-z0-9\-_]+(\.[a-z0-9\-_]+)*)@([a-z0-9\-_]+(\.[a-z0-9\-_]+)*)\.[A-Z]{2,6}))\b#i'; But because of groups changing around $replace should also be change to: $replace = '<a href="mailto:\\1">\\1</a>';

Paul Over a year ago

Lol, I don't even want to go there. People will probably stick mostly to ascii for their email addresses and domain names for a few more years... I hope.

|

Paul · Accepted Answer · 2011-07-17 23:11:38Z

1

Try using this for your search string instead:

$search = '#(^|\b)([A-Z0-9_\-.]+@[A-Z0-9_\-.]+\.[A-Z]{2,5})($|\b)#i';

edited Jul 17, 2011 at 23:11

answered Jul 17, 2011 at 22:54

Paul

142k28 gold badges285 silver badges272 bronze badges

3 Comments

Jonathan Over a year ago

Works great! Could you explain what was wrong with my search regex?

Paul Over a year ago

Hey, I just made a change to account for the fact that email addresses can contain dots.

Wrikken Over a year ago

You made the change that they can contain consecutive dots.

Collectives™ on Stack Overflow

Preg_replace regex in PHP gives unexpected empty result

2 Answers 2

8 Comments

3 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

8 Comments

3 Comments

Your Answer

Sign up or log in

Post as a guest

Related