Help tweaking my Regex

MartinKool

To be honest, today is the first day I ever touch regular expressions. I'm creating one that will add a target attribute to external links. What I have so far works, but needs a little extra:

$string:

My name is Martin. This is 1 link: <a href="http://www.seacrest.nl/test/">Test</a>. Here's another link <A HREF="../../test.html">Internal</a> and yet another <A href="http://www.testsite.nl/" target="">External</a>

Regex part:

eregi_replace("<a href=\"http://([^{\"]+)\"([^{>]*)>([^<]+)</a>",}} "<a href=\"http://\1\" target=\"new\">\3</a>", $string);

What it does so far is add a target to external links, provided they start with http://. However many internal links also start with http:// and then point to www.seacrest.nl/

My question is what do I add to exclude links that have http://www.seacrest.nl in them?

I've tried:

eregi_replace("<a href=\"http://([^{(www.seacrest.nl)][^{\"]+)\"([^{>]*)>([^<]+)</a>",}}} "<a href=\"http://\1\" target=\"new\">\3</a>", $string);

But that's just wrong and I know it. Can anyone help me out?

Thanks in advance.

xblue

if you'd switch to perl-style regexp, you could employ a negative lookahead for this.

syntax:
(?!www.seacrest.nl)
(will make sure that the following sequence is different from www.seacrest.nl without "consuming" it)

with ereg-functions it is probably much harder to accomplish.

MartinKool

Hmm unfortunately the perl style regex puzzles me. I'll give it a try though. Any chance you could post the perl equivalent of my snippet there?

xblue

don't give up, the basics are pretty similar.

first difference:
you'll need to put your pattern into delimiters. this may be almost every special character. forward slash is fairly common, but can be inconvenient when the pattern contains slashes as well.

the first try should not be hard to set up at all, replace "eregi" with "preg", leave your pattern as it is but add e.g. § and § at the beginning and end, and insert "i" at the end after the delimiter, too (this makes it match case-insensitive, as the "i" in the eregi function name). then add the lookahead and see what we get.

MartinKool

I see, it seems to be working. It were those delimiters that bugged me. What are they for?

With preg_replace and your suggestion for the exception everything works as planned. Thanks alot.

xblue

the one at the end is needed to separate the pattern from the modifiers (that i you added for case-insensitive matching).

the one at the beginning is needed because you are allowed to choose the delimiter yourself - the regex machine takes the first one to tell which character will indicate the end of the pattern.

once you think about it like that, I found it fairly convincing. Hope you agree.