Regexp help needed

[Anon]

I've got 2 regexp problems that I need help with. The first is for my discussion forum. I'm using the following regexp code to replace convert urls into clickable links (http://www.php.net/ into <a href="http://www.php.net/">http://www.php.net/</a>):

$mypattern = "http://(([A-Za-z0-9/.?=_~-])*)([a-zA-Z0-9/])";

$it = ereg_replace($mypattern,"<a href=\"\0\">\0</a>",$it);

Where $it is the string containing the text. This works fine, except when $it already contains a URL in the format <a href="http://www.php.net/">http://www.php.net/</a>, as then I end up with:

<a href="<A href="http://www.php.net/"><A href="http://www.php.net/">http://www.php.net/</a></a>">http://www.php.net/</a>

So, I need to modify the pattern so that it won't convert the URL if it is preceeded by href=". I don't know how to do this.

The second problem is (I hope) a little bit simpler. I have some text that I'm parsing and replacing certain words with URLs. For example, "UKF" being replaced by "<a href="http://www.ukfortress.com/">UKF</a>".

The problem here is how to stop words containing "UKF" being included - i.e. I don't want "UKFortress" to be replaced by "<a href="http://www.ukfortress.com/">UKF</a>ortress".

Any help is greatly appreciated.

[Anon]

Tim,

you can solve both these problems by using Perl Compatible Regular Expressions (PCRE). There is an extensive manual on the PHP site (http://www.php.net) or you can use the Perl manpages (http://www.perl.com). The book to read is "Mastering Regular Expressions" by Jeffrey Friedl, O'Reilly.

For your first problem you need to use a method called negative lookbehind, which is an assertion that a sub-pattern does not occour before the pattern being matched.

<code>
$vText = preg_replace(
"#(?<!href=\")http://[-\w\d/.?=_~]*[\w\d/]#i",
"<a href=\"\0\">\0</a>",
$vText);
</code>

Note that this method will not work if there is an anchor in the orginal text containing the text "http://" in the body of the tag. E.g.

<a href="http://php.net">http://php.net</a>

will not work.

For your second problem, you need to use the \b modifier which specifies a word boundary.

<code>
$vText = preg_replace(
"#\bUKF\b#", "whatever", $vText
);
</code>

If you need more help, send me an email.

Jonathan Kahn
[e-mail]

[Anon]

Thanks for the help, however preg_replace was introducted in PHP version 3.0.9 and the server my site's hosted on is running version 3.0.1.9. I'll see if it's possible for it to be upgraded.

Incidently, what's the difference between preg_replace() and ereg_replace() ?

[Anon]

Tim Fountain wrote:

Incidently, what's the difference between >preg_replace() and ereg_replace() ?

preg() use Perl Compatible Regular Expressions (PCRE), while ereg use PHP POSIX-compatible regular expressions. There are a number of features exclusive to PCRE, such as lookahead and lookbehind, which you will not find in the ereg_* functions. However, the word-bounday solution should be possible with ereg_replace().

If you need a powerful regex engine, it's worth upgrading to a newer PHP.