Ampersand regexp? 1

[Kirsle]

A problem I run into sometimes on my websites is, they fail the W3C HTML Validator on the grounds that I tend to use a lot of &'s in hyperlinks without using &

i.e.

<a href="file.cgi?color=blue&font=Arial">

they want it to be:

<a href="file.cgi?color=blue[COLOR=red]&amp;[/color]font=Arial">

And by the time I think to validate it, there are many pages to go through (imho, it shouldn't complain when an & is used within an href), so I decided to try using a regexp on the content before it gets printed to the browser (this is a CGI website)

Does anybody know of a simple regexp? The one I had to settle with is kinda blech:

$cuvou->{template} =~ s/&#/__amp__number__/ig;
$cuvou->{template} =~ s/</__amp__lt__/ig;
$cuvou->{template} =~ s/>/__amp__gt__/ig;
$cuvou->{template} =~ s/&/__amp__amp__/ig;
$cuvou->{template} =~ s/"/__amp__quot__/ig;
$cuvou->{template} =~ s/'/__amp__apos__/ig;
$cuvou->{template} =~ s/&/&/ig;
$cuvou->{template} =~ s/__amp__number__/&#/ig;
$cuvou->{template} =~ s/__amp__lt__/</ig;
$cuvou->{template} =~ s/__amp__gt__/>/ig;
$cuvou->{template} =~ s/__amp__amp__/&/ig;
$cuvou->{template} =~ s/__amp__quot__/"/ig;
$cuvou->{template} =~ s/__amp__apos__/'/ig;

Thanks in advance.

-------------
Cuvou.com | The NEW Kirsle.net

 

[Kirsle]

I revised that one with a three-liner:

$cuvou->{template} =~ s/&(amp|lt|gt|copy|#\d+);/__goodamp__$1__/ig;
$cuvou->{template} =~ s/&/&/ig;
$cuvou->{template} =~ s/__goodamp__(.+?)__/&$1;/ig;

But there's gotta be a good one-liner out there. :|

-------------
Cuvou.com | The NEW Kirsle.net

 

[MillerH]

Not to rain in on your regex parade (I love such problems) but what about  's and the literally hundreds of other HTML Character Entities. What about JavaScript code and the need for AND logic?

It is not going to be possible to create a series, let alone a single regex to fix this particular problem. Instead, if you should focus your attack. Use HTML:arser or some other parsing module to apply a regex to link urls only.

However, if you insist on wanting a regex, just use a negative lookahead assertion.

my $entities = join '|', qw(amp lt gt copy #\d+ nbsp);
$cuvou->{template} =~ s/&(?!(?:$entities);)/&/ig;

 

[ishnid]

Or just use semicolons instead of ampersands for your URLs. Assuming your CGI script is written correctly, it should support that just as well, without the encoding issues.

 

[KevinADC]

Yea, I tend to use semi-colons myself these days for similar reasons and because its the new-and-improved way of delimiting URI data strings.

------------------------------------------
- Kevin, perl coder unexceptional!