my $re = qr/
             <p>         # paragraph-open
             (           # Capture
              (?:         # group
               (?!<\/p>)    # Make sure we aren't at a paragraph-close
               (?!\[tab\])  # and we're not at [tab]
               .            # consume a char
              )*          # any number of times
              \[tab\]     # consume [tab] (hooray!)
              .*?        # anything up to paragraph-close
             )
             <\/p>               /x;
[download]

Thanks a lot for the code and explanation. I think this is exactly what I need. I'll mess with it some more and see if I can break it.

If your HTML is actually XHTML-compliant, you could use XML::Twig to parse it, and then do something like this:

my @tagged_paragraphs = $twig->get_xpath('//p[string()=~/\[tag\]/');
my @texts = map { $_->text() } @tagged_paragraphs;
[download]

my @arr = split /<\/p>/, $string;
@matches = grep { /\[tab\]/ } @arr;
[download]

one extra line but much easier for me. If I was concerned about it looking neat, I'd put the code inside a subroutine called match_tabs().

One minor difference is that splitting this way removes the close-paragraph, but not the open-paragraph, so you need to remove it.

my @arr = grep {/\[tab\]/ and s/^<p>//} split /<\/p>/, $contents;
[download]

If you wanted to retain both tags, you could do so by splitting on a lookbehind expression:

my @arr = grep /\[tab\]/, split /(?<=<\/p>)/, $contents;
[download]

You'd also retain the paragraph-close if you used it for $/ and read the string as a file.

{
  local $/ = '</p>';
  open (STR, '<', \$contents) or die "Opening string: $!\n";
  @arr = grep /^<p>/ && /\[tab\]/, <STR>;
  print "read $_\n" for @arr;
}
[download]

But now I'm just getting silly.

---

Caution: Contents may have been coded under pressure.

You seem to be a bit focussed on using big regexes :P

map { s/<P>// }  grep { /\[tab\]/ } split /<\/p>/, $string;
[download]

___________________
Jeremy
I didn't believe in evil until I dated it.

hmmm, actually makes sense. Luckily most of the time I will be looking exactly for what I originally showed in my example. This should work ok. I'll have to mess with it and see if there's anything I didn't think of that may pop up.

Thanks.

Update: OK, that doesn't fix your real problem. I'll have to think a bit more about it. In this simple case you can just use [^<>] instead of a dot everywhere between <p> and </p>.

The old truth that HTML shouldn't be parsed with regexes still holds.

My problem with the first match is it matches too much...

original:
"<p>no tabs here</p><p>column 1[tab]column 2</p><p>no tabs here</p>"

result of first regex:
no tabs here</p><p>column 1[tab]column 2
[download]

What I want it to match is...

column 1[tab]column 2
[download]

Since that is the set of para's with the string [tab].

My first shot was too fast, here's a working solution:

my $contents = "<p>no tabs here</p><p>column 1[tab]column 2</p><p>no t
+abs here</p>";

if ($contents =~ m{(<p>[^<>]*\[tab\][^<>]*</p>)} ){
    print "Matched '$1'\n;";
}
[download]

I understand your concern of parsing html with regexes, but this string will never be a full set of html. It will have some html (of course with the para's) but using a parser would be overkill for this situation.