[RESOLVED] Regex Help

areid

Hey everyone. I need a little help with a regular expression that I have been trying to figure out.

I want to match everything between : and ; that can include a-z0-9#\s but also must include at least one +,-,* or /.

: #ffffff + 2; MATCH
: #ffffff + #000000 3; MATCH
: #ffffff + #000000; MATCH
: 2px + 2 3; MATCH

: 2px; NO MATCH
: #ffffff; NO MATCH

Thanks for any help!

NogDog

Untested:

preg_match('@(?<=:)[a-z0-9#\s]*[-*+/][a-z0-9#\s]*(?=;)@', $text, $matches);

Caveat, this will match on something like ":+;". If something must appear on one either or both sides of that special character, change the "*" to a "+" on the character class on that side.

areid

Thanks for the reply, but it didn't work.

Yes there will be text on both sides of the : and ; I want to parse some CSS.

so, i would need to match just the width attribute and also any other variations explained in my original post.

#test {	width: 2px + 2 + 3; color:#ffffff; }

Again, thanks for your reply and any help!

CoderDan

I think all you need to do is add in your 'required chars' to the end match, ie, instead of:
(?<=🙂[a-z0-9#\s][-+/][a-z0-9#\s](?=😉
use:
(?<=🙂[a-z0-9#\s][-+/][-+/a-z0-9#\s]*(?=😉

nrg_alpha

I do wonder why the look ahead/behind assertions though. Perhaps I'm missing something?

Perhaps something along the lines of:

$css = '#test {	width: 2px + 2 + 3; color:#ffffff; }';
preg_match_all('~:([a-z0-9#\s]+[-*+/][-*+/a-z0-9#\s]+);~', $css, $matches);
$matches[1] = array_map('trim', $matches[1]); // trim off initial / trailing spaces...
echo '<pre>'.print_r($matches[1], true);

I use preg_match_all in the even you want all instances of what you seek.

In either case, is that #test line a css expression? If so, they are discouraged.

EDIT - Given that the last css attribute between { and } doesn't have to contain a semi-colon, we could modify the pattern to reflect this:

preg_match_all('~:([a-z0-9#\s]+[-*+/][-*+/a-z0-9#\s]+)(?:;|\R*})~', $css, $matches);

NogDog

nrg_alpha;10924216 wrote:
I do wonder why the look ahead/behind assertions though. Perhaps I'm missing something?
...

The OP said s/he wanted everything between the ":" and ";", so by using assertions they would not be included in the actual match, giving you everything between, but not including those characters.

nrg_alpha

NogDog;10924224 wrote:
The OP said s/he wanted everything between the ":" and ";", so by using assertions they would not be included in the actual match, giving you everything between, but not including those characters.

Right. I understand. I don't think I phrased my question as well as I could have. I guess what I meant was why use assertions when one could include : and ; in the pattern, but just not involve them in a capture?

I suppose the draw back to this method is that it involves a capture (where as the assertion method doesn't). Granted, speed might be an advantage to the capture method (I can't be sure though, as I never tested it). I always figured assertions is costly in performance. but it might not be an issue though...

areid

CoderDan;10924205 wrote:
I think all you need to do is add in your 'required chars' to the end match, ie, instead of:
(?<=🙂[a-z0-9#\s][-+/][a-z0-9#\s](?=😉
use:
(?<=🙂[a-z0-9#\s][-+/][-+/a-z0-9#\s]*(?=😉

Thank you. Worked perfectly!

NogDog

nrg_alpha;10924250 wrote:
Right. I understand. I don't think I phrased my question as well as I could have. I guess what I meant was why use assertions when one could include : and ; in the pattern, but just not involve them in a capture?

I suppose the draw back to this method is that it involves a capture (where as the assertion method doesn't). Granted, speed might be an advantage to the capture method (I can't be sure though, as I never tested it). I always figured assertions is costly in performance. but it might not be an issue though...

Off-hand I'm not aware of either way being faster. I just figure why capture something you don't need? With the assertion it's one less sub-pattern and therfore one less array element in which something needs to be stored. But whichever way you get to work is fine. As the XP developers say, code whatever works, and leave optimization for last. 🙂