Deleting Everything expect in Parentheses

Sops21

I have text in a string. The text looks like this:

sometext here (important information here)

I need a way to pull the text between the parentheses out and store in in its own string. Any ideas?

dagon

until some one posts something better:

preg_match('#\((.*)\)#', $string, $matches);
print_r($matches);

Sops21

sick that did the trick. Thanks for the help.

bradgrafelman

Just to go a different route and use a one-liner:

$test = 'sometext here (important information here) and here';

$test = preg_replace('/^.*\((.*)\).*$/', '$1', $test);

echo $test; // important information here

Either way, don't forget to mark this thread resolved (if it is).

nrg_alpha

try:

$str = 'sometext here (important information here)';
preg_match('#\(([^)]+)\)#', $str, $match);
echo $match[1];

output:

important information here

Just a note on using .* In general, it is frowned upon to use this, as it tends to be potentially inefficient. You can read up about this here in a previous post I made.

Not included in that link is another reason why . isn't the best way to do things which is a matter of accuracy (but I don't suspect that would be the case in this situation). Since is a greedy quantifier, it will keep matching up to the last instance of whatever you are seeking, which could in turn yield to improper results... so if the string had something like:

$str = 'sometext here (important information here) some more text (yep, another set of parenthesis).';

You can start to guess what happens when throwing around .* carelessly.

Instead, making use of negated character classes (or making . lazy: .?) is prefered.

bradgrafelman

Gah, I've heard plenty of comments about using a negated character class as opposed to '.' in regexp's, but the '.' is an old habit and as the saying goes, old habits are hard to break.

Also, nice catch on the lazy quantifier - not entirely sure why I didn't think of that before. Guess that goes to show you the importance of concepts such as unit testing to cover a variety of cases. Heh. :o

nrg_alpha

Indeed old habits do die hard.

I learned regex by reading the Mastering Regular Expressions book (which issues the warning of using .) Good read if people are truly interested in really getting on the ball with regex. It does a fantastic job in revealing how regex engines think*.

Weedpacket

bradgrafelman wrote:
Also, nice catch on the lazy quantifier - not entirely sure why I didn't think of that before. Guess that goes to show you the importance of concepts such as unit testing to cover a variety of cases. Heh.

But for that case to be properly addressed it's necessary to know what to do about multiple parenthesised phrases. Is the first one the one that's actually wanted? Or do you preg_match_all() and let the caller sort 'em out (Note that you've just changed the return type)?

nrg_alpha

The sample with two sets of parenthesis was just to demonstrate the ill effects of using .* instead of a negated character class / lazy quantifier. It was a hypothetical circumstance.

Indeed, if the OP runs into situations where there might be multiple instances of content encased within parenthesis, then preg_match_all with perhaps some additional screening methods involving those captures might be necessary (unless only the first set is the only desired capture). I am assuming that none of this is the case however (then again, we know what they say about assuming...)

The point was more geared towards potential regex engine performance / accuracy issues. I was thinking (assuming?) in the context of having those single set of parenthesis nested within a much larger block of text and not at the end of a simple single lined string.

cupboy

I tried writing a C# version of the code in post #4

eng = Regex.Replace(x, @"/^.((.)).*$/", "$1");

And it does nothing. Just returns the value of x as is.