Select the inner brace only if it has outer brace

Question

This is my text:

This [is] some [d[um]my] text. How to [se[le]ct i[nn]er b]race wi[th[out s]ele[ct]ing th]e outer b[race [in] a tex]t

The regex for the above text which is required must be highlighted as below

This [is] s[o]me [d[um]my] text. How to [se[le]ct i[nn]er b]race wi[th[out s]ele[ct]ing th]e outer b[race [in] a tex]t

As you can see, the regex must highlight only the braces which has parent braces. The braces which do not have parent braces must not be selected.

For example [is] and s[o]me has no parent brace hence they must not be highlighted. But [d[um]my], [se[le]ct i[nn]er b]race have parent brace hence the brackets along with the text inside must be selected.

I have tried the below PCRE regex:

\[[^\[]+?]

https://regex101.com/r/xR0wM3/12

But it is also highlighting the braces which do not have outer brace. That is the only issue to be solved, all the other highlighting of text is working perfectly. In the example provided the change must be made such that it must not select the braces which do not have parent brace. i.e, in the example [is] is being selecting which is out of scope of the requirement. If this is solved then my requirement is completed.

Quote the regex you've tried **in** your question, not just linked. Links rot, and people shouldn't have to go off-site to see what you've tried. (Not that a regex101 link isn't a useful *addition*.) *Edit*: I've done it for you this time. — T.J. Crowder, Aug 17 '15 at 09:59
You've tagged `javascript` and `php`. You need to choose, their regular expression engines are similar, but different, and one of the differences (PHP has lookbehind, JavaScript doesn't) may well be relevant. Please also don't tag wildly: `performance` is clearly not a relevant tag, nor is `html`. (I've removed them.) — T.J. Crowder, Aug 17 '15 at 10:00
I would do it in two steps: match parent braces then match inner braces. Now the question is: are you using PHP or JS? If you're using PHP you might use a recursive pattern otherwise you will need to write a parser in JS. In either cases you could write a parser. Have fun. — HamZa, Aug 17 '15 at 10:08

Wiktor Stribiżew · Accepted Answer · 2015-08-17T11:26:26.587

3

Keeping in mind that

There will be only one parent brace i.e only one nested level.

You can use the following regex in PHP:

(?:\[|(?!^)\G).*?(\[[^\[\]]*\])

See demo

The (?:\[|(?!^)\G) part will make sure we only match [...] that are inside another pair of [...].

A bit more optimized variant without capture group and using \K (that omits the whole initial part of the match):

(?:\[|(?!^)\G)[^\[\]]*\K\[[^\[\]]*\]

See demo 2

An approach for JavaScript includes 2 steps:

We extract those substrings with parent parentheses with var re = /[^\[]+(\[(?:[^\[\]]|\[[^\[\]]*\])*\])/g;
Then, we extract all inner [...] substrings from those chunks with rx = /\[[^\[\]]+\](?=(?:[^\[\]]*(?:\[[^\[\]]*\][^\[\]]*)*\]))/g;.

var re = /[^\[]+(\[(?:[^\[\]]|\[[^\[\]]*\])*\])/g; 
var str = 'This [is] some [d[um]my] text. How to [se[le]ct i[nn]er b]race wi[th[out s]ele[ct]ing th]e outer b[race [in] a tex]t';
var m;
 
while ((m = re.exec(str)) !== null) {
    if (m.index === re.lastIndex) {
        re.lastIndex++;
    }
    rx = /\[[^\[\]]+\](?=(?:[^\[\]]*(?:\[[^\[\]]*\][^\[\]]*)*\]))/g;
    var n;
    while ((n = rx.exec(m[1])) !== null) {
          if (n.index === rx.lastIndex) {
             rx.lastIndex++;
        }
        document.getElementById("r").innerHTML += n[0]+"<br/>";
    }
}

<div id="r"/>

A couple of words about the second regex: (?=(?:[^\[\]]*(?:\[[^\[\]]*\][^\[\]]*)*\])) look-ahead is making sure there are either characters other than [ and ] ([^\[\]]*), or [...] substrings (\[[^\[\]]*\]) and then a closing ] should follow. It could be written as (?=(?:[^\[\]]|\[[^\[\]]*\])*\]), but an unwrapped version I am using is more efficient (though looks very untidy. This is JS, sorry.)

edited Aug 17 '15 at 11:26

answered Aug 17 '15 at 10:13

Wiktor Stribiżew

607,720
39
448
563

Nothing extraordinary, here is [a post explaining `\G` operator](http://stackoverflow.com/questions/21971701/when-is-g-useful-application-in-a-regex). The main point is that it helps find texts inside some delimiters. This operator is not compatible with JS, BTW. – Wiktor Stribiżew Aug 17 '15 at 10:21
@stribizhev This is working fine in PHP platform. But not working in javascript platform – Shiva Prasad Aug 17 '15 at 10:55
Yes, as I said that is only for PHP. A pure 1-regex approach only works with PHP. Do you want me to also devise a multiple step approach in JS then? It might take some time. – Wiktor Stribiżew Aug 17 '15 at 10:57
Yes please. The same regex functionality in javascript is required – Shiva Prasad Aug 17 '15 at 10:58
Do you always have just 1 nested level of brackets? Or can there be more? – Wiktor Stribiżew Aug 17 '15 at 11:03
There will be only one parent brace i.e only one nested level. But multiple child braces inside a parent brace like [se[le]ct i[nn]er b]race, with 2 child brace inside a single parent brace – Shiva Prasad Aug 17 '15 at 11:07
I posted a JS suggestion. – Wiktor Stribiżew Aug 17 '15 at 11:20
You are awesome. Got it. Thanks a lot :) – Shiva Prasad Aug 17 '15 at 11:22

Select the inner brace only if it has outer brace

1 Answers1