Regex (regular expressions), replace the second occurence in javascript

Question

This is an example of the string that's being worked with:

xxxxxx[xxxxxx][7][xxxxxx][9][xxxxxx]

I'm having a little trouble matching the second occurrence of a match, I want to return the 2nd square brackets with a number inside. I have some regex finding the first square backets with numbers in a string:

\[+[0-9]+\]

This returns [7], however I want to return [9].

I'm using Javascript's replace function, the following regex matches the second occurrence (the [9]) in regex testeing apps, however it isn't replaced correctly in the Javascript replace function:

(?:.*?(\[+[0-9]+\])){2}

My question is how do I use the above regex to replace the [9] in Javasctipt or is there another regex that matches the second occurrence of a number in square brackets.

Cheers!

Solution 1

If xxx is just any string, and not necessarily a number, then this might be what you want:

(\[[0-9]+\]\[.*?\])\[([0-9]+)\]

This looks for the second number in []. Replace it with $1[<replacement>]. Play with it on rubular.

Your regular expression fails to work as intended because groups followed by + only end up holding the last [xxx].

Solution 2

Try

result = subject.replace(/(\[\d\]\[[^\]]+\])\[\d\]/, "$1[replace]");

As a commented regex:

(       # capture the following in backref 1:
\[\d\]  # first occurrence of [digit]
\[      # [
 [^\]]+ # any sequence of characters except ]
\]      # ]
)       # end of capturing group
\[\d\]  # match the second occurence of [digit]

If the number of [xxx] groups between the first and second [digit] group is variable, then use

result = subject.replace(/(\[\d\](?:\[[^\]]+\])*?)\[\d\]/, "$1[replace]");

By surrounding the part that matches the [xxx] groups with (non-capturing) parentheses and the lazy quantifier *? I'm asking the regex engine to match as few of those groups as possible, but as many as necessary so the next group is a [digit] group.

Comments

[Anon.]

Why are you using {3} as a quantifier? This doesn't appear to be a "there will always be three tags before the one we want, and one after" situation, shouldn't you base the regex on the actual requirements stated by the asker? Personally, I would use /(\[\d+\](?:\[.*?\D.*?\])*)(\[\d+\])/\1[replacement]/

[moinudin]

@Anon We don't know that xxx isn't a number. I assumed it was, but maybe I'm wrong.

[Hemlock]

Not sure if this is important, but this fails on [1][2][3] - replaces the last [3]. Maybe the OP should provide some tests...

[moinudin]

I've added another regex which will replace the second [<number>].

[Anon.]

@marcog: The asker mentioned he wanted "the second match", so I think we can presume that [xxx] doesn't match the same thing as [9]. Though it is unclear whether [xxx] can be things like [123], or whether it must have something that isn't a digit in it.

[moinudin]

@Anon Do you agree with my new regex?

[Anon.]

@marcog: You're missing a backslash near the end of the first group, but other than that it looks good.

[moinudin]

@Anon Technically it's not required, but I'll add it in.