ansaurus

Question

Matching inner pattern an unlimited amount of times within outer pattern

Answer 1

+2 A:

Try this:

INDICATOR(\s+([a-z0-9]+))+

Here the repeating pattern is wrapped in a group and quantified using + to allow one or more repetitions of the expression inside the group. But you won’t get every match of the inner group with this but only the last match (or to be more specific: it depends on the implementation you’re using).

Gumbo 2010-07-31 15:51:44

@Gumbo: Basically I just need to match the full string and then I will reparse it using only the inner regex to get the data out. This initial regex is for validation.

James 2010-07-31 15:57:21

@James: Then I suggest using so called non-capturing groups `(?:…)` instead of “normal” groups to avoid the costs of storing the captured string.

Gumbo 2010-07-31 16:15:37

@Gumbo: Does this give me the advantage of not having to re-parse the string I can just iterate over the groups?

James 2010-07-31 16:23:08

No you will always have to re-parse the group if you want the individual values. Regex cannot recursively capture values in that matter. If you use a quantifier on a capture group, the group only contains the last value matched.

Cags 2010-07-31 16:55:34

Say the pattern might have a different separator instead of a space i.e. a comma or a hyphen. Would I just replace the `\s+` with something like `[(\s+),-]` ?

James 2010-07-31 17:00:41

@James: If you would allow one or more spaces but only one single comma/hyphen, you would need `(\s+|[,-])`. Otherwise, if it is always just one character, `[\s,-]`.

Gumbo 2010-07-31 17:02:55

@Cags: There are implementations that store each single match of a repetition (I think [.NET’s CaptureCollection](http://msdn.microsoft.com/en-us/library/system.text.regularexpressions.capturecollection\(v=VS.71\).aspx) provides such a behavior).

Gumbo 2010-07-31 17:06:39

@Gumbo , ah true, for some reason I'd got it in my head he was using PHP.

Cags 2010-07-31 17:31:50

James 2010-07-31 17:52:17

I ended up with the following `INDICATOR((\s+)?([,-])?(\s+)?([a-z0-9]+))+`. This would allow all variations of `INDICATOR AA, BB`, `INDICATOR AA,BB`, `INDICATOR AABB`, `INDICATOR AA , BB`, `INDICATOR AA BB` which keeps my pattern flexible. I need to try support all these different situations as I can't guarentee how the users will send the information in a strict format.

James 2010-07-31 17:55:00

@Gumbo: Just one more thing (not really a necessity but would be nice). Is it possible to restrict the amount of patterns I match? i.e instead of matching all, can I restrict it to only match the first 20?

James 2010-07-31 18:23:50

@James: Yes, you can use the quantifier `{n,m}` to allow n to m repetitions. So try `INDICATOR(\s*[,-\s]\s*[a-z0-9]+){1,20}`.

Gumbo 2010-07-31 18:32:52

@Gumbo: Excellent, thanks for your time!

James 2010-07-31 21:58:58

ansaurus

tags:

views:

answers:

Matching inner pattern an unlimited amount of times within outer pattern

related questions