Java regex help

Question

I have a string containing n substrings in the following format which I want to match:

{varName:param1, param2, param2}

Requirements are as follows:

Only the varName (inside the curly brackets) is mandatory
No limit on the number of parameters
No restrictions on whitespace inside curly brackets apart from var and param names which must not contain whitespace

I would like to be able to capture the varName and each of the parameters separately.

I have come up with a regex that is nearly there, but not quite. Any help would be appreciated.

You say you've got a regex which is "nearly there". Perhaps you could post what you've managed so far? — Anon.
– Anon., Commented Feb 8, 2011 at 0:54
Problem with this is that it matches the varName (group1) the first param (group2) and the last param (group3) only — Tom
– Tom, Commented Feb 8, 2011 at 1:07
\{([a-zA-Z]+)(?:\s*:\s*([^,\s]+))?(?:\s*,\s*([^,\s]+))*\s*\} — Tom
– Tom, Commented Feb 8, 2011 at 1:30
As every regex has a fixed number of groups, you can't use it to capture an unlimited number of substrings. You need some sort of loop or split. — maaartinus
– maaartinus, Commented Feb 8, 2011 at 1:31

Brian Agnew · Accepted Answer · 2011-02-08 00:57:02Z

2

I'm wondering whether it would be easier to simply use String.split() judiciously, rather than battle with regexps for the above. The delimiters (colons/whitespace/commas) seem well-defined.

answered Feb 8, 2011 at 0:57

Brian Agnew

273k38 gold badges342 silver badges443 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

Tom Over a year ago

perhaps, though I do need to use a regex to find the whole patterns in the surrounding string, so thought I might as well match the inner parts at the same time

Voo Over a year ago

Agreed, regex really aren't the solution for this problem. Unreadable oneliners (and are there any other regexs?) are horrible to maintain and the syntax is simple enough to be easily parsed in a few lines.

Alan Moore · Accepted Answer · 2011-02-08 01:59:33Z

1

String s = "blah blah\n{varName:param1, param2, param2}\nblah";

Pattern p = Pattern.compile(
  "\\{([a-zA-Z]+)(?:\\s*:\\s*([^,\\s]+(?:\\s*,\\s*[^,\\s]+)*))\\}"
);
Matcher m = p.matcher(s);
if (m.find())
{
  String varName = m.group(1);
  String[] params = m.start(2) != -1
                  ? m.group(2).split("[,\\s]+")
                  : new String[0];

  System.out.printf("var: %s%n", varName);
  for (String param : params)
  {
    System.out.printf("param: %s%n", param);
  }
}

If you're holding out for a way to match the string and break out all the components with one regex, don't bother; this is as good as it gets (unless you switch to Perl 6). As for performance, I wouldn't worry about that until it becomes a problem.

answered Feb 8, 2011 at 1:59

Alan Moore

75.6k13 gold badges109 silver badges161 bronze badges

1 Comment

Tom Over a year ago

Yep, thanks everyone for your help. I see now that I was barking up the wrong tree.

toolkit · Accepted Answer · 2011-02-08 01:06:29Z

1

How about a regex, AND a Scanner ?

import java.util.Scanner;

public class Regex {

  public static void main(String[] args) {  
    String string = "{varName: param1, param2, param2}";   
    Scanner scanner = new Scanner(string);
    scanner.useDelimiter("[\\s{:,}]+");
    System.out.println("varName: " + scanner.next());
    while (scanner.hasNext()) {
      System.out.println("param: " + scanner.next());
    }
  }
}

answered Feb 8, 2011 at 1:06

toolkit

50.4k18 gold badges113 silver badges140 bronze badges

1 Comment

Tom Over a year ago

Interesting that, haven't come across the Scanner class before. Reading up on it now.

Anon. · Accepted Answer · 2011-02-08 01:10:11Z

1

A quick solution in psuedocode:

string.match(/{(\w+):([\w\s,]+)}/);
varName = matches[1];
params = matches[2].split(',');

answered Feb 8, 2011 at 1:10

Anon.

60.2k9 gold badges84 silver badges86 bronze badges

1 Comment

Tom Over a year ago

yes, thought of that and this post was last hurrah before giving up and doing that. Worried a little about the performance of a separate split because it will be quite high volume - though, tbh, have no idea about that.

Jochen Bedersdorfer · Accepted Answer · 2011-02-08 01:01:28Z

0

Post what you have so far. You can test it very easily on this website: http://www.regexplanet.com/simple/index.html

answered Feb 8, 2011 at 1:01

Jochen Bedersdorfer

4,15226 silver badges27 bronze badges

1 Comment

Tom Over a year ago

Added what I have got so far above

Voo · Accepted Answer · 2011-02-08 01:31:44Z

0

Ok I've got a solution in regex that seems to work just fine:

\{\s*([^\{\},\s]+)\s*(?:(?::\s*([^\{\},\s]+)\s*)(?:,\s*([^\{\},\s]+)\s*)*)?\}

Or to even keep the pretense of being able to understand it:

name = [^\{\},\s]+

ws = \s*

\{ws(name)ws(?:(?::ws(name)ws)(?:,ws(name)ws)*)?\}

I wouldn't recommend it but short testing seems to indicate that it works - nice brain teaser for 3am in the morning ;)

PS: If you're comparing the split solution to this or something similar I'd be interested in hearing if there were any performance differences - I don't think regex would be especially performant.

answered Feb 8, 2011 at 1:31

Voo

30.4k13 gold badges91 silver badges165 bronze badges

Collectives™ on Stack Overflow

Java regex help

6 Answers 6

2 Comments

1 Comment

1 Comment

1 Comment

1 Comment

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

6 Answers 6

2 Comments

1 Comment

1 Comment

1 Comment

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Related