Name: bsC130419 Date: 06/13/2001


java version "1.4.0-beta"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.4.0-beta-b65)
Java HotSpot(TM) Client VM (build 1.4.0-beta-b65, mixed mode)


As the sample code output below shows, any text that is
matched by an independent group or a conditional group is
automatically captured. I think this is incorrect, for
two reasons. First, it isn't how these constructs work in
Perl, where they first appeared, so anyone who is already
familiar with them will be tripped up by this unexpected
behavior. Second, while the regex author can choose to
capture any part of an expression by enclosing it in
parentheses, there is no way to tell these constructs NOT
to capture; you're taking away some of the author's control
over the matching process, for no good reason.

----------------------- sample code ---------------------------

import java.util.regex.*;

public class CaptureTest
{
  Pattern pattern;
  Matcher matcher;
  
  public CaptureTest()
  {
    // independent group
    tryit("x+(?>y+)z+");

    // conditional
    tryit("x+(?(?=y)y+|z+)");

    // non-capturing group, for comparison
    tryit("x+(?:y+)z+");
  }
  
  public void tryit(String regex)
  {
    System.out.println();
    System.out.println(regex);
    try
    {
      pattern = Pattern.compile(regex);
      matcher = pattern.matcher("xxxyyyzzz");
      if (matcher.find())
      {
        System.out.println(matcher.group(1));
      }
    }
    catch (Exception ex)
    {
      System.out.println("error: " + ex.getMessage());
    }
  }

  public static void main(String[] argv)
  {
    new CaptureTest();
  }
}

------------------------- output -----------------------------
$ java CaptureTest

x+(?>y+)z+
yyy

x+(?(?=y)y+|z+)
yyy

x+(?:y+)z+
error: No group 1
(Review ID: 126515)
======================================================================