REGEX: question about finding Overlapping matches using regular expressions

Question

I have the following problem.
Say for my pattern I use:
Pattern pattern = Pattern.compile("AAA");
Matcher matcher = pattern.matcher("AAAAAA");when I run a loop
while (matcher.find())
System.out.println("Match Found: "+matcher.start()+" "+matcher.end());I get 2 Hits shown in the following output:
Match Found: 0 3
Match Found: 3 6
therefore the regex is seeing the first AAA then the second AAA.
I want it to find the other AAA's in there that are overlapping the other two finds i.e. I want the output to find
AAA from 0 to 3
AAA from 1 to 4
AAA from 2 to 5 and finally
AAA from 3 to 6
thereby including the overlapping finds.
How can I do this using regex? what am I missing that prevents the overlapping matches to be found? Do I need a quantifier?
Thanks for the help!

Answer 1

While the solutions above work fine with the given input, they don't really find all overlapping matches. They just find the longest possible match at each start position. Here's a more thorough approach:import java.util.*;
import java.util.regex.*;
public class Test
  public static List<String> matchAllWays(String rgx, String str)
    Pattern p = Pattern.compile(rgx);
    Matcher m = p.matcher(str);
    List<String> result = new ArrayList<String>();
    int len = str.length();
    int start = 0;
    int end = len;
    while (start < len && m.region(start, len).find())
      start = m.start();
      do
        result.add(m.group());
        end = m.end() - 1;
      } while (end > start && m.region(start, end).find());
      start++;
    return result;
  public static void main(String[] args)
    List<String> matches = matchAllWays("a.*a", "abracadabra");
    System.out.println(matches);
}This approach requires JDK 1.5 or later; that's when the regions API was added to Matcher.