Hash marks in Regular Expressions

Question

Given this example code, what do you expect as output?import java.util.regex.Matcher;
import java.util.regex.Pattern;
public class HashTest
   * @param args
  public static void main(String[] args)
    String testChar = "#";
    String testInput = "this string is # split by a hash";
    String[] stringArray = testInput.split( testChar );
    for (String s : stringArray)
      System.out.println( s );
    Pattern p = Pattern.compile( testChar );
    Matcher m = p.matcher( testInput );
    if ( m.lookingAt() )
      System.out.println( m.groupCount() );
    p = Pattern.compile( "(#)(.*)" );
    m = p.matcher( testInput );
    if ( m.lookingAt() )
      for (int i = 0; i <= m.groupCount(); i++)
        System.out.println( m.group( i ) );
}The output I get is
this string is
split by a hash
replacing the last pattern with
p = Pattern.compile( "(.*?)(#)(.*)" );I get
this string is
split by a hash
this string is # split by a hash
this string is
split by a hash
This is very unexpected behavior and if anyone could come up with a convincing solution, I'd appreciate it.

Answer 1

As for matches(), I think it was because they were also adding regex-related wrapper methods to String. Part of the contract for the find() method is that, if you call it multiple times without resetting the matcher, each match attempt will start searching at the position where the last match ended. But that wouldn't work in String, because there was no way to ensure that the variable it's being called on this time still refers the same string it referred to the last time. So, if it can only be meaninfully called once anyway, they might as well have it match the whole string. That's my guess.
I can't even guess what they were thinking about with lookingAt(). I found it perfectly useless until JDK 1.5, when they added the "regions" API and the usePattern(Pattern) method to Matcher. Now it's useful for scanning text with multiple regexes, which is how they use it in java.util.Scanner, and how I use it in my syntax-highlighting editor. Before, I had to prepend "\G" to every regex and, for each match attempt, create a new Matcher and call its find(int) method--yech!.