Can a non-empty string have a hashcode of zero?

2019-02-07 21:28发布

问题:

By "non-empty", I mean in this question a string which contains at least one non-zero character.

For reference, here's the hashCode implementation :

1493    public int hashCode() {
1494        int h = hash;
1495        if (h == 0) {
1496            int off = offset;
1497            char val[] = value;
1498            int len = count;
1499
1500            for (int i = 0; i < len; i++) {
1501                h = 31*h + val[off++];
1502            }
1503            hash = h;
1504        }
1505        return h;
1506    }

and the algorithm is specified in the documentation.

Before an integer overflow occurs, the answer is easy: it's no. But what I'd like to know is if, due to integer overflow, it's possible for a non-empty string to have a hashcode of zero? Can you construct one?

What I'm looking for would ideally be a mathematical demonstration (or a link to one) or a construction algorithm.

回答1:

Sure. The string f5a5a608 for example has a hashcode of zero.

I found that through a simple brute force search:

public static void main(String[] args){
    long i = 0;
    loop: while(true){
        String s = Long.toHexString(i);
        if(s.hashCode() == 0){
            System.out.println("Found: '"+s+"'");
            break loop;
        }
        if(i % 1000000==0){
            System.out.println("checked: "+i);              
        }
        i++;
    }       
}

Edit: Joseph Darcy, who worked on the JVM, even wrote a program that can construct a string with a given hashcode (to test the implementation of Strings in switch/case statements) by basically running the hash algorithm in reverse.



回答2:

just be care of that int h;. It may overflow, every string that satisfy h % 2^31 == 0 may lead to this.

public class HelloWorld {
    public static void main(String []args) {
       System.out.println("\u0001!qbygvW".hashCode());
        System.out.println("9 $Ql(0".hashCode());
        System.out.println(" #t(}lrl".hashCode());
        System.out.println(" !!#jbw}a".hashCode());
        System.out.println(" !!#jbw|||".hashCode());
        System.out.println(" !!!!Se|aaJ".hashCode());
        System.out.println(" !!!!\"xurlls".hashCode());
    }
}

A lot of strings...