How to perform string Diffs in Java?

Viewed 81736

I need to perform Diffs between Java strings. I would like to be able to rebuild a string using the original string and diff versions. Has anyone done this in Java? What library do you use?

String a1; // This can be a long text
String a2; // ej. above text with spelling corrections
String a3; // ej. above text with spelling corrections and an additional sentence

Diff diff = new Diff();
String differences_a1_a2 = Diff.getDifferences(a,changed_a);
String differences_a2_a3 = Diff.getDifferences(a,changed_a);    
String[] diffs = new String[]{a,differences_a1_a2,differences_a2_a3};
String new_a3 = Diff.build(diffs);
a3.equals(new_a3); // this is true
8 Answers

Apache Commons has String diff

org.apache.commons.lang.StringUtils

StringUtils.difference("foobar", "foo");

As Torsten Says you can use

org.apache.commons.lang.StringUtils;

System.err.println(StringUtils.getLevenshteinDistance("foobar", "bar"));

If you need to deal with differences between big amounts of data and have the differences efficiently compressed, you could try a Java implementation of xdelta, which in turn implements RFC 3284 (VCDIFF) for binary diffs (should work with strings too).

Use the Levenshtein distance and extract the edit logs from the matrix the algorithm builds up. The Wikipedia article links to a couple of implementations, I'm sure there's a Java implementation among in.

Levenshtein is a special case of the Longest Common Subsequence algorithm, you might also want to have a look at that.

Apache Commons Text now has StringsComparator:

StringsComparator c = new StringsComparator(s1, s2);
c.getScript().visit(new CommandVisitor<Character>() {

    @Override
    public void visitKeepCommand(Character object) {
        System.out.println("k: " + object);
    }

    @Override
    public void visitInsertCommand(Character object) {
        System.out.println("i: " + object);
    }

    @Override
    public void visitDeleteCommand(Character object) {
        System.out.println("d: " + object);
    }
});
Related