Collator compares strings weird

Question

I have a collection of strings and need to sort it. I'm using the Collator. But the output is weird.

final Collator collator = Collator.getInstance(Locale.US);

List<String> data = new ArrayList<String>();

data.add("1Z5800701_AB");
data.add("1Z5800701_AC");
data.add("1Z5800701-A");
data.add("1Z5800701 A");
data.add("1Z5800701B");
data.add("1Z5800701A");
data.add("1Z5800701 - A");

Collections.sort(data, new Comparator<String>() {

    @Override
    public int compare(String o1, String o2) {
        return collator.compare(o1, o2);
    }
});

for (String s : data) {
    System.out.println(s);
}

And the output is:

1Z5800701_AB
1Z5800701_AC
1Z5800701A
1Z5800701 A
1Z5800701 - A
1Z5800701-A
1Z5800701B

The last one string '1Z5800701B' should be after '1Z5800701A'. What am I missing here?

Answer 1

It's a matter of the locale used, you can reproduce the same behavior in the bash shell with LC_ALL=en_US sort. The point is that the "word separators" are treated differently from "word characters" in this locale (i.e. you can't always say that character X sorts before or after character B - it depends on context). The result is if you have 1Z5800701 <optional separators> A, it sorts before 1Z5800701 <optional separators> B, that's why 1Z5800701B comes after all combinations where the A comes after the digits, optionally separated by "separators". You can also see some more examples of "not obvious" orderings in this Wikipedia articles