如何在Java中实现unicode感知选择排序算法

Question

I study selection sort algorithm. Consider this implementation:

http://algs4.cs.princeton.edu/21elementary/Selection.java.html

I have a text file which has unicode words like that.

$ more words.txt
şeftali içel ırak üzüm uzun çorba çimen ufuk

When I run the program it doesn't sorts unicode characters correctly.

$ java-algs4 Selection < words.txt
içel
ufuk
uzun
çimen
çorba
üzüm
ırak
şeftali

My first attempt was to use a collator.

import java.util.*;
import java.text.*;

public class StringSorter
{
    public static void sortStrings(Collator c, String[] words)
    {
        String tmp;

        for (int i = 0; i < words.length; ++i)
        {
            for (int j = 0; j < words.length; ++j)
            {
                if (c.compare(words[i], words[j]) < 0)
                {
                    tmp = words[i];
                    words[i] = words[j];
                    words[j] = tmp;
                }
            }
        }
    }

    public static void printStrings(String[] words)
    {
        for (int i = 0; i < words.length; ++i)
        {
            System.out.println(words[i]);
        }
    }

    public static void main(String[] args)
    {
        Collator tr_TRCollator = Collator.getInstance(new Locale("tr", "TR"));

        String[] words = {"şeftali", "içel", "ırak", "üzüm", "uzun", "çorba", "çimen", "ufuk"};
        sortStrings(tr_TRCollator, words);
        printStrings(words);
    }

}

This program sorts words correctly as expected.

$ java-algs4 StringSorter
çimen
çorba
ırak
içel
şeftali
ufuk
uzun
üzüm

My question is how should we implement unicode aware selection sort algorithm in Java?

Also Selection.class sort method takes a Comparator object as a second parameter. Is it possible to write our own implementation of Comparator interface so that it should be able to sort unicode elements correctly.

 public static void sort(Object[] a, Comparator c)

Any help would be appreciated. Thanks..

Answer 1

Collator类实现Comparator接口，因此您可以将tr_TRCollator传递给Selection.sort作为第二个参数。

Answer 2

You could Normalize the Strings and do a unicode comparison if they other wise match.

String[] words = "şeftali içel ırak üzüm uzun çorba çimen ufuk".split(" ");
Arrays.sort(words, Comparator.comparing((String w) -> 
                                        Normalizer.normalize(w, Normalizer.Form.NFD))
                             .thenComparing(Comparator.naturalOrder()));
Stream.of(words).forEach(System.out::println);

prints

çimen
çorba
içel
şeftali
ufuk
uzun
üzüm
ırak

This is close but it doesn't consider ı to be like i

Answer 3

The important point is that your 2nd example uses locale settings. Sort order for strings is locale dependent and has nothing to do with the unicode codepoints of the characters. Even countries that use the same language, say Austria, Germany and Switzerland have subtle differences in string sorting order.