jubeat++での日本語ソート実装

HiraganaComparator.java

package net.sue445.jubeatplusplus.comparator;

import java.io.Serializable;
import java.util.Comparator;
import java.util.HashMap;
import java.util.Map;

/**
 * ひらがなをソートするためのコンパレータ
 * @author sue445
 *
 */
public class HiraganaComparator implements Comparator<String>, Serializable {
  private static final long serialVersionUID = 1L;


	/**
	 * 伸ばし棒
	 */
	private static char HYPHEN = 'ー';

	/**
	 * 対応文字
	 */
	private static final char[][] SUPPORT_CHAR_ARRAY = {
		{'あ', 'あ'},
		{'ぁ', 'あ'},
		{'い', 'い'},
		{'ぃ', 'い'},
		{'う', 'う'},
		{'ぅ', 'う'},
		{'え', 'え'},
		{'ぇ', 'え'},
		{'お', 'お'},
		{'ぉ', 'お'},
		{'か', 'あ'},
		{'が', 'あ'},
		{'き', 'い'},
		{'ぎ', 'い'},
		{'く', 'う'},
		{'ぐ', 'う'},
		{'け', 'え'},
		{'げ', 'え'},
		{'こ', 'お'},
		{'ご', 'お'},
		{'さ', 'あ'},
		{'ざ', 'あ'},
		{'し', 'い'},
		{'じ', 'い'},
		{'す', 'う'},
		{'ず', 'う'},
		{'せ', 'え'},
		{'ぜ', 'え'},
		{'そ', 'お'},
		{'ぞ', 'お'},
		{'た', 'あ'},
		{'だ', 'あ'},
		{'ち', 'い'},
		{'ぢ', 'い'},
		{'つ', 'う'},
		{'づ', 'う'},
		{'っ', 'う'},
		{'て', 'え'},
		{'で', 'え'},
		{'と', 'お'},
		{'ど', 'お'},
		{'な', 'あ'},
		{'に', 'い'},
		{'ぬ', 'う'},
		{'ね', 'え'},
		{'の', 'お'},
		{'は', 'あ'},
		{'ば', 'あ'},
		{'ぱ', 'あ'},
		{'ひ', 'い'},
		{'び', 'い'},
		{'ぴ', 'い'},
		{'ふ', 'う'},
		{'ぶ', 'う'},
		{'ぷ', 'う'},
		{'へ', 'え'},
		{'べ', 'え'},
		{'ぺ', 'え'},
		{'ほ', 'お'},
		{'ぼ', 'お'},
		{'ぽ', 'お'},
		{'ま', 'あ'},
		{'み', 'い'},
		{'む', 'う'},
		{'め', 'え'},
		{'も', 'お'},
		{'や', 'あ'},
		{'ゃ', 'あ'},
		{'ゆ', 'う'},
		{'ゅ', 'う'},
		{'よ', 'お'},
		{'ょ', 'お'},
		{'ら', 'あ'},
		{'り', 'い'},
		{'る', 'う'},
		{'れ', 'え'},
		{'ろ', 'お'},
		{'わ', 'あ'},
		{'ゐ', 'い'},
		{'ゑ', 'え'},
		{'を', 'お'},
		{'ん', 'ん'},
	};

	static class SupportCharDto{
		/**
		 * 表示順
		 */
		int priority;

		/**
		 * 文字
		 */
		//char character;

		/**
		 * 伸ばした時の音
		 */
		char beforeHyphen;
	};
	private static SupportCharDto OTHER = new SupportCharDto();

	/**
	 * charから表示順へのマッピング
	 */
	private static final Map<Character, SupportCharDto> CHAR_TO_DTO = new HashMap<Character, SupportCharDto>();
	static{
		for(int index = 0; index < SUPPORT_CHAR_ARRAY.length; index++){
			SupportCharDto dto = new SupportCharDto();
			dto.priority = index + 1;
			dto.beforeHyphen = SUPPORT_CHAR_ARRAY[index][1];

			CHAR_TO_DTO.put(SUPPORT_CHAR_ARRAY[index][0], dto);
		}

		OTHER.priority = 10000;
		OTHER.beforeHyphen = '無';
	}


	/**
	 * {@inheritDoc}
	 */
	@Override
	public int compare(String str1, String str2) {
		for(int index = 0; index < str1.length() && index < str2.length(); index++){
			int priority1 = findPriority(str1, index);
			int priority2 = findPriority(str2, index);

			if(priority1 != priority2){
				return priority1 - priority2;
			}
		}

		// 短い方を辞書順で先とみなす
		return str1.length() - str2.length();
	}

	/**
	 * 表示順DTOを取得する
	 * @param c
	 * @return
	 */
	// package private
	SupportCharDto findSupportDto(char c){
		SupportCharDto dto = CHAR_TO_DTO.get(c);
		if(dto == null){
			return OTHER;
		}

		return dto;
	}

	/**
	 * 表示順を取得する
	 * @param str
	 * @param index
	 * @return
	 */
	// package private
	int findPriority(String str, int index){
		char currentChar = str.charAt(index);

		if(currentChar == HYPHEN && index > 0){
			// 伸ばし棒の前の文字から母音を判定する
			char beforeChar = str.charAt(index - 1);
			SupportCharDto dto = findSupportDto(beforeChar);
			SupportCharDto beforeDto = findSupportDto(dto.beforeHyphen);
			return beforeDto.priority;
		}

		SupportCharDto currentDto = findSupportDto(currentChar);
		return currentDto.priority;
	}

	/**
	 * ひらがなのみで構成されているかどうか
	 * @param str
	 * @return
	 */
	public static boolean isHiragaraOnly(String str){
		for(int index = 0; index < str.length(); index++){
			char c = str.charAt(index);
			if(c == HYPHEN){
				continue;
			}

			SupportCharDto dto = CHAR_TO_DTO.get(c);
			if(dto == null){
				return false;
			}
		}

		return true;
	}

}

楽曲マスタ

http://jubeatplusplus.appspot.com/file/music.xls

利用例

int ret = new HiraganaComparator(). compare("うぃりあむてるじょきょく", "すうぃーとれいん");