Java实现中国象棋的示例代码

@Slf4j
public class BoardPanel extends JPanel implements LambdaMouseListener {

    /**
     * 用于标记棋盘走棋痕迹
     */
    private final transient TraceMarker traceMarker;
    /**
     * 当前走棋开始坐标位置对应棋子
     */
    private transient ChessPiece curFromPiece;
    /**
     * 场景
     */
    private transient Situation situation;

    /**
     * Create the panel.
     */
    public BoardPanel() {
        setBorder(new EmptyBorder(5, 5, 5, 5));
        setLayout(null);
        // 初始化标记符
        traceMarker = new TraceMarker(BoardPanel.this);
        // 添加鼠标事件
        addMouseListener(this);
    }

    /**
     * 更新标记
     */
    public void updateMark(Place from, Place to) {
        // 更新标记
        curFromPiece = null;
        // 更改标记
        traceMarker.endedStep(from, to);
    }

    /**
     * 初始化所有标记
     */
    public void initMark() {
        traceMarker.initMarker();
    }

    /**
     * 添加棋子
     */
    public void init(Situation situation) {
        this.situation = situation;
        // 移除所有组件
        this.removeAll();
        // 添加棋子
        situation.getPieceList().forEach(it -> add(it.getComp()));
        situation.getSituationRecord().getEatenPieceList().forEach(it -> add(it.getComp()));
        // 初始化标记符
        traceMarker.initMarker();
        repaint();
    }

    /**
     * @param e 鼠标按压事件对象
     */
    @Override
    public void mouseReleased(MouseEvent e) {
        // 位置
        Place pointerPlace = ChessDefined.convertLocationToPlace(e.getPoint());
        if (pointerPlace == null) {
            return;
        }
        if (situation.winner() != null) {
            log.warn("已经存在胜利者: {}, 无法走棋", situation.winner());
            return;
        }
        // 当前走棋方
        @NonNull Part pointerPart = situation.getNextPart();
        // 当前焦点棋子
        ChessPiece pointerPiece = situation.getChessPiece(pointerPlace);
        // 通过当前方和当前位置判断是否可以走棋
        // step: form
        if (curFromPiece == null) {
            // 当前焦点位置有棋子且是本方棋子
            if (pointerPiece != null && pointerPiece.piece.part == pointerPart) {
                // 本方棋子, 同时是from指向
                curFromPiece = pointerPiece;
                traceMarker.setMarkFromPlace(pointerPlace);
                // 获取toList
                MyList<Place> list = curFromPiece.piece.role.find(new AnalysisBean(situation.generatePieces()), pointerPart, pointerPlace);
                traceMarker.showMarkPlace(list);
                ChessAudio.CLICK_FROM.play();
                log.info("true -> 当前焦点位置有棋子且是本方棋子");
                final ListPool listPool = ListPool.localPool();
                listPool.addListToPool(list);
                return;
            }
            log.warn("warning -> from 焦点指示错误");
            return;
        }
        if (pointerPlace.equals(curFromPiece.getPlace())) {
            log.warn("false -> from == to");
            return;
        }
        // 当前焦点位置有棋子且是本方棋子
        if (pointerPiece != null && pointerPiece.piece.part == pointerPart) {
            assert curFromPiece.piece.part == pointerPart : "当前焦点位置有棋子且是本方棋子 之前指向了对方棋子";
            // 更新 curFromPiece
            curFromPiece = pointerPiece;
            traceMarker.setMarkFromPlace(pointerPlace);
            MyList<Place> list = curFromPiece.piece.role.find(new AnalysisBean(situation.generatePieces()), pointerPart, pointerPlace);
            traceMarker.showMarkPlace(list);
            ChessAudio.CLICK_FROM.play();
            log.info("true -> 更新 curFromPiece");
            ListPool.localPool().addListToPool(list);
            return;
        }
        final StepBean stepBean = StepBean.of(curFromPiece.getPlace(), pointerPlace);
        // 如果不符合规则则直接返回
        final Piece[][] pieces = situation.generatePieces();
        if (!curFromPiece.piece.role.rule.check(pieces, pointerPart, stepBean.from, stepBean.to)) {
            // 如果当前指向棋子是本方棋子
            log.warn("不符合走棋规则");
            return;
        }
        // 如果达成长拦或者长捉, 则返回
        final StepBean forbidStepBean = situation.getForbidStepBean();
        if (forbidStepBean != null && forbidStepBean.from == stepBean.from && forbidStepBean.to == stepBean.to) {
            ChessAudio.MAN_MOV_ERROR.play();
            log.warn("长拦或长捉");
            return;
        }
        AnalysisBean analysisBean = new AnalysisBean(pieces);
        // 如果走棋后, 导致两个 BOSS 对面, 则返回
        if (!analysisBean.isBossF2FAfterStep(curFromPiece.piece, stepBean.from, stepBean.to)) {
            ChessAudio.MAN_MOV_ERROR.play();
            log.warn("BOSS面对面");
            return;
        }
        /* 模拟走一步棋, 之后再计算对方再走一步是否能够吃掉本方的 boss */
        if (analysisBean.simulateOneStep(stepBean, bean -> bean.canEatBossAfterOneAiStep(Part.getOpposite(pointerPart)))) {
            ChessAudio.MAN_MOV_ERROR.play();
            log.warn("BOSS 危险");
            if (!Application.config().isActiveWhenBeCheck()) {
                return;
            }
        }
        // 当前棋子无棋子或者为对方棋子, 且符合规则, 可以走棋
        Object[] objects = new Object[]{stepBean.from, stepBean.to, PlayerType.PEOPLE};
        final boolean sendSuccess = Application.context().getCommandExecutor().sendCommandWhenNotRun(CommandExecutor.CommandType.LocationPiece, objects);
        if (!sendSuccess) {
            log.warn("命令未发送成功: {} ==> {}", CommandExecutor.CommandType.LocationPiece, Arrays.toString(objects));
        }
    }

    @Override
    public void paintComponent(Graphics g) {
        super.paintComponent(g);
        Image img = ChessImage.CHESS_BOARD.getImage();
        int imgWidth = img.getWidth(this);
        int imgHeight = img.getHeight(this);// 获得图片的宽度与高度
        int fWidth = getWidth();
        int fHeight = getHeight();// 获得窗口的宽度与高度
        int x = (fWidth - imgWidth) / 2;
        int y = (fHeight - imgHeight) / 2;
        // 520 576 514 567
        log.debug(String.format("%s,%s,%s,%s,%s,%s", imgWidth, imgHeight, fWidth, fHeight, x, y));
        g.drawImage(img, 0, 0, null);
    }

}

@Slf4j
public class CommandExecutor {

    /**
     * 异步调用线程, 来处理走棋命令
     */
    private final CtrlLoopThreadComp ctrlLoopThreadComp;
    private final BoardPanel boardPanel;
    /**
     * 是否持续运行标记
     */
    private volatile boolean sustain;

    public CommandExecutor(BoardPanel boardPanel) {
        this.boardPanel = boardPanel;
        this.ctrlLoopThreadComp = CtrlLoopThreadComp.ofRunnable(this::loop)
                .setName("CommandExecutor")
                .catchFun(CtrlLoopThreadComp.CATCH_FUNCTION_CONTINUE);
    }

    /**
     * 下一步骤命令
     */
    private CommandType nextCommand;
    /**
     * 下一步骤命令的参数
     */
    private Object nextParamObj;

    private volatile boolean isRun;

    /**
     * @param commandType 命令类型
     */
    public void sendCommand(@NonNull CommandType commandType) {
        sendCommand(commandType, null);
    }

    /**
     * @param commandType 命令类型
     * @param paramObj    命令参数
     */
    public synchronized void sendCommand(@NonNull CommandType commandType, Object paramObj) {
        this.nextCommand = commandType;
        this.nextParamObj = paramObj;
        sustain = false;
        this.ctrlLoopThreadComp.startOrWake();
    }

    /**
     * 只有在 线程没有运行的情况下, 才能添加成功
     *
     * @param commandType 命令类型
     * @param paramObj    命令参数
     * @return 是否添加成功
     */
    public synchronized boolean sendCommandWhenNotRun(@NonNull CommandType commandType, Object paramObj) {
        if (isRun) {
            return false;
        }
        sendCommand(commandType, paramObj);
        return true;
    }

    private void loop() {
        final CommandType command;
        final Object paramObj;
        synchronized (this) {
            command = this.nextCommand;
            paramObj = this.nextParamObj;
            this.nextCommand = null;
            this.nextParamObj = null;
        }
        if (command != null) {
            isRun = true;
            try {
                log.debug("处理事件[{}] start", command.getLabel());
                consumerCommand(command, paramObj);
                log.debug("处理事件[{}] end ", command.getLabel());
            } catch (Exception e) {
                log.error("执行命令[{}]发生异常", command.getLabel(), e);
                new Thread(() -> JOptionPane.showMessageDialog(boardPanel, e.getMessage(), e.toString(), JOptionPane.ERROR_MESSAGE)).start();
            }
        } else {
            this.ctrlLoopThreadComp.pause();
            isRun = false;
        }
    }

    /**
     * 运行
     */
    private void consumerCommand(final CommandType commandType, Object paramObj) {
        switch (commandType) {
            case SuspendCallBackOrAiRun:
                break;
            case CallBackOneTime:
                Application.context().rollbackOneStep();
                break;
            case AiRunOneTime:
                if (Application.context().aiRunOneTime() != null) {
                    log.debug("已经决出胜方!");
                }
                break;
            case SustainCallBack:
                sustain = true;
                while (sustain) {
                    if (!Application.context().rollbackOneStep()) {
                        sustain = false;
                        break;
                    }
                    Throws.con(Application.config().getComIntervalTime(), Thread::sleep).logThrowable();
                }
                break;
            case SustainAiRun:
                sustain = true;
                while (sustain) {
                    if (Application.context().aiRunOneTime() != null) {
                        log.debug("已经决出胜方, AI执行暂停!");
                        sustain = false;
                        break;
                    }
                    Throws.con(Application.config().getComIntervalTime(), Thread::sleep).logThrowable();
                }
                break;
            case SustainAiRunIfNextIsAi:
                sustain = true;
                while (sustain) {
                    // 如果下一步棋手不是 AI, 则暂停
                    if (!PlayerType.COM.equals(Application.config().getPlayerType(Application.context().getSituation().getNextPart()))) {
                        sustain = false;
                        log.debug("下一步棋手不是 AI, 暂停!");
                    } else if (Application.context().aiRunOneTime() != null) {
                        log.debug("已经决出胜方, AI执行暂停!");
                        sustain = false;
                    } else {
                        Throws.con(Application.config().getComIntervalTime(), Thread::sleep).logThrowable();
                    }
                }
                break;
            case LocationPiece:
                final Object[] params = (Object[]) paramObj;
                Place from = (Place) params[0];
                Place to = (Place) params[1];
                PlayerType type = (PlayerType) params[2];
                Application.context().locatePiece(from, to, type);
                sendCommand(CommandExecutor.CommandType.SustainAiRunIfNextIsAi);
                break;
            default:
                throw new ShouldNotHappenException("未处理的命令: " + commandType);
        }
    }

    /**
     * 命令支持枚举(以下命令应当使用同一个线程运行, 一个事件结束之后, 另一个事件才能开始运行.)
     */
    @SuppressWarnings("java:S115")
    public enum CommandType {
        SuspendCallBackOrAiRun("停止撤销|AI计算"),
        CallBackOneTime("撤销一步"),
        SustainCallBack("持续撤销"),
        AiRunOneTime("AI计算一步"),
        SustainAiRun("AI持续运行"),
        SustainAiRunIfNextIsAi("COM角色运行"),
        LocationPiece("ui落子命令");

        @Getter
        private final String label;

        CommandType(String label) {
            this.label = label;
        }
    }

}

@NoArgsConstructor(access = AccessLevel.PRIVATE)
@Slf4j
public class AlphaBeta {

    private static final int MAX = 100_000_000;
    /**
     * 这里要保证 Min + Max = 0, 哪怕是微不足道的差距都可能导致发生错误
     */
    private static final int MIN = -MAX;

    /**
     * 根据棋子数量, 动态调整搜索深度
     *
     * @param pieceNum 棋子数量
     * @return 调整搜索深度差值
     */
    public static int searchDeepSuit(final int pieceNum) {
        // 根据棋子数量, 动态调整搜索深度
        if (pieceNum > 20) {
            return -2;
        } else if (pieceNum <= 4) {
            return 4;
        } else if (pieceNum <= 8) {
            return 2;
        }
        return 0;
    }

    /**
     * 生成待选的列表，就是可以下子的空位, 如果 deep > 2 则对搜索结果进行排序.
     *
     * @param analysisBean 棋盘分析对象
     * @param curPart      当前走棋方
     * @param deep         搜索深度
     * @return 可以下子的空位集合
     */
    private static MyList<StepBean> geneNestStepPlaces(final AnalysisBean analysisBean, final Part curPart, final int deep) {
        final Piece[][] pieces = analysisBean.pieces;
        // 是否杀棋
        MyList<StepBean> stepBeanList = ListPool.localPool().getAStepBeanList();
        for (int x = 0; x < ChessDefined.RANGE_X; x++) {
            for (int y = 0; y < ChessDefined.RANGE_Y; y++) {
                final Piece fromPiece = pieces[x][y];
                if (fromPiece != null && fromPiece.part == curPart) {
                    final Place from = Place.of(x, y);
                    // TO DO 考虑下此处添加至集合的做法 在计算时 是否有优化空间.
                    final MyList<Place> list = fromPiece.role.find(analysisBean, curPart, from);
                    if (list.isEmpty()) {
                        ListPool.localPool().addListToPool(list);
                        continue;
                    }
                    final Object[] elementData = list.eleTemplateDate();
                    for (int i = 0, len = list.size(); i < len; i++) {
                        stepBeanList.add(StepBean.of(from, (Place) elementData[i]));
                    }
                    ListPool.localPool().addListToPool(list);
                }
            }
        }
        // 是否排序, 如果搜索深度大于2, 则对结果进行排序
        // 排序后的结果, 进入极大极小值搜索算法时, 容易被剪枝.
        if (deep > 2) {
            orderStep(analysisBean, stepBeanList, curPart);
        }

        return stepBeanList;
    }

    /**
     * 对 空位列表 进行排序, 排序后的空位列表, 进入极大极小值搜索算法时, 容易被剪枝.
     *
     * @param analysisBean 棋盘分析对象
     * @param stepBeanList 可以下子的空位列表
     * @param curPart      当前走棋方
     */
    private static void orderStep(final AnalysisBean analysisBean, final MyList<StepBean> stepBeanList, final Part curPart) {
        final Piece[][] srcPieces = analysisBean.pieces;
        // 进入循环之前计算好循环内使用常量
        MyList<DoubleBean<Integer, StepBean>> bestPlace = ListPool.localPool().getADoubleBeanList();
        // 对方棋手
        final Part oppositeCurPart = Part.getOpposite(curPart);
        int best = MIN;

        final Object[] objects = stepBeanList.eleTemplateDate();
        for (int i = 0; i < stepBeanList.size(); i++) {
            final StepBean item = (StepBean) objects[i];
            final Place to = item.to;
            // 备份
            final Piece eatenPiece = srcPieces[to.x][to.y];
            int score;
            // 判断是否胜利
            if (eatenPiece != null && eatenPiece.role == Role.BOSS) {
                score = MAX;
            } else {
                // 走棋
                final int invScr = analysisBean.goForward(item.from, to, eatenPiece);
                DebugInfo.incrementAlphaBetaOrderTime();
                // 评分
                score = negativeMaximumWithNoCut(analysisBean, oppositeCurPart, -best);
                // 退回上一步
                analysisBean.backStep(item.from, to, eatenPiece, invScr);
            }
            // 这里添加进所有的分数
            bestPlace.add(new DoubleBean<>(score, item));
            if (score > best) { // 找到一个更好的分，就把以前存的位子全部清除
                best = score;
            }
        }
        /* 排序后返回 */
        // 这样排序是正确的, 可以有效消减数量
        bestPlace.sort((o1, o2) -> o2.getO1() - o1.getO1());

        stepBeanList.clear();
        bestPlace.forEach(dou -> stepBeanList.add(dou.getO2()));

        ListPool.localPool().addListToDoubleBeanListPool(bestPlace);
    }


    /**
     * 负极大值搜索算法(不带剪枝算法)
     *
     * @param analysisBean 局势分析对象
     * @param curPart      当前走棋方
     * @return 负极大值搜索算法计算分值
     */
    private static int negativeMaximumWithNoCut(AnalysisBean analysisBean, Part curPart, int alphaBeta) {
        // 1. 初始化各个变量
        final Piece[][] pieces = analysisBean.pieces;
        int best = MIN;
        // 2. 生成待选的列表，就是可以下子的列表
        MyList<StepBean> stepBeanList = geneNestStepPlaces(analysisBean, curPart, 1);

        final Object[] objects = stepBeanList.eleTemplateDate();
        for (int i = 0, len = stepBeanList.size(); i < len; i++) {
            final StepBean item = (StepBean) objects[i];
            Place from = item.from;
            Place to = item.to;
            // 备份
            Piece eatenPiece = pieces[to.x][to.y];
            int score;
            // 判断是否胜利
            if (eatenPiece != null && eatenPiece.role == Role.BOSS) {
                score = MAX;
            } else {
                // 走棋
                final int invScr = analysisBean.goForward(from, to, eatenPiece);
                DebugInfo.incrementAlphaBetaOrderTime();
                score = analysisBean.getCurPartEvaluateScore(curPart);
                // 退回上一步
                analysisBean.backStep(from, to, eatenPiece, invScr);
            }
            if (score > best) { // 找到一个更好的分，就更新分数
                best = score;
            }
            if (score > alphaBeta) { // alpha剪枝
                break;
            }
        }
        ListPool.localPool().addListToStepBeanListPool(stepBeanList);
        return -best;
    }


    /**
     * 奇数层是电脑(max层)thisSide, 偶数层是human(min层)otherSide
     *
     * @param srcPieces 棋盘
     * @param curPart   当前走棋方
     * @param deep      搜索深度
     * @param forbidStep 禁止的步骤(长捉或长拦)
     * @return 下一步的位置
     */
    public static Set<StepBean> getEvaluatedPlace(final Piece[][] srcPieces, final Part curPart, final int deep, final StepBean forbidStep) {
        // 1. 初始化各个变量
        final AnalysisBean analysisBean = new AnalysisBean(srcPieces);
        // 2. 获取可以下子的空位列表
        MyList<StepBean> stepBeanList = geneNestStepPlaces(analysisBean, curPart, deep);
        // 3. 移除不该下的子
        stepBeanList.remove(forbidStep);
        // 进入循环之前计算好循环内使用常量
        Set<StepBean> bestPlace = new HashSet<>();
        int best = MIN;
        // 对方棋手
        final Part oppositeCurPart = Part.getOpposite(curPart);
        // 下一深度
        final int nextDeep = deep - 1;
        log.debug("size : {}, content: {}", stepBeanList.size(), stepBeanList);
        final Object[] objects = stepBeanList.eleTemplateDate();
        for (int i = 0, len = stepBeanList.size(); i < len; i++) {
            StepBean item = (StepBean) objects[i];
            final Place to = item.to;
            // 备份
            final Piece eatenPiece = srcPieces[to.x][to.y];
            int score;
            // 判断是否胜利
            if (eatenPiece != null && eatenPiece.role == Role.BOSS) {
                // 步数越少, 分值越大
                score = MAX + deep;
            } else {
                // 走棋
                final int invScr = analysisBean.goForward(item.from, to, eatenPiece);
                // 评分
                if (deep <= 1) {
                    score = analysisBean.getCurPartEvaluateScore(curPart);
                } else {
                    score = negativeMaximum(analysisBean, oppositeCurPart, nextDeep, -best);
                }
                // 退回上一步
                analysisBean.backStep(item.from, to, eatenPiece, invScr);
            }
            if (score == best) { // 找到相同的分数, 就添加这一步
                bestPlace.add(item);
            }
            if (score > best) { // 找到一个更好的分，就把以前存的位子全部清除
                best = score;
                bestPlace.clear();
                bestPlace.add(item);
            }
        }
        ListPool.end();
        ListPool.localPool().addListToStepBeanListPool(stepBeanList);
        return bestPlace;
    }

    /**
     * 奇数层是电脑(max层)thisSide, 偶数层是human(min层)otherSide
     *
     * @param srcPieces 棋盘
     * @param curPart   当前走棋方
     * @param deep      搜索深度
     * @param forbidStep 禁止的步骤(长捉或长拦)
     * @return 下一步的位置
     */
    public static Set<StepBean> getEvaluatedPlaceWithParallel(final Piece[][] srcPieces, final Part curPart, final int deep, final StepBean forbidStep) {
        // 1. 初始化各个变量
        final AnalysisBean srcAnalysisBean = new AnalysisBean(srcPieces);
        // 2. 获取可以下子的空位列表
        MyList<StepBean> stepBeanList = geneNestStepPlaces(srcAnalysisBean, curPart, deep);
        // 3. 移除不该下的子
        stepBeanList.remove(forbidStep);
        // 进入循环之前计算好循环内使用常量
        final Set<StepBean> bestPlace = new HashSet<>();
        final AtomicInteger best = new AtomicInteger(MIN);
        // 对方棋手
        final Part oppositeCurPart = Part.getOpposite(curPart);
        // 下一深度
        final int nextDeep = deep - 1;
        log.debug("size : {}, content: {}", stepBeanList.size(), stepBeanList);

        Arrays.stream(stepBeanList.toArray()).parallel().filter(Objects::nonNull).map(StepBean.class::cast).forEach(item -> {
            log.debug("并行流 ==> Thread : {}", Thread.currentThread().getId());
            final Piece[][] pieces = ArrayUtils.deepClone(srcPieces);
            final AnalysisBean analysisBean = new AnalysisBean(pieces);

            final Place to = item.to;
            // 备份
            final Piece eatenPiece = pieces[to.x][to.y];
            int score;
            // 判断是否胜利
            if (eatenPiece != null && eatenPiece.role == Role.BOSS) {
                // 步数越少, 分值越大
                score = MAX + deep;
            } else {
                // 走棋
                final int invScr = analysisBean.goForward(item.from, to, eatenPiece);
                // 评分
                if (deep <= 1) {
                    score = analysisBean.getCurPartEvaluateScore(curPart);
                } else {
                    score = negativeMaximum(analysisBean, oppositeCurPart, nextDeep, -best.get());
                }
                // 退回上一步
                analysisBean.backStep(item.from, to, eatenPiece, invScr);
            }
            if (score == best.get()) { // 找到相同的分数, 就添加这一步
                synchronized (bestPlace) {
                    bestPlace.add(item);
                }
            }
            if (score > best.get()) { // 找到一个更好的分，就把以前存的位子全部清除
                best.set(score);
                synchronized (bestPlace) {
                    bestPlace.clear();
                    bestPlace.add(item);
                }
            }
            ListPool.end();
        });
        ListPool.localPool().addListToStepBeanListPool(stepBeanList);
        ListPool.end();
        return bestPlace;
    }


    /**
     * 负极大值搜索算法
     *
     * @param analysisBean 局势分析对象
     * @param curPart      当前走棋方
     * @param deep         搜索深度
     * @param alphaBeta    alphaBeta 剪枝分值
     * @return 负极大值搜索算法计算分值
     */
    private static int negativeMaximum(AnalysisBean analysisBean, Part curPart, int deep, int alphaBeta) {
        // 1. 初始化各个变量
        final Piece[][] pieces = analysisBean.pieces;
        int best = MIN;
        // 对方棋手
        final Part oppositeCurPart = Part.getOpposite(curPart);
        // 下一深度
        final int nextDeep = deep - 1;
        // 2. 生成待选的列表，就是可以下子的列表
        final MyList<StepBean> stepBeanList = geneNestStepPlaces(analysisBean, curPart, deep);

        final Object[] objects = stepBeanList.eleTemplateDate();
        for (int i = 0, len = stepBeanList.size(); i < len; i++) {
            final StepBean item = (StepBean) objects[i];
            Place from = item.from;
            Place to = item.to;
            // 备份
            Piece eatenPiece = pieces[to.x][to.y];
            int score;
            // 判断是否胜利
            if (eatenPiece != null && eatenPiece.role == Role.BOSS) {
                // 步数越少, 分值越大
                score = MAX + deep;
            } else {
                // 走棋
                final int invScr = analysisBean.goForward(from, to, eatenPiece);
                // 评估
                if (deep <= 1) {
                    score = analysisBean.getCurPartEvaluateScore(curPart);
                } else {
                    score = negativeMaximum(analysisBean, oppositeCurPart, nextDeep, -best);
                }
                // 退回上一步
                analysisBean.backStep(from, to, eatenPiece, invScr);
            }
            if (score > best) { // 找到一个更好的分，就更新分数
                best = score;
            }
            if (score > alphaBeta) { // alpha剪枝
                break;
            }
        }
        ListPool.localPool().addListToStepBeanListPool(stepBeanList);
        return -best;
    }

}