java - 合并排序算法的问题

Question

合并排序算法不起作用。不断收到 IndexOutOFBoundsError。我认为问题可能是因为没有哨兵。当 leftarray 和 rightarray 将自身复制到 k 时，我需要创建一个循环，其中一个数组用完数字，这会导致错误。我需要一些帮助来创建这个哨兵。

private static <T> void mergeSort(Comparable<? extends T>[] items, int begIndx, int endIndx) {
    if (items.length > 1) {
        int midIndx = items.length / 2;
        @SuppressWarnings("unchecked")
        T[] left = (T[]) new Object[midIndx];
        @SuppressWarnings("unchecked")
        T[] right = (T[]) new Object[items.length - midIndx];

        for (int i = 0; i < midIndx; i++) {
            left[i] = (T) items[i];
        }
        for (int i = midIndx; i < items.length; i++) {
            right[i] = (T) items[i];
        }

        mergeSort(items, begIndx, midIndx);
        mergeSort(items, midIndx + 1, endIndx);
        merge(items, begIndx, midIndx, endIndx);
    }
}

@SuppressWarnings("unchecked")
private static <T> void merge(Comparable<? extends T>[] array,
                              int begIndx, int midIndx, int endIndx) {
    int sizeOfLeft = midIndx - begIndx + 1;
    int sizeOfRight = endIndx - midIndx;

    /// change to generic later
    @SuppressWarnings("unchecked")
    T[] leftArr = (T[]) new Object[sizeOfLeft + 1];
    @SuppressWarnings("unchecked")
    T[] rightArr = (T[]) new Object[sizeOfRight + 1];

    for (int i = 0; i < sizeOfLeft; i++) {
        leftArr[i] = (T) array[begIndx + i];
    }
    for (int j = 0; j < sizeOfRight; j++) {
        rightArr[j] = (T) array[midIndx + j + 1];
    }

    int i = 0;
    int j = 0;

    // changed to less than or equal to rather than "less than"
    // this is because this is a zero based index system
    // and because endeIndex is not a length but an index,
    // you need to populate it.
    for (int k = begIndx; k <= endIndx; k++) {
        // use comparable here
        if (((Integer) leftArr[i]).compareTo((Integer) rightArr[j]) <= 0) {
            array[k] = (Comparable<? extends T>) leftArr[i];
            i = i + 1;
        } else if (((Integer) leftArr[i]).compareTo((Integer) rightArr[j]) >= 0) {
            /// just replaces it so don't use comparable
            array[k] = (Comparable<? extends T>) rightArr[j];
            j = j + 1;
        } else if ((Integer) leftArr[sizeOfLeft] == null) {
            array[k] =  (Comparable<? extends T>) rightArr[j];
            j = j + 1;
        } else if ((Integer) rightArr[sizeOfRight] == null) {
            array[k] = (Comparable<? extends T>) leftArr[i];
            i = i + 1;
        }
    }
}

我创建了一个整数数组arr = new Integer[5]。所以这 5 个数字应该被排序。

Answer 1

您的代码中存在多个问题:

• 尚不清楚索引 endIndx 是否包含或排除。如果排除 endIndx，代码会简单得多，并且对数组进行排序的初始调用很简单:mergesort(arr, 0, arr.length);

• mergesort 中的初始测试不正确:您应该测试切片是否包含超过 1 个元素，而不是测试数组长度:

  if (endIndx - begIndx > 1)
  

• left 和 right 数组在 mergesort 中未使用。

• 在merge中不需要在数组left和right中分配额外的元素，比较更简单数组索引值与子数组的长度。这种哨兵方法令人困惑，不应该使用。

这是一个简化版本:

private static <T> void mergeSort(Comparable<? extends T>[] items, int begIndx, int endIndx) {
    if (endIndx - begIndx > 1) {
        int midIndx = items.length / 2;
        mergeSort(items, begIndx, midIndx);
        mergeSort(items, midIndx, endIndx);
        merge(items, begIndx, midIndx, endIndx);
    }
}

@SuppressWarnings("unchecked")
private static <T> void merge(Comparable<? extends T>[] array,
                              int begIndx, int midIndx, int endIndx) {
    int sizeOfLeft = midIndx - begIndx;
    int sizeOfRight = endIndx - midIndx;

    /// change to generic later
    @SuppressWarnings("unchecked")
    T[] leftArr = (T[]) new Object[sizeOfLeft];
    @SuppressWarnings("unchecked")
    T[] rightArr = (T[]) new Object[sizeOfRight];

    for (int i = 0; i < sizeOfLeft; i++) {
        leftArr[i] = (T)array[begIndx + i];
    }
    for (int j = 0; j < sizeOfRight; j++) {
        rightArr[j] = (T)array[midIndx + j];
    }

    int i = 0;
    int j = 0;
    int k = begIndx;

    while (i < sizeOfLeft && j < sizeOfRight) {
        /// use comparable to compare actual values
        if ((Integer)leftArr[i]).compareTo((Integer)rightArr[j]) <= 0) {
            array[k] = (Comparable<? extends T>)leftArr[i];
            i++;
            k++;
        } else {
            array[k] = (Comparable<? extends T>)rightArr[j];
            j++;
            k++;
        }
    }
    while (i < sizeOfLeft) {
        array[k] = (Comparable<? extends T>)leftArr[i];
        i++;
        k++;
    }
    while (j < sizeOfRight) {
        array[k] = (Comparable<? extends T>)rightArr[j];
        j++;
        k++;
    }
}