📜  B树中的删除操作

📅  最后修改于: 2021-04-17 10:20:53             🧑  作者: Mango

建议参考以下帖子作为该帖子的先决条件。

B-树|设置1(简介)
B-树|套装2(插入)

B树是多路搜索树的一种。因此,如果您通常对多路搜索树不熟悉,那么最好继续看一下IIT-Delhi的这段视频讲座。一旦您清楚了多向搜索树的基础,B-Tree操作将更容易理解。

以下解释和算法的来源是Clifford Stein,Thomas H.Cormen,Charles E.Leiserson,Ronald L.Rivest撰写的《算法简介》第三版。

删除过程:
从B树中删除比插入要复杂得多,因为我们可以从任何节点(而不只是叶子)中删除键,并且当从内部节点中删除键时,我们将不得不重新排列该节点的子级。

在插入过程中,我们必须确保删除操作不违反B树属性。正如我们必须确保节点不会由于插入而变得太大一样,我们也必须确保在删除过程中节点不会变得太小(除了允许根的根数小于最小t-1之外)键)。就像如果要插入密钥的路径上的节点已满,可能需要备份简单的插入算法一样,如果路径上的节点(不是根节点),则必须备份一种简单的删除方法要删除密钥的位置的密钥数量最少。

删除过程从以x为根的子树中删除密钥k。该过程保证了,无论何时它在节点x上递归调用,x中的键数至少为最小度t。请注意,此条件比通常的B树条件所需的最小密钥多一个密钥,因此有时在递归递归到该子节点之前,有时可能必须将一个密钥移入一个子节点。这种增强的条件使我们可以一次向下删除树中的密钥,而不必“备份”(有一个例外,我们将对此进行解释)。您应该理解以下有关从B树中删除的规范,并应理解,如果根节点x曾经成为没有键的内部节点(这种情况可能在情况2c和3b中发生,那么我们将删除x,并且x是唯一的子x .c1成为树的新根,将树的高度减一,并保留该树的根包含至少一个键的属性(除非树为空)。

我们概述了删除如何与从B树中删除键的各种情况一起使用。

1.如果密钥k在节点x中并且x是叶子,请从x删除密钥k。

2.如果密钥k在节点x中,并且x是内部节点,请执行以下操作。

a)如果在节点x中在k之前的子y至少具有t个键,则在以y为根的子树中找到k的前任k0。递归删除k0,然后用x中的k0替换k。 (我们可以找到k0并在一次向下传递中将其删除。)

b)如果y的键数少于t,则对称地检查节点x中紧随k的子级z。如果z至少具有t个键,则在以z为根的子树中找到k的后继k0。递归删除k0,然后用x中的k0替换k。 (我们可以找到k0并在一次向下传递中将其删除。)

c)否则,如果y和z都只有t-1个键,则将k和所有z合并到y中,以便x失去k和指向z的指针,并且y现在包含2t-1个键。然后释放z并从y中递归删除k。

3.如果内部节点x中不存在密钥k,则如果k中根本没有k,则确定必须包含k的适当子树的根xc(i)。如果xc(i)只有t-1个键,请根据需要执行步骤3a或3b,以确保我们下降到包含至少t个键的节点。然后,通过递归x的适当子代来完成。

a)如果xc(i)仅具有t-1个键,但具有至少t个键的直接同级,则通过将一个键从x向下移动到xc(i),再将一个键从xc中移动来给xc(i)一个额外的键(i)的直接左或右同级向上到x,并将适当的子指针从同级移动到xc(i)。

b)如果xc(i)和xc(i)的两个直接同级都具有t-1个密钥,则将xc(i)与一个同级合并,这涉及将密钥从x向下移动到新的合并节点中,成为中位数该节点的密钥。

由于B树中的大多数键都在叶子中,因此删除操作最常用于从叶子中删除键。然后,递归删除过程将向下进行一次操作,而不必进行备份。但是,当在内部节点中删除密钥时,该过程将向下通过树,但可能必须返回到删除密钥的节点,才能用其前任或后继替换密钥(情况2a和2b)。

下图说明了删除过程。
BTreeDelet1
BTreeDelet2

执行:
以下是删除过程的C++实现。

/* The following program performs deletion on a B-Tree. It contains functions
   specific for deletion along with all the other functions provided in the
   previous articles on B-Trees. See https://www.geeksforgeeks.org/b-tree-set-1-introduction-2/
   for previous article.
  
   The deletion function has been compartmentalized into 8 functions for ease
   of understanding and clarity
  
   The following functions are exclusive for deletion
   In class BTreeNode:
    1) remove
    2) removeFromLeaf
    3) removeFromNonLeaf
    4) getPred
    5) getSucc
    6) borrowFromPrev
    7) borrowFromNext
    8) merge
    9) findKey
  
   In class BTree:
     1) remove
  
  The removal of a key from a B-Tree is a fairly complicated process. The program handles
  all the 6 different cases that might arise while removing a key.
  
  Testing: The code has been tested using the B-Tree provided in the CLRS book( included
  in the main function ) along with other cases.
  
  Reference: CLRS3 - Chapter 18 - (499-502)
  It is advised to read the material in CLRS before taking a look at the code. */
  
#include
using namespace std;
  
// A BTree node
class BTreeNode
{
    int *keys;  // An array of keys
    int t;      // Minimum degree (defines the range for number of keys)
    BTreeNode **C; // An array of child pointers
    int n;     // Current number of keys
    bool leaf; // Is true when node is leaf. Otherwise false
  
public:
  
    BTreeNode(int _t, bool _leaf);   // Constructor
  
    // A function to traverse all nodes in a subtree rooted with this node
    void traverse();
  
    // A function to search a key in subtree rooted with this node.
    BTreeNode *search(int k);   // returns NULL if k is not present.
  
    // A function that returns the index of the first key that is greater
    // or equal to k
    int findKey(int k);
  
    // A utility function to insert a new key in the subtree rooted with
    // this node. The assumption is, the node must be non-full when this
    // function is called
    void insertNonFull(int k);
  
    // A utility function to split the child y of this node. i is index
    // of y in child array C[].  The Child y must be full when this
    // function is called
    void splitChild(int i, BTreeNode *y);
  
    // A wrapper function to remove the key k in subtree rooted with
    // this node.
    void remove(int k);
  
    // A function to remove the key present in idx-th position in
    // this node which is a leaf
    void removeFromLeaf(int idx);
  
    // A function to remove the key present in idx-th position in
    // this node which is a non-leaf node
    void removeFromNonLeaf(int idx);
  
    // A function to get the predecessor of the key- where the key
    // is present in the idx-th position in the node
    int getPred(int idx);
  
    // A function to get the successor of the key- where the key
    // is present in the idx-th position in the node
    int getSucc(int idx);
  
    // A function to fill up the child node present in the idx-th
    // position in the C[] array if that child has less than t-1 keys
    void fill(int idx);
  
    // A function to borrow a key from the C[idx-1]-th node and place
    // it in C[idx]th node
    void borrowFromPrev(int idx);
  
    // A function to borrow a key from the C[idx+1]-th node and place it
    // in C[idx]th node
    void borrowFromNext(int idx);
  
    // A function to merge idx-th child of the node with (idx+1)th child of
    // the node
    void merge(int idx);
  
    // Make BTree friend of this so that we can access private members of
    // this class in BTree functions
    friend class BTree;
};
  
class BTree
{
    BTreeNode *root; // Pointer to root node
    int t;  // Minimum degree
public:
  
    // Constructor (Initializes tree as empty)
    BTree(int _t)
    {
        root = NULL;
        t = _t;
    }
  
    void traverse()
    {
        if (root != NULL) root->traverse();
    }
  
    // function to search a key in this tree
    BTreeNode* search(int k)
    {
        return (root == NULL)? NULL : root->search(k);
    }
  
    // The main function that inserts a new key in this B-Tree
    void insert(int k);
  
    // The main function that removes a new key in thie B-Tree
    void remove(int k);
  
};
  
BTreeNode::BTreeNode(int t1, bool leaf1)
{
    // Copy the given minimum degree and leaf property
    t = t1;
    leaf = leaf1;
  
    // Allocate memory for maximum number of possible keys
    // and child pointers
    keys = new int[2*t-1];
    C = new BTreeNode *[2*t];
  
    // Initialize the number of keys as 0
    n = 0;
}
  
// A utility function that returns the index of the first key that is
// greater than or equal to k
int BTreeNode::findKey(int k)
{
    int idx=0;
    while (idxn < t)
            fill(idx);
  
        // If the last child has been merged, it must have merged with the previous
        // child and so we recurse on the (idx-1)th child. Else, we recurse on the
        // (idx)th child which now has atleast t keys
        if (flag && idx > n)
            C[idx-1]->remove(k);
        else
            C[idx]->remove(k);
    }
    return;
}
  
// A function to remove the idx-th key from this node - which is a leaf node
void BTreeNode::removeFromLeaf (int idx)
{
  
    // Move all the keys after the idx-th pos one place backward
    for (int i=idx+1; in >= t)
    {
        int pred = getPred(idx);
        keys[idx] = pred;
        C[idx]->remove(pred);
    }
  
    // If the child C[idx] has less that t keys, examine C[idx+1].
    // If C[idx+1] has atleast t keys, find the successor 'succ' of k in
    // the subtree rooted at C[idx+1]
    // Replace k by succ
    // Recursively delete succ in C[idx+1]
    else if  (C[idx+1]->n >= t)
    {
        int succ = getSucc(idx);
        keys[idx] = succ;
        C[idx+1]->remove(succ);
    }
  
    // If both C[idx] and C[idx+1] has less that t keys,merge k and all of C[idx+1]
    // into C[idx]
    // Now C[idx] contains 2t-1 keys
    // Free C[idx+1] and recursively delete k from C[idx]
    else
    {
        merge(idx);
        C[idx]->remove(k);
    }
    return;
}
  
// A function to get predecessor of keys[idx]
int BTreeNode::getPred(int idx)
{
    // Keep moving to the right most node until we reach a leaf
    BTreeNode *cur=C[idx];
    while (!cur->leaf)
        cur = cur->C[cur->n];
  
    // Return the last key of the leaf
    return cur->keys[cur->n-1];
}
  
int BTreeNode::getSucc(int idx)
{
  
    // Keep moving the left most node starting from C[idx+1] until we reach a leaf
    BTreeNode *cur = C[idx+1];
    while (!cur->leaf)
        cur = cur->C[0];
  
    // Return the first key of the leaf
    return cur->keys[0];
}
  
// A function to fill child C[idx] which has less than t-1 keys
void BTreeNode::fill(int idx)
{
  
    // If the previous child(C[idx-1]) has more than t-1 keys, borrow a key
    // from that child
    if (idx!=0 && C[idx-1]->n>=t)
        borrowFromPrev(idx);
  
    // If the next child(C[idx+1]) has more than t-1 keys, borrow a key
    // from that child
    else if (idx!=n && C[idx+1]->n>=t)
        borrowFromNext(idx);
  
    // Merge C[idx] with its sibling
    // If C[idx] is the last child, merge it with with its previous sibling
    // Otherwise merge it with its next sibling
    else
    {
        if (idx != n)
            merge(idx);
        else
            merge(idx-1);
    }
    return;
}
  
// A function to borrow a key from C[idx-1] and insert it
// into C[idx]
void BTreeNode::borrowFromPrev(int idx)
{
  
    BTreeNode *child=C[idx];
    BTreeNode *sibling=C[idx-1];
  
    // The last key from C[idx-1] goes up to the parent and key[idx-1]
    // from parent is inserted as the first key in C[idx]. Thus, the  loses
    // sibling one key and child gains one key
  
    // Moving all key in C[idx] one step ahead
    for (int i=child->n-1; i>=0; --i)
        child->keys[i+1] = child->keys[i];
  
    // If C[idx] is not a leaf, move all its child pointers one step ahead
    if (!child->leaf)
    {
        for(int i=child->n; i>=0; --i)
            child->C[i+1] = child->C[i];
    }
  
    // Setting child's first key equal to keys[idx-1] from the current node
    child->keys[0] = keys[idx-1];
  
    // Moving sibling's last child as C[idx]'s first child
    if(!child->leaf)
        child->C[0] = sibling->C[sibling->n];
  
    // Moving the key from the sibling to the parent
    // This reduces the number of keys in the sibling
    keys[idx-1] = sibling->keys[sibling->n-1];
  
    child->n += 1;
    sibling->n -= 1;
  
    return;
}
  
// A function to borrow a key from the C[idx+1] and place
// it in C[idx]
void BTreeNode::borrowFromNext(int idx)
{
  
    BTreeNode *child=C[idx];
    BTreeNode *sibling=C[idx+1];
  
    // keys[idx] is inserted as the last key in C[idx]
    child->keys[(child->n)] = keys[idx];
  
    // Sibling's first child is inserted as the last child
    // into C[idx]
    if (!(child->leaf))
        child->C[(child->n)+1] = sibling->C[0];
  
    //The first key from sibling is inserted into keys[idx]
    keys[idx] = sibling->keys[0];
  
    // Moving all keys in sibling one step behind
    for (int i=1; in; ++i)
        sibling->keys[i-1] = sibling->keys[i];
  
    // Moving the child pointers one step behind
    if (!sibling->leaf)
    {
        for(int i=1; i<=sibling->n; ++i)
            sibling->C[i-1] = sibling->C[i];
    }
  
    // Increasing and decreasing the key count of C[idx] and C[idx+1]
    // respectively
    child->n += 1;
    sibling->n -= 1;
  
    return;
}
  
// A function to merge C[idx] with C[idx+1]
// C[idx+1] is freed after merging
void BTreeNode::merge(int idx)
{
    BTreeNode *child = C[idx];
    BTreeNode *sibling = C[idx+1];
  
    // Pulling a key from the current node and inserting it into (t-1)th
    // position of C[idx]
    child->keys[t-1] = keys[idx];
  
    // Copying the keys from C[idx+1] to C[idx] at the end
    for (int i=0; in; ++i)
        child->keys[i+t] = sibling->keys[i];
  
    // Copying the child pointers from C[idx+1] to C[idx]
    if (!child->leaf)
    {
        for(int i=0; i<=sibling->n; ++i)
            child->C[i+t] = sibling->C[i];
    }
  
    // Moving all keys after idx in the current node one step before -
    // to fill the gap created by moving keys[idx] to C[idx]
    for (int i=idx+1; in += sibling->n+1;
    n--;
  
    // Freeing the memory occupied by sibling
    delete(sibling);
    return;
}
  
// The main function that inserts a new key in this B-Tree
void BTree::insert(int k)
{
    // If tree is empty
    if (root == NULL)
    {
        // Allocate memory for root
        root = new BTreeNode(t, true);
        root->keys[0] = k;  // Insert key
        root->n = 1;  // Update number of keys in root
    }
    else // If tree is not empty
    {
        // If root is full, then tree grows in height
        if (root->n == 2*t-1)
        {
            // Allocate memory for new root
            BTreeNode *s = new BTreeNode(t, false);
  
            // Make old root as child of new root
            s->C[0] = root;
  
            // Split the old root and move 1 key to the new root
            s->splitChild(0, root);
  
            // New root has two children now.  Decide which of the
            // two children is going to have new key
            int i = 0;
            if (s->keys[0] < k)
                i++;
            s->C[i]->insertNonFull(k);
  
            // Change root
            root = s;
        }
        else  // If root is not full, call insertNonFull for root
            root->insertNonFull(k);
    }
}
  
// A utility function to insert a new key in this node
// The assumption is, the node must be non-full when this
// function is called
void BTreeNode::insertNonFull(int k)
{
    // Initialize index as index of rightmost element
    int i = n-1;
  
    // If this is a leaf node
    if (leaf == true)
    {
        // The following loop does two things
        // a) Finds the location of new key to be inserted
        // b) Moves all greater keys to one place ahead
        while (i >= 0 && keys[i] > k)
        {
            keys[i+1] = keys[i];
            i--;
        }
  
        // Insert the new key at found location
        keys[i+1] = k;
        n = n+1;
    }
    else // If this node is not leaf
    {
        // Find the child which is going to have the new key
        while (i >= 0 && keys[i] > k)
            i--;
  
        // See if the found child is full
        if (C[i+1]->n == 2*t-1)
        {
            // If the child is full, then split it
            splitChild(i+1, C[i+1]);
  
            // After split, the middle key of C[i] goes up and
            // C[i] is splitted into two.  See which of the two
            // is going to have the new key
            if (keys[i+1] < k)
                i++;
        }
        C[i+1]->insertNonFull(k);
    }
}
  
// A utility function to split the child y of this node
// Note that y must be full when this function is called
void BTreeNode::splitChild(int i, BTreeNode *y)
{
    // Create a new node which is going to store (t-1) keys
    // of y
    BTreeNode *z = new BTreeNode(y->t, y->leaf);
    z->n = t - 1;
  
    // Copy the last (t-1) keys of y to z
    for (int j = 0; j < t-1; j++)
        z->keys[j] = y->keys[j+t];
  
    // Copy the last t children of y to z
    if (y->leaf == false)
    {
        for (int j = 0; j < t; j++)
            z->C[j] = y->C[j+t];
    }
  
    // Reduce the number of keys in y
    y->n = t - 1;
  
    // Since this node is going to have a new child,
    // create space of new child
    for (int j = n; j >= i+1; j--)
        C[j+1] = C[j];
  
    // Link the new child to this node
    C[i+1] = z;
  
    // A key of y will move to this node. Find location of
    // new key and move all greater keys one space ahead
    for (int j = n-1; j >= i; j--)
        keys[j+1] = keys[j];
  
    // Copy the middle key of y to this node
    keys[i] = y->keys[t-1];
  
    // Increment count of keys in this node
    n = n + 1;
}
  
// Function to traverse all nodes in a subtree rooted with this node
void BTreeNode::traverse()
{
    // There are n keys and n+1 children, travers through n keys
    // and first n children
    int i;
    for (i = 0; i < n; i++)
    {
        // If this is not leaf, then before printing key[i],
        // traverse the subtree rooted with child C[i].
        if (leaf == false)
            C[i]->traverse();
        cout << " " << keys[i];
    }
  
    // Print the subtree rooted with last child
    if (leaf == false)
        C[i]->traverse();
}
  
// Function to search key k in subtree rooted with this node
BTreeNode *BTreeNode::search(int k)
{
    // Find the first key greater than or equal to k
    int i = 0;
    while (i < n && k > keys[i])
        i++;
  
    // If the found key is equal to k, return this node
    if (keys[i] == k)
        return this;
  
    // If key is not found here and this is a leaf node
    if (leaf == true)
        return NULL;
  
    // Go to the appropriate child
    return C[i]->search(k);
}
  
void BTree::remove(int k)
{
    if (!root)
    {
        cout << "The tree is empty\n";
        return;
    }
  
    // Call the remove function for root
    root->remove(k);
  
    // If the root node has 0 keys, make its first child as the new root
    //  if it has a child, otherwise set root as NULL
    if (root->n==0)
    {
        BTreeNode *tmp = root;
        if (root->leaf)
            root = NULL;
        else
            root = root->C[0];
  
        // Free the old root
        delete tmp;
    }
    return;
}
  
// Driver program to test above functions
int main()
{
    BTree t(3); // A B-Tree with minium degree 3
  
    t.insert(1);
    t.insert(3);
    t.insert(7);
    t.insert(10);
    t.insert(11);
    t.insert(13);
    t.insert(14);
    t.insert(15);
    t.insert(18);
    t.insert(16);
    t.insert(19);
    t.insert(24);
    t.insert(25);
    t.insert(26);
    t.insert(21);
    t.insert(4);
    t.insert(5);
    t.insert(20);
    t.insert(22);
    t.insert(2);
    t.insert(17);
    t.insert(12);
    t.insert(6);
  
    cout << "Traversal of tree constructed is\n";
    t.traverse();
    cout << endl;
  
    t.remove(6);
    cout << "Traversal of tree after removing 6\n";
    t.traverse();
    cout << endl;
  
    t.remove(13);
    cout << "Traversal of tree after removing 13\n";
    t.traverse();
    cout << endl;
  
    t.remove(7);
    cout << "Traversal of tree after removing 7\n";
    t.traverse();
    cout << endl;
  
    t.remove(4);
    cout << "Traversal of tree after removing 4\n";
    t.traverse();
    cout << endl;
  
    t.remove(2);
    cout << "Traversal of tree after removing 2\n";
    t.traverse();
    cout << endl;
  
    t.remove(16);
    cout << "Traversal of tree after removing 16\n";
    t.traverse();
    cout << endl;
  
    return 0;
}

输出:

Traversal of tree constructed is
 1 2 3 4 5 6 7 10 11 12 13 14 15 16 17 18 19 20 21 22 24 25 26
Traversal of tree after removing 6
 1 2 3 4 5 7 10 11 12 13 14 15 16 17 18 19 20 21 22 24 25 26
Traversal of tree after removing 13
 1 2 3 4 5 7 10 11 12 14 15 16 17 18 19 20 21 22 24 25 26
Traversal of tree after removing 7
 1 2 3 4 5 10 11 12 14 15 16 17 18 19 20 21 22 24 25 26
Traversal of tree after removing 4
 1 2 3 5 10 11 12 14 15 16 17 18 19 20 21 22 24 25 26
Traversal of tree after removing 2
 1 3 5 10 11 12 14 15 16 17 18 19 20 21 22 24 25 26
Traversal of tree after removing 16
 1 3 5 10 11 12 14 15 17 18 19 20 21 22 24 25 26