Open In App

Level order traversal in spiral form using stack and multimap

Improve
Improve
Like Article
Like
Save
Share
Report

Given a binary tree of N nodes, the task is to print level order traversal in a spiral form. In spiral form, nodes at the first and second level of tree are printed normally (left to right), after which nodes at alternate levels are printed in reverse order. Examples:

Input: N = 3

      1
     / \
    3   2

Output: 1 3 2 Explanation: Nodes at level 0 printed in normal order (1) Nodes at level 1 printed in normal order (3, 2) Hence, spiral order is (1, 3, 2) Input: N = 5

       10
      /  \
     20  30
    /  \
   40  60

Output: 10 20 30 60 40 Explanation: Nodes at level 0 printed in normal order (10) Nodes at level 1 printed in normal order (20, 30) Nodes at level 2 printed in reverse order (60, 40) Hence, spiral order is (10, 20, 30, 60, 40)

Naive Approach: A naive approach for this problem has already been discussed in this article. The basic idea is to use recursion and a flag variable, using which nodes of alternate levels are printed in reverse order and finally spiral form is obtained. Time complexity: O(N2) Auxiliary Space: O(1) Efficient Approach: In this approach, stack and multimap are used. A multimap container in C++ stores the (key, value) pairs in ascending order, sorted according to the key. For every node of a given tree, if we put (level, node) in a multimap, then it will store these nodes sorted according to their level. For example, a given tree is: For this tree, the multimap would look like:

Key(Level)     Value(Element)
0               1
1               2
1               3
2               7
2               6
2               5
2               4

Detailed steps of this approach are as follows:

  1. Traverse the given tree and insert all (level, node) pairs in a multimap and then traverse this multimap.
  2. If the level is odd, print the nodes in order in which they are present in the multimap.
  3. If the level is even, push all elements of the current level to a stack then pop all elements from the stack and print them. It gives the reverse order.

Finally, this level order traversal will result in the required spiral form. Below is the implementation of the above approach: 

CPP




// C++ implementation of
// the above approach
 
#include <bits/stdc++.h>
using namespace std;
 
// Tree Node
struct Node {
    int data;
    Node* left;
    Node* right;
};
 
// Utility function to
// create a new Tree Node
Node* newNode(int val)
{
    Node* temp = new Node;
    temp->data = val;
    temp->left = NULL;
    temp->right = NULL;
 
    return temp;
}
 
void printSpiral(Node* root);
 
// Function to build tree
// from given nodes
Node* buildTree(string str)
{
    // Corner Case
    if (str.length() == 0
        || str[0] == 'N')
        return NULL;
 
    // Vector to store nodes
    // after splitting space
    vector<string> ip;
 
    istringstream iss(str);
    for (string str; iss >> str;)
        ip.push_back(str);
 
    // Creating root of the tree
    Node* root = newNode(stoi(ip[0]));
 
    // Push the root to the queue
    queue<Node*> queue;
    queue.push(root);
 
    // Start from second element
    int i = 1;
    while (!queue.empty()
        && i < ip.size()) {
 
        // Get and remove the
        // front of the queue
        Node* currNode = queue.front();
        queue.pop();
 
        // Get the current node's
        // value from the string
        string currVal = ip[i];
 
        // If left child is not null
        if (currVal != "N") {
 
            // Create the left child
            // for the current node
            currNode->left = newNode(stoi(currVal));
 
            // Push it to the queue
            queue.push(currNode->left);
        }
 
        // For the right child
        i++;
        if (i >= ip.size())
            break;
        currVal = ip[i];
 
        // If the right child is not null
        if (currVal != "N") {
 
            // Create the right child
            // for the current node
            currNode->right = newNode(stoi(currVal));
 
            // Push it to the queue
            queue.push(currNode->right);
        }
        i++;
    }
 
    return root;
}
 
// Globally defined multimap
multimap<int, int> m;
 
// Function to fill multimap
void fillMultiMap(Node* root, int level)
{
 
    if (root == NULL)
        return;
 
    else {
        m.insert(pair<int, int>(
            level, root->data));
        fillMultiMap(
            root->left, level + 1);
        fillMultiMap(
            root->right, level + 1);
    }
}
 
void printSpiral(Node* root)
{
    m.clear();
    fillMultiMap(root, 0);
    stack<int> s;
 
    map<int, int>::iterator it
        = m.begin();
 
    // Traversing multimap
    while (it != m.end()) {
 
        // If level is odd
        if ((it->first) % 2 != 0) {
 
            // Printing same order
            while (!s.empty()) {
                cout << s.top() << " ";
                s.pop();
            }
            cout << it->second << " ";
        }
 
        // Otherwise, pushing to stack
        else {
            s.push(it->second);
        }
        it++;
    }
 
    // Pop from stack
    // to get reverse order
    while (!s.empty()) {
        cout << s.top() << " ";
        s.pop();
    }
 
    return;
}
 
// Driver code
int main()
{
 
    // Tree input
    string s = "1 2 3 7 6 5 4";
 
    // Build tree form given nodes
    Node* root = buildTree(s);
 
    // Print spiral form
    printSpiral(root);
 
    return 0;
}


Python3




# Python implementation of
# the above approach
from typing import List
import queue
import collections
 
 
# Tree Node
class Node:
    def __init__(self, val: int = 0, left: 'Node' = None, right: 'Node' = None):
        self.data = val
        self.left = left
        self.right = right
 
 
def print_spiral(root: Node) -> None:
    m = collections.defaultdict(list)
 
    def fill_multimap(root: Node, level: int) -> None:
        if root is None:
            return
 
        m[level].append(root.data)
        fill_multimap(root.left, level + 1)
        fill_multimap(root.right, level + 1)
 
    fill_multimap(root, 0)
     
    # Traversing multimap
    for level, level_nodes in m.items():
        # If level is odd
        if level % 2 == 0:
            level_nodes.reverse()
 
        print(" ".join(str(x) for x in level_nodes), end=" ")
 
    return
 
 
# Function to build tree from given nodes
def build_tree(nodes: List[str]) -> Node:
    n = len(nodes)
    # Corner Case
    if n == 0 or nodes[0] == 'N':
        return None
 
    # Creating root of the tree
    root = Node(int(nodes[0]))
     
    # Push the root to the queue
    q = queue.Queue()
    q.put(root)
 
    # Start from second element
    i = 1
    while not q.empty() and i < n:
        # Get and remove the
        # front of the queue
        curr_node = q.get()
         
        # If left child is not null
        if nodes[i] != 'N':
            # Create the left child
            # for the current node
            curr_node.left = Node(int(nodes[i]))
             
            # Push it to the queue
            q.put(curr_node.left)
             
        # For the right child
        i += 1
 
        if i >= n:
            break
         
        # If the right child is not null
        # to get reverse order
        if nodes[i] != 'N':
            # Create the right child
            # for the current node
            curr_node.right = Node(int(nodes[i]))
            # Push it to the queue
            q.put(curr_node.right)
        i += 1
 
    return root
 
 
# Driver code
if __name__ == '__main__':
    # Tree input
    s = "1 2 3 7 6 5 4".split()
 
    # Build tree from given nodes
    root = build_tree(s)
 
    # Print spiral form
    print_spiral(root)
 
# This code is contributed by Utkarsh


Output:

1 2 3 4 5 6 7

Time complexity: O(NlogN) Auxiliary Space: O(N)



Last Updated : 26 Apr, 2023
Like Article
Save Article
Previous
Next
Share your thoughts in the comments
Similar Reads