## Introduction

Whenever we create trees, we need to be careful about memory allocations and de-allocation. Just out of a requirement, I thought of trying C++ STL (vectors and deques) to create a small BST. This article is the obvious consequence of that thought.

## Background

BSTs are one of the most widely used data structures. C is the preferred language, however with the advent of C++ especially with C++11, it seems to be interesting to use C++. However, nothing related to C++11 is used in this article and can be used by C++98 compilers.

## Using the Code

To create a BST, we need a BST data structure. Traditional BST data structures contain pointer to left and right sub-tree. Since I’ll be using vector and not pointers, I'll use vector indexes as pointer to left and right sub-tree.

struct bst
{
unsigned int data;
int leftIdx;
int rightIdx;
};

Now, I’ll write various functions used in creating the BST and its child nodes. The first function will be used to create a BST. It will initialize the data structure with incoming data and reset the left and right index with `-1`

(similar to making the pointers `NULL`

).

void MakeNode(vector<struct> &v1, int aData)
{
struct bst b1 = { aData, -1, -1 };
v1.push_back(b1);
}

The functions below will set the left and right child of a node. While setting, it assigns the correct index of left and right child on to the root node:

void setleft(vector <struct>&v1, int currIndex, int aData)
{
unsigned int leftIndex = v1.size();
v1[currIndex].leftIdx = leftIndex;
struct bst b1 = { aData, -1, -1 };
v1.push_back(b1);
}
void setright(vector<struct> &v1, int currIndex, int aData)
{
unsigned int rightIndex = v1.size();
v1[currIndex].rightIdx = rightIndex;
struct bst b1 = { aData, -1, -1 };
v1.push_back(b1);
}

The function below will be used to insert the data into the BST. Simple traversal to all the nodes till we find the appropriate place to insert the elements and call left and right functions defined above.

void Insert(vector<struct bst> &v1, int aData)
{
if(v1.size() == 0)
{
cout<<"Note is not made yet. MakeNode first..."<<endl;
return;
}
unsigned int currentIdx = 0;
while ( currentIdx < v1.size() )
{
if(aData <= v1[currentIdx].data)
{
if( v1[currentIdx].leftIdx == -1)
{
setleft(v1, currentIdx, aData);
break;
}
else
currentIdx = v1[currentIdx].leftIdx;
}
else
{
if(v1[currentIdx].rightIdx == -1)
{
setright(v1, currentIdx, aData);
break;
}
else
currentIdx = v1[currentIdx].rightIdx;
}
}
}

The code below will be used for traversing the BST in in-order, pre-order, and post-order. The index is used as a starting point.

void InTrav(vector <struct bst> &v1, unsigned int Idx)
{
if(v1[Idx].leftIdx != -1)
InTrav(v1,v1[Idx].leftIdx );
cout<<v1[Idx].data<<endl;
if( v1[Idx].rightIdx != -1)
InTrav(v1, v1[Idx].rightIdx);
}
void PreTrav(vector <struct bst> &v1, unsigned int Idx)
{
cout<<v1[Idx].data<<endl;
if(v1[Idx].leftIdx != -1)
PreTrav(v1,v1[Idx].leftIdx );
if( v1[Idx].rightIdx != -1)
PreTrav(v1, v1[Idx].rightIdx);
}
void PostTrav(vector <struct bst> &v1, unsigned int Idx)
{
if(v1[Idx].leftIdx != -1)
PostTrav(v1,v1[Idx].leftIdx );
if( v1[Idx].rightIdx != -1)
PostTrav(v1, v1[Idx].rightIdx);
cout<<v1[Idx].data<<endl;
}

The main program can be a simple one, for example:

int main()
{
vector <struct bst> v1;
MakeNode(v1, 30);
Insert(v1, 20);
Insert(v1, 6);
Insert(v1, 40);
Insert(v1, 35);
Insert(v1, 16);
Insert(v1, 7);
InTrav(v1, 0);
PreTrav(v1,0);
PostTrav(v1,0);
return 0;
}

## Points of Interest

- The same code can be used for STL deque also. I haven’t checked it for any other containers.
- The performance will be slow as compared to raw pointers unless some vector (STL) optimizations are done. I didn't try that.
- You can use it for small BST which you can delete in one shot without worrying about memory de-allocating.
- Deletion of an element in BST will be updated in the next post.

## History