Well, I've been at it for a while...trying to figure out an algorithm to insert my list of random numbers into a binary tree.
This is what I have gotten so far:
NodePtr and Tree are pointers to a node
NodePtr CreateTree(FILE * fpData)
{
int in;
fscanf(fpData, "%i", &in);
Tree T开发者_运维百科 = (NodePtr)malloc(sizeof(Node));
T->Left = NULL;
T->Right = NULL;
T->value = in;
while((fscanf(fpData, "%i", &in)) != EOF)
{
InsertInTree(in, T);
printf("\n %p", T);
}
return T;
}
void InsertInTree(int value,Tree T)
{
if(T == NULL)
{
T->Left = (NodePtr)malloc(sizeof(Node));
T->Left->Left = NULL;
T->Left->Right = NULL;
T->Left->value = value;
printf("\n %i ", value);
return;
}
if(T->Left == NULL)
{
InsertInNull(value, T->Left);
}
else if(T->Right == NULL)
{
InsertInNull(value, T->Right);
}
else
{
if(T->Left->Left == NULL || T->Left->Right == NULL) InsertInTree(value, T->Left);
else InsertInTree(value, T->Right);
}
}
I'm lost on what to do if the both children of a particular node are not null. What I did here works for a small amount of numbers (1,2,3,5,6) but if the list is larger it becomes unbalanced and wrong.
Is it meant to be a search-tree? I don't see any if (value < T->Value)
conditions.
And you have an InsertNull (not shown). That shouldn't be necessary, 1 function should be enough.
To address your main problem, use a pointer-to-pointer parameter or, more elegant, always return a new Tree:
//untested, no balancing
Tree InsertValue(Tree t, int value)
{
if (t == null)
t = // create and return new node
else
{
if (value < t->Value)
t->Left = InsertValue(t->Left, value);
else
t->Right = InsertValue(t->Left, value);
}
return t;
}
And in CreateTree:
Tree t = InsertValue(null, in);
Since the assignment is not for a sorted tree, you can populate it in a breadth-first manner. This means the first thing inserted is always the root, the next is the first node at the next level so it looks like this:
0
1 2
3 4 5 6
Here is an article that explains it further:
http://www.cs.bu.edu/teaching/c/tree/breadth-first/
Simple insertion in a binary tree and keeping a binary tree balanced are different problems. I suggest you start with the first problem and just focus on keeping order properties correct within the tree. Your are not far from that.
Then you should have a look at classical implementations for red-black trees, well studied and efficient way of keeping trees balanced, but with a cost, it's more complex.
精彩评论