Passing arguments in a BST insert function by extracting data using sscanf - c

Question

I am writing a program that reads students (id, name, surname, grade) from a text file (line by line) and then stores them to a Binary Search Tree by using id as a key. To read each line i use fgets() and to extract the words from the line is use sscanf().
struct TreeNode* root = NULL;
FILE *text;
char *id, *onoma, *epitheto, *word, *line;
onoma = (char *)malloc(20 * sizeof(char));
epitheto = (char *)malloc(30 * sizeof(char));
id = (char *)malloc(9 * sizeof(char));
float vathmos;
text = fopen("students.txt", "r");
if (text == NULL) {
printf("Cannot read from the file!");
exit(1);
}
This is the loop where the data are extracted for each student:
while (fgets(line, 50, text) != NULL) {
printf("%d \n", root);
sscanf(line, "%s %s %s %f", id, onoma, epitheto, &vathmos);
printf("%s %s %s %.3f \n", id, onoma, epitheto, vathmos);
root = Insert(root, id);
}
And this is the insert function for the node:
TreeNode *Insert(struct TreeNode* root, char *data) {
if (root == NULL) { // empty tree
root = CreateNewNode(data);
}
// if data to be inserted is lesser, insert in left subtree.
else if ((strcmp(data, root->id)) <= 0) {
root->left = Insert(root->left,data);
}
// else, insert in right subtree.
else if ((strcmp(data, root->id)) > 0) {
root->right = Insert(root->right,data);
}
return root;
}
When I insert nodes "by hand" e.g.:
root = Insert(root, "AY881159");
root = Insert(root, "AA564510");
root = Insert(root, "AB784123");
the program works and the nodes are created and the tree can be manipulated.
But when the tree is created in the fgets() loop by getting the data from the sscanf(), there is a problem. While the variables store the data correctly (that's why I have the printf() after the sscanf() to check this), the root seems to reset and only the last student is kept in the tree.
Any ideas?
The code for the nodes is:
typedef struct TreeNode {
char *id;
struct TreeNode *left;
struct TreeNode *right;
} TreeNode;
and
TreeNode *CreateNewNode(char *data) {
struct TreeNode *NewNode = (TreeNode *)malloc(sizeof(TreeNode));
NewNode->id = data;
NewNode->left = NewNode->right = NULL;
return NewNode;
}

Answer 1

The code you posted cannot be compiled, it makes it more difficult to answer questions.
You create all nodes in the loop from the same id buffer. You need to make a copy of the buffer, either when calling Insert of preferably in the CreateNewnode() function. You did not post the code for that, nor did you post the definition of type TreeNode. Here is a possibility:
TreeNode *CreatNewNode(const char *data) {
TreeNode *node = calloc(1, sizeof(*node));
if (node != NULL) {
node->id = strdup(data);
node->left = node->right = NULL;
}
return node;
}
There is no need to allocate the arrays for the parse phase, local char arrays are fine for this, but data you store to the tree should be duplicated so you can reuse the buffers from the parsing code. Make the argument to Insert a const char *data to indicate the buffer will not be modified, nor owned by the tree after the call.
You must pass extra information to scanf to prevent buffer overflow.
Here is a modified version of the calling code:
int main(void) {
struct TreeNode *root = NULL;
FILE *text;
char id[9], onoma[20], epitheto[30], line[256];
float vathmos;
text = fopen("students.txt", "r");
if (text == NULL) {
printf("Cannot read from the file!");
exit(1);
}
// This is the loop where the data are extracted for each student:
while (fgets(line, sizeof line, text) != NULL) {
printf("%d \n", root);
if (sscanf(line, "%8s %19s %29s %f", id, onoma, epitheto, &vathmos) == 4) {
printf("%s %s %s %.3f \n", id, onoma, epitheto, vathmos);
root = Insert(root, id);
} else {
printf("invalid line: %s", line);
}
// I'm curious how you are going to store the other data...
}
...
}
The Insert function can be simplified:
TreeNode *Insert(struct TreeNode *root, const char *data) {
if (root == NULL) { // empty tree
root = CreateNewNode(data);
} else {
if (strcmp(data, root->id) <= 0) {
// if data to be inserted is lesser or equal, insert in left subtree.
root->left = Insert(root->left, data);
} else {
// else insert in the right subtree
root->right = Insert(root->right, data);
}
}
return root;
}
A better API for InsertNode would be to take a pointer to the root pointer and return a pointer to the new node:
TreeNode *Insert(struct TreeNode **nodep, const char *data) {
while (*nodep != NULL) {
if (strcmp(data, (*nodep)->id) <= 0) {
nodep = &(*nodep)->left;
} else {
nodep = &(*nodep)->right;
}
}
return *nodep = CreateNewNode(data);
}