char* function returning Null - c

Question

I am currently working on a spam filter program that takes in and reads text files. In the initializeTraining function, I call the preprocess function which reads in each string from each line in a given text-file.
Once the newDict function is executed from the line first=newDict(string, NULL);, the program, however, returns an error stating that there is a load of null pointer of type 'char' at line while(string[i] !='\0' && i<WORDLENGTH) { in the newDict function.
It seems that the preprocess function is returning null pointers despite the fact that it still takes in the passed-in strings from the text file. Is there something that I'm doing wrong in the preprocess function?
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
/*
WORDLENGTH is the length of the word in the linked list named as dictionary
MAILSEPARATOR is the totken to differentiate the mails that are included in one file.
it is also the token to update updated in linkedlist
*/
#define WORDLENGTH 20
#define MAILSEPARATOR "#####"
/*
define DEBUG as 0 to disable debug mode and to 1 to enable the mode.
*/
#define DEBUG 0
typedef struct dictionary dict;
typedef dict* word_dict;
typedef enum {false, true} bool;
/*
linked list, count is for the total word count and
occur is the numbers of the mails that had the word
*/
struct dictionary{
char word[WORDLENGTH];
int occur;
int count;
word_dict next;
bool updated;
};
// if there is no matching words after searching, create a new node
word_dict newDict(char *string, word_dict next){
word_dict target = (word_dict)malloc(sizeof(dict));
int i = 0;
while(string[i] !='\0' && i<WORDLENGTH) {
target->word[i] = string[i];
i++;
}
target->count = 1;
target->next = next;
target->occur = 1;
target->updated = true;
return target;
}
/*
preprocessor, convert string to lowercase
and trim the puctuations at the back
*/
char* preprocess(char* string){
#if DEBUG
printf("\nbefore preprocess, string: %s \n", string);
#endif
int i=0;
while(string[i] != '\0') { // convert to lower case
if (string[i] >= 65 && string[i] < 90) {
string[i] += 32;
i++;
}
while(true) {
i--;
if(i < 0) {
#if DEBUG
printf("word of only punctuations \n");
#endif
return NULL;
} else if((string[i] >= 97 && string[i] <= 122) || (string[i] >= 48 && string[i] <= 57)){
string[i+1]='\0';
break;
}
}
i=0;
while(true) {
if ((string[i] >= 97 && string[i] <= 122) || (string[i] >= 48 && string[i] <= 57)){
break;
} else {
string = &string[i+1];
}
i++;
}
}
#if DEBUG
printf("_after preprocess, string: %s\n", string);
#endif
return string;
}
/*
initialize training
reads the sample mails and creates a linked list of
the percentages of the words occuring in the sample mails
*/
word_dict initializeTraining(char* filename){
FILE *fp = NULL;
fp = fopen(filename, "r");
if(fp == NULL) {
printf("no file found\n");
return NULL;
}
char* string;
string = (char*)malloc(sizeof(char)*50);
word_dict first = NULL;
fscanf(fp, "%s\n", string);
string = preprocess(string);
first = newDict(string, NULL);
while(fscanf(fp,"%s", string) == 1) {
first = searchDict(string, first);
}
fclose(fp);
free(string);
return first;
}
/*
tests whether the mail is pam or not
takes the filename of the test mail,
returns true or false depending on the email's content
*/
bool bayesian_spam_filter(char * filename_for_test_email) {
word_dict spamDict=initializeTraining("spam.txt");
word_dict nonspamDict=initializeTraining("not_spam.txt");
#if DEBUG
printDict(spamDict);
printDict(nonspamDict);
#endif
FILE *stream=NULL;
stream = fopen(filename_for_test_email, "r");
if(stream == NULL){
printf("no file found\n");
return false;
}
char* string;
string = (char*)malloc(sizeof(char)*50);
int ps, pn; // probability of spam mail and non-spam mail
double prob = 0.5;
while(fscanf(stream,"%s", string) == 1){
char* tempString; // for handling the errors happening from string being null during preprocessing
tempString = preprocess(string);
if(tempString == NULL){
continue;
}
if((ps = searchTest(tempString, spamDict)) != 0) {
if((pn = searchTest(tempString, nonspamDict)) != 0) {
printf("ps:%3d, pn:%3d, %s\n", ps, pn, tempString);
prob = prob * (double) ps / ((prob* (double)ps + (1 - prob) * (double) pn));
printf("this prob: %.10f\n", prob);
}
}
}
//printf("%d, %d \n", pSProduct, pNProduct);
//proba=(float)(pSProduct/(pSProduct+pNProduct));
printf("Probability of mail being spam: %.10f\n", prob);
fclose(stream);
free(string);
if (prob > 0.9) {
return true;
}
return false;
}

Answer 1

It seems that the preprocess function is returning null pointers
Hardly surprising when it contains a line return NULL;. At that point, you should instead set the first character of the string to '\0' and return it since the surrounding code expects to get the string returned in all cases.
Another problem can be found in this section:
i=0;
while(true) {
if ((string[i] >= 97 && string[i] <= 122) || (string[i] >= 48 && string[i] <= 57)){
break;
} else {
string = &string[i+1];
}
i++;
}
Let's say the string is ".a" and go through the loop. Since the first character is not a letter, we do not break, instead we update the string pointer so that it points to "a" now. Then, we increment i. In the next iteration, string[i] is the null terminator, which is not a letter, so we continue. Since we are past the string's data, what follows is undefined behavior.
The simple fix for this is to not increment i and rather stick with [0] since you always want to remove from the beginning. The proper fix is to use i but without incrementing the string pointer, since you want to free it later – you must call free on the pointer returned by malloc, therefore modifying the pointer leads to undefined behavior! Instead of returning the string from preprocess, return the offset from the start (as counted by i), this way freeing the string later will work properly. The calling code would then look like this:
int offset = preprocess(string);
first = newDict(string + offset, NULL);