Right algorithm but wrong implementation? - c

Question

I am a beginner in C with some experience in python and java. I want to solve a problem with C. The problem goes like this:
Take an input as a sentence with words separated by blank spaces only (assume lower case only), re-write the sentence with the following rules:
1) If a word occurs the first time, keep it the same.
2) If the word occurs twice, replace the second occurrence with the word being copied twice (e.g. two --> twotwo).
3) If the word occurs three times or more, delete all the occurrences after the second one.
Print the output as a sentence. The maximum lengths of the input sentence and each individual word is 500 chars and 50 chars.
Exemplary input: jingle bells jingle bells jingle all the way
Exemplary output: jingle bells jinglejingle bellsbells all the way
The approach I take is:
1) Read the input, separate each word and put them into an array of char pointers.
2) Use nested for loop to go through the array. For each word after the first word:
A - If there is no word before it that is equal to it, nothing happens.
B - If there is already one word before it that is equal to it, change the word as its "doubled form".
C - If there is already a "doubled form" of itself that exists before it, delete the word (set the element to NULL.
3) Print the modified array.
I am fairly confident about the correctness of this approach. However, when I actually write the code:
'''
int main()
{
char input[500];
char *output[500];
// Gets the input
printf("Enter a string: ");
gets(input);
// Gets the first token, put it in the array
char *token = strtok(input, " ");
output[0] = token;
// Keeps getting tokens and filling the array, untill no blank space is found
int i = 1;
while (token != NULL) {
token = strtok(NULL, " ");
output[i] = token;
i++;
}
// Processes the array, starting from the second element
int j, k;
char *doubled;
for (j = 1; j < 500; j++) {
strcpy(doubled, output[j]);
strcat(doubled, doubled); // Create the "doubled form"
for (k = 0; k < j; k++) {
if (strcmp(output[k], output[j]) == 0) { // Situation B
output[j] = doubled;
}
if (strcmp(output[k], doubled) == 0) { // Situation C
output[j] = ' ';
}
}
}
// Convert the array to a string
char *result = output[0]; // Initialize a string with the first element in the array
int l;
char *blank_space = " "; // The blank spaces that need to be addded into the sentence
for (l = 1; l < 500; l++) {
if (output[l] != '\0'){ // If there is a word that exists at the given index, add it
strcat(result, blank_space);
strcat(result, output[l]);
}
else { // If reaches the end of the sentence
break;
}
}
// Prints out the result string
printf("%s", result);
return 0;
}
'''
I did a bunch of tests on each individual block. There are several issues:
1) When processing the array, strcmp, strcat, and strcpy in the loop seem to give Segmentation fault error reports.
2) When printing the array, the words did not show the order that they are supposed to do.
I am now frustrated because it seems that the issues are all coming from some internal structural defects of my code and they are very much related to the memory mechanism of C which I am not really too familiar with. How should I fix this?

Answer 1

One issues jumps out at me. This code is wrong:
char *doubled;
for (j = 1; j < 500; j++) {
strcpy(doubled, output[j]);
strcat(doubled, doubled); // Create the "doubled form"
doubled doesn't point to any actual memory. So trying to copy data to where it points is undefined behavior and will almost certainly cause a SIGSEGV - and it will corrupt memory if it doesn't cause a SIGSEGV.
That needs to be fixed - you can't copy a string with strcpy() or strcat() to an pointer that doesn't point to actual memory.
This would be better, but still not ideal as no checking is done to ensure there's no buffer overflow:
char doubled[ 2000 ];
for (j = 1; j < 500; j++) {
strcpy(doubled, output[j]);
strcat(doubled, doubled); // Create the "doubled form"
This is also a problem with doubled defined like that:
if (strcmp(output[k], output[j]) == 0) { // Situation B
output[j] = doubled;
}
That just points output[j] at doubled. The next loop iteration will overwrite doubled, and the data that output[j] still points to to will get changed.
This would address that problem:
if (strcmp(output[k], output[j]) == 0) { // Situation B
output[j] = strdup( doubled );
}
strdup() is a POSIX function that, unsurprising, duplicates a string. That string will need to be free()'d later, though, as strdup() is the same as:
char *strdup( const char *input )
{
char *duplicate = malloc( 1 + strlen( input ) );
strcpy( duplicate, input );
return( duplicate );
}
As pointed out, strcat(doubled, doubled); is also a problem. One possible solution:
memmove(doubled + strlen( doubled ), doubled, 1 + strlen( doubled ) );
That copies the contents of the doubled string to the memory starting at the original '\0' terminator. Note that since the original '\0' terminator is part of the string, you can't use strcpy( doubled + strlen( doubled ), doubled );. Nor can you use memcpy(), for the same reason.

Answer 2

Your code invokes undefined behavior in several places by writing to memory that is not yet owned, and is likely the cause of the segmentation fault you are seeing. For example, in this segment:
char *result = output[0]; // Initialize a string with the first element in the array
int l;
char *blank_space = " "; // The blank spaces that need to be addded into the sentence
for (l = 1; l < 500; l++) {
if (output[l] != '\0'){ // If there is a word that exists at the given index, add it
strcat(result, blank_space);
strcat(result, output[l]);
By itself...
char *result = output[0]; //creates a pointer, but provides no memory.
...is not sufficient to receive content such as
strcat(result, blank_space);
strcat(result, output[l]);
It needs memory:
char *result = malloc(501);//or more space if needed.
if(result)
{
//now use strcat or strcpy to add content