Question

With this structure definition:

struct word {
    char *cont; //content of the word
    char *wsp; //whitespace following the word
    int ctr;
};

for a word, I’m trying to write a function to get the first word from stdin with all the whitespace following it. Here it is:

struct word *getword(){
    char cont[WORDLIM];
    char wsp[WORDLIM];
    cont[0] = '\0';
    wsp[0] = '\0';
    if (peekchar() == EOF) return NULL; //peekchar defined elsewhere as getting a char and ungetc-ing it

    REPEAT{ //macro for for(;;)
            char c = getchar();
            char buf[2];
            buf[0]=c;
            buf[1]='\0';
            if (c == '\n' || c == ' '){
                    strcat(wsp, buf);
                    if (peekchar() != '\n' && peekchar() != ' '){
                            struct word *toret;
                            toret = malloc(sizeof(struct word));
                            toret->cont = cont;
                            toret->wsp = wsp;
                            toret->ctr = -1;
                            printf("---%s---\n", toret->wsp); //just for debugging
                            return toret;
                    }
                    continue;
            }
    }
    printf("PANIC PANIC PANIC THIS IS GOING WROOOOONG!!!!!\n");
    return NULL;
}

Now the fun thing is that I can get the correct output of whitespace in the just for debugging line, but when I attempt to access getword()->wsp, I get random garbage. Even more interestingly, getword()->cont works…

I’m a really new newbie to C…what did I do wrong?

Answer 1

REPEAT{ //macro for for(;;)

Lose that macro post-haste. Using the preprocessor to de-uglify C syntax is a recipe for heartburn. Just use for(;;) or while(1). The compiler is usually smart enough to turn it into an unconditional jump.

As for your problem:

toret->cont = cont;
toret->wsp = wsp;

The cont and wsp arrays are declared locally to the getword function; once the function exits, the arrays cease to exist, and any pointers to them are no longer valid. What you’ll need to do is allocate memory for the cont and wsp members of toret in addition to the memory for toret itself.

toret = malloc(sizeof(struct word));
if (toret) // ALWAYS check the result of a `malloc` call
{
  toret->cont = malloc(strlen(cont) + 1);
  if (toret->cont)
    strcpy(toret->cont, cont);
  toret->wsp = malloc(strlen(wsp) + 1);
  if (toret->wsp)
    strcpy(toret->wsp, wsp);
  ...
}

Note that when you need to free the memory for the struct, you’ll first need to free the memory for the cont and wsp members.