Using 'regex.h'

[TheDenominater]

I'm trying to do some input parsing using <regex.h> , but I can't get it work quite the way I want.

In this case I'm trying to match a line that starts with any amount of whitespace, then a '#'. Right now I'm trying to use:

int isComment(char *line)
{
  regex_t* temp;
  temp = malloc(sizeof(regex_t));
  regcomp(temp, "[:blanc:]+#", 0);
  
  if (regexec(temp, line, 0, NULL , 0) == 0)
  {
    printf("%s", line);
    return 0;
  }
  else
    return 1;
 }

But this isn't working, any ideas why? Am I using :space: correctly?

Edit: I also want to do this to other lines, including some where I need to replace the original string with the new pattern, but can't figure out how to do this.

[cas]

There are a few problems:
1. It's spelled “blank”
2. Your [:blank:] needs to be inside of brackets, so: [[:blank:]]
3. + matches a + in basic regular expressions. Either use \+ (which means you need \\+ in a string literal to escape the \), or tell regcomp to use extended regular expressions with the REG_EXTENDED flag

In short, use either of the following:

regcomp(temp, "[[:blank:]]\\+#", 0);
regcomp(temp, "[[:blank:]]+#", REG_EXTENDED);

I've also got a couple of suggestions for using regex(3). You want to regfree() your regex_t or else you'll leak memory. You also don't need to make it a pointer and allocate space (unless you need it to live beyond the lifetime of the function, which you don't for this example). It just makes extra work for you to do it that way. Instead:

regex_t temp;
regcomp(&temp, "whatever", 0);
...
regfree(&temp);

[itCbitC]

I'm trying to do some input parsing using <regex.h> , but I can't get it work quite the way I want.

In this case I'm trying to match a line that starts with any amount of whitespace, then a '#'.

Does "any amount of whitespace" mean zero or more whitespace characters? If so the "+" should really be a "*", as in

pattern == "^[[:blank:]]*#";

The code segment below is executed inside of isComment() every time a line of input is read.

  regex_t* temp;
  temp = malloc(sizeof(regex_t));
  regcomp(temp, "[:blanc:]+#", 0);

Instead compile the regexp once and execute many times by moving the code into the function that calls isComment(), as in

  regex_t *temp = (regex_t *) malloc(sizeof(regex_t));
  regcomp(temp, "^[[:blank:]]*#", 0);
  isComment(temp, line);

[TheDenominater]

Thanks for your help.

So what I've got now is:

int isComment(char* line)
{
  regex_t temp;
  regcomp(&temp, "[:blanc:]*#",REG_EXTENDED);
  if ( regexec(&temp, line, 0, NULL , 0) == 0)
  {
    regfree(&temp);
    printf("%s", line);
    return 0;
  }
  else
    {
    regfree(&temp);
    return 1;
    }
 }

(I use the function in so many different places I'm not sure I should take the declaration out of it, it would get really messy..)

However, the second regfree() causes a segfault. I'm pretty sure it needs to be there. Is that correct? And I also can't get the '^' to work.

[itCbitC]

Did you not read the post by cas. He pointed out the typo ie "blank" NOT "blanc".
Don't regfree() temp as it's a local variable and will be freed automatically by the stack popping.

[MK27]

Don't regfree() temp as it's a local variable and will be freed automatically by the stack popping.

I would not count on that as it is used by regcomp, and the regex.h docs I have seen recommend freeing it in all cases.

However, I don't use regex.h that much, maybe someone knows the sure answer to this one -- I guess a valgrind test or something will tell you.

[cas]

You still have multiple errors that I've pointed out to you.

At the very least, check the return value of regcomp(). You're proceeding whether or not it returns error, and since your regular expression is invalid, it will be returning an error.

[TheDenominater]

Oh, not sure how I missed that. I still can't get the anchor '^' to work. I'm using:

regcomp(&temp, "^[:blank:]*#",REG_EXTENDED);

When I include '^' it misses every line for some reason.

[cas]

I would not count on that as it is used by regcomp, and the regex.h docs I have seen recommend freeing it in all cases.

However, I don't use regex.h that much, maybe someone knows the sure answer to this one -- I guess a valgrind test or something will tell you.

You're correct. regcomp() will allocate memory, and that memory needs to be freed regardless of where the regex_t was allocated. It's just like storing the return value of malloc() into a local pointer. The pointer itself is destroyed automatically, but what it points to needs to be manually freed.

[itCbitC]

Oh, not sure how I missed that. I still can't get the anchor '^' to work. I'm using:

regcomp(&temp, "^[:blank:]*#",REG_EXTENDED);

When I include '^' it misses every line for some reason.

It's not the anchor but as cas pointed out it's your regexp syntax that is totally wrong.
Read "[[:blank:]]" NOT "[:blank:]". Do you see the difference??

[Subsonics]

Instead compile the regexp once and execute many times by moving the code into the function that calls isComment(), as in

  regex_t *temp = (regex_t *) malloc(sizeof(regex_t));
  regcomp(temp, "^[[:blank:]]*#", 0);
  isComment(temp, line);

That's a good advice, it seems like an unecessary amount of work to do every time the function is called, especially if it's called from with in a loop.

[TheDenominater]

Ok, I re-read all the posts to make sure I didn't miss anything.

int isComment(char* line)
{
  regex_t temp;
  
  if(regcomp(&temp, "^[[:blank:]]*#",REG_EXTENDED) != 0)
  {
    printf("Incorrect regex\n");
    exit(EXIT_FAILURE);
  }

  if ( regexec(&temp, line, 0, NULL , 0) == 0)
  {
    regfree(&temp);
    printf("%s", line);
    return 0;
  }
  else
  {
    regfree(&temp);
    return 1;
  }
 }

The only problem is that regfree(&temp) in the else causes big problems. Am I not doing that correctly?

[itCbitC]

I would not count on that as it is used by regcomp, and the regex.h docs I have seen recommend freeing it in all cases.

However, I don't use regex.h that much, maybe someone knows the sure answer to this one -- I guess a valgrind test or something will tell you.

Whoops! I was thinking stack allocation but you're correct regfree() complements regcomp().

[itCbitC]

Perhaps then have only one regfree() placed right after the call to regexec() as in

int isComment(char* line)
{
    int rc;
    regex_t temp;
  
    if(regcomp(&temp, "^[[:blank:]]*#",REG_EXTENDED) != 0)
    {
        printf("Incorrect regex\n");
        exit(EXIT_FAILURE);
    }

    rc = regexec(&temp, line, 0, NULL , 0);
    regfree(&temp);
  
    if (rc == 0) {
        printf("%s", line);
        return 0;
    }
}