Thread: How get all numbers by regex?

How get all numbers by regex?

If I input string like "34df4354sdf234ds23", after regexec I get only:
submatch: 1
match 0: 34


How get something like this:
submatch: 4
match 0: 34
match 0: 4354
match 0: 234
match 0: 23

???

Code:

    
    regex_t p;
    regmatch_t *pmatch;
    int rcomp_err, rexec_err;
    char string[BUFSIZ+1];
    int i;

    rcomp_err = regcomp(&p, "([0-9])*",
                REG_EXTENDED | REG_NEWLINE);

    pmatch = alloca(sizeof(regmatch_t) * (p.re_nsub+1));
    if(!pmatch) {
        perror("alloca");
    }

    printf("Input a string: ");
    fgets(string, sizeof(string), stdin);

    rexec_err = regexec(&p, string, p.re_nsub, pmatch, 0);
    printf("submatch: %i\n",p.re_nsub);
    if(rexec_err) {
        printf("errrror");
    } else {
        /* match succeeded */
        for(i = 0; i < p.re_nsub; i++) {
            /* print the matching portion(s) of the string */
            if(pmatch[i].rm_so != -1) {
                char * submatch;
                size_t matchlen = pmatch[i].rm_eo - pmatch[i].rm_so;
                submatch = malloc(matchlen+1);
                strncpy(submatch, string+pmatch[i].rm_so, matchlen);
                submatch[matchlen] = '\0';
                printf("match %i: %s\n", i, submatch);
                free(submatch);
            }
        }
    }

regexec() will not find multiple instances of your parenthesized expression. There are multiple elements in pmatch because you can have multiple sets of parentheses, not for any other reason.

What it comes down to is that you'll have to keep calling regexec() until you run out of matches. Keep a pointer to the string you're searching. When you find a match, extract it, then advance the pointer past that match. Call regexec() again with the new value that is past the old match.

For your code you don't need to use parentheses at all. pmatch's first element (assuming a match was found) will be the substring that matched the entire regular expression. And speaking of regular expressions, you want [0-9]+, not [0-9]*, being that the latter will always match (the string "abc", after all, does start with zero or more digits).

You are completely overthinking this.

Code:

regex_t re;
char file[] = "hello777world99yew 8 I8909do23!";
const char *p = file;
regmatch_t match;

if(regcomp(&re, "[0-9]+", REG_EXTENDED) != 0) exit(1);

while(regexec(&re, p, 1, &match, 0) == 0)
{
  printf("%.*s\n", (int)(match.rm_eo - match.rm_so), &p[match.rm_so]);
  p += match.rm_eo; /* or p = &p[match.rm_eo]; */
}

That's all you need. Inside the loop, rm_eo - rm_so is the length of the match, and &p[rm_so] is the start of the match. The entire match (the whole expression) is always the first element in “match”.

As I told you before, you don't need parentheses for this. You only care about whether the entire expression matches.

Originally Posted by cas

You are completely overthinking this.

Code:

regex_t re;
char file[] = "hello777world99yew 8 I8909do23!";
const char *p = file;
regmatch_t match;

if(regcomp(&re, "[0-9]+", REG_EXTENDED) != 0) exit(1);

while(regexec(&re, p, 1, &match, 0) == 0)
{
  printf("%.*s\n", (int)(match.rm_eo - match.rm_so), &p[match.rm_so]);
  p += match.rm_eo; /* or p = &p[match.rm_eo]; */
}

That's all you need. Inside the loop, rm_eo - rm_so is the length of the match, and &p[rm_so] is the start of the match. The entire match (the whole expression) is always the first element in “match”.

As I told you before, you don't need parentheses for this. You only care about whether the entire expression matches.

THANKS for example!!!