Question

I’ve been struggling with regular expressions in C (just /usr/include/regex.h).

---

I have (let’s say) hundreds of regexps and one of them can match input string.
Currently I’m doing it (generating it actually) like this: hundreds of do-while with match inside, break if not matching and going to another. One by one:

do {
    if ( regex_match(str, my_regex1) != MY_REGEX_SUCCESS ) DO_FAIL; //break
    ...
    if ( sscanf(str, " %d.%d.%d.%d / %d ", &___ip1, &___ip2, &___ip3, &___ip4, &___pref) != 5 ) DO_FAIL; //break
    ...
} while (0);

do {
    if ( regex_match(str, my_regex2) != MY_REGEX_SUCCESS ) DO_FAIL; //break
    ...
    ...
} while (0);

do {
    if ( regex_match(str, my_regex3) != MY_REGEX_SUCCESS ) DO_FAIL; //break
    ...
    ...
} while (0);

---

What I’d like to have is something like:

const char * match1 = "^([[:space:]]*)([$]([._a-zA-Z0-9-]{0,118})?[._a-zA-Z0-9])([[:space:]]*)$";
const char * match2 = "^([[:space:]]*)(target|origin)([[:space:]]*):([[:space:]]*)([$]([._a-zA-Z0-9-]{0,118})?[._a-zA-Z0-9])([[:space:]]*):([[:space:]]*)\\*([[:space:]]*)$";
const char * match3 = "^([[:space:]]*)(target|origin)([[:space:]]*):([[:space:]]*)([$]([._a-zA-Z0-9-]{0,118})?[._a-zA-Z0-9])([[:space:]]*)/([[:space:]]*)(([0-2]?[0-9])|(3[0-2]))([[:space:]]*):([[:space:]]*)(([1-9][0-9]{0,3})|([1-5][0-9]{4})|(6[0-4][0-9]{3})|(65[0-4][0-9]{2})|(655[0-2][0-9])|(6553[0-5]))([[:space:]]*)$";
char * my_match;
asprintf(&my_match, "(%s)|(%s)|(%s)", match1, match2, match3);


int num_gr = give_me_number_of_regex_group(str, my_match)
switch (num_gr) {
    ...
}

and don’t have an idea how to do that…

Any suggestions?
Thanks!

Answer 1

I assume your regex_match is some combination of regcomp and regexec. To enable grouping, you need to call regcomp with the REG_EXTENDED flag, but without the REG_NOSUB flag (in the third argument).

regex_t compiled;
regcomp(&compiled, "(match1)|(match2)|(match3)", REG_EXTENDED);

Then allocate space for the groups. The number of groups is stored in compiled.re_nsub. Pass this number to regexec:

size_t ngroups = compiled.re_nsub + 1;
regmatch_t *groups = malloc(ngroups * sizeof(regmatch_t));
regexec(&compiled, str, ngroups, groups, 0);

Now, the first invalid group is the one with a -1 value in both its rm_so and rm_eo fields:

size_t nmatched;
for (nmatched = 0; nmatched < ngroups; nmatched++)
    if (groups[nmatched].rm_so == (size_t)(-1))
        break;

nmatched is the number of parenthesized subexpressions (groups) matched. Add your own error checking.