regex_match fails to find square brackets

Question

I am trying to do regex_match on a string which have square brackets([...]) inside it.

Things I have tried so far:

Normal matching
Backslashing the square brackets with 1 slash
Backslashing the square brackets with 2 slashes

Code to repro:

#include <iostream>
#include <cstring>
#include <regex>

using namespace std;

int main () {
  std::string str1 = "a/b/c[2]/d";
  std::string str2 = "(.*)a/b/c[2]/d(.*)";
  std::regex e(str2);

  std::cout << "str1 = " << str1 << std::endl;
  std::cout << "str2 = " << str2 << std::endl;
  if (regex_match(str1, e)) {
    std::cout << "matched" << std::endl;
  }
}

This is the error message I get every time I compile it.

terminate called after throwing an instance of 'std::regex_error'
what():  regex_error
Aborted (core dumped)

I was told by stack overflow members that gcc 4.8 or earlier version of it are known to be buggy. So, I needed to update it to latest version.

I have created an Ideone fiddle where compiler should not be issue. Even there, I do not see regex_match happening.

@WiktorStribiżew, Thread model: posix gcc version 4.8.2 (Ubuntu 4.8.2-19ubuntu1) — Hemant Bhargava, Apr 13 '16 at 11:50
So, all the answers are not helpful. Just upgrade your gcc compiler to some recent version. — Wiktor Stribiżew, Apr 13 '16 at 11:51
@WiktorStribiżew, It might be. But then should'nt it work on ideone.com. I tested code there and it did not match at all. Btw, I did not get any regex_error there. At : https://ideone.com/fJEKAF — Hemant Bhargava, Apr 13 '16 at 11:56
See [this demo](https://ideone.com/LfZ09p), you made some mistakes in the code. — Wiktor Stribiżew, Apr 13 '16 at 11:57
@WiktorStribiżew, where are you using square brackets in your regex/strings to be compare? What you wrote it just a plain and simple regex which btw works with 4.8 as well. — Hemant Bhargava, Apr 13 '16 at 13:18
If you need to actually match square brackets, you do need to escape them. Please see my answer below, and if there are any questions, please drop a comment there. — Wiktor Stribiżew, Apr 13 '16 at 13:34
@To SO users: This is not a dupe of [*Is gcc 4.8 or earlier buggy about regular expressions?*](http://stackoverflow.com/questions/12530406/is-gcc-4-8-or-earlier-buggy-about-regular-expressions), please consider reopening. — Wiktor Stribiżew, Apr 15 '16 at 08:22

Wiktor Stribiżew · Accepted Answer · 2016-04-13T13:29:56.280

2

The main problem you have is the outdated gcc compiler: you need to upgrade to some recent version. 4.8.x just does not support regex as it should.

Now, the code you should be using is:

#include <iostream>
#include <cstring>
#include <regex>

using namespace std;

int main () {
    std::string str1 = "a/b/c[2]/d";
    std::string str2 = R"(a/b/c\[2]/d)";
    std::regex e(str2);

    std::cout << "str1 = " << str1 << std::endl;
    std::cout << "str2 = " << str2 << std::endl;
    if (regex_search(str1, e)) {
        std::cout << "matched" << std::endl;
    }
}

See the IDEONE demo

Use

regex_search instead of regex_match to search for partial matches (regex_match requires a full string match)
The [2] in the regex pattern matches a literal 2 ([...] is a character class matching 1 character from the range/list specified in the character class). To match the literal square brackets, you need to escape the [ and you do not have to escape ]: R"(a/b/c\[2]/d)".

edited Apr 13 '16 at 13:29

answered Apr 13 '16 at 12:03

Wiktor Stribiżew

607,720
39
448
563

where are you using square brackets in your regex/strings to be compare? What you wrote it just a plain and simple regex which btw works with 4.8 as well – Hemant Bhargava Apr 13 '16 at 13:19
1

Completely make sense after you introduced slashes in your code. Thanks! – Hemant Bhargava Apr 13 '16 at 15:38
You may read more on what characters should be escaped for the regex engine to treat them literally [here](http://regular-expressions.info/characters.html). – Wiktor Stribiżew Apr 13 '16 at 15:45
Glad I could help. If my answer proved useful, please also consider upvoting. – Wiktor Stribiżew Apr 13 '16 at 21:36

score 0 · Answer 2 · answered Apr 13 '16 at 11:12

0

Well they should definitely be escaped by using a backslash. Unfortunately since backslash is itself special in a literal string you need two backslashes. So the regex should look like "(.*)a/b/c\\[2\\]/d(.*)".

answered Apr 13 '16 at 11:12

Some programmer dude

400,186
35
402
621

I am afraid to tell it do not work. – Hemant Bhargava Apr 13 '16 at 11:16
The forward slashes might also need escaping... ``"\\/"`` – BitTickler Apr 13 '16 at 11:21
@BitTickler. Nope that is not true. Write this code without square brackets and it would match. Even with forward slashes. – Hemant Bhargava Apr 13 '16 at 11:23
On https://regex101.com/, this one seems to do the job: ``(.*)a\/b\/c\[2\]\/d(.*)`` Some more backslashing added to accomodate the c-string literal stuff and you might be there. – BitTickler Apr 13 '16 at 11:23
@BitTickler, For c++? It might work for perl/python. – Hemant Bhargava Apr 13 '16 at 11:24
is that not an ISO standard or something? Lol - I never use regex and assumed they all work the same everywhere. – BitTickler Apr 13 '16 at 11:25
The error I get when testing is indeed `std::regex_constants::error_escape`, but I'm not sure *what* needs to be escaped. – Some programmer dude Apr 13 '16 at 11:29
Oh and for those who wonders, it's not the `std::regex_match` call that throws, but the `std::regex` object construction. – Some programmer dude Apr 13 '16 at 11:31

score 0 · Answer 3 · answered Apr 13 '16 at 11:23

0

Raw string literals often simplify cases where one would otherwise have to have complex escape sequences:

#include <iostream>
#include <cstring>
#include <regex>

using namespace std;

int main () {
    std::string str1 = "a/b/c[2]/d";
    std::string str2 = R"regex((.*)a/b/c[2]/d(.*))regex";
    std::regex e(str2);

    std::cout << "str1 = " << str1 << std::endl;
    std::cout << "str2 = " << str2 << std::endl;
    if (regex_match(str1, e)) {
        std::cout << "matched" << std::endl;
    }
}

expected output:

str1 = a/b/c[2]/d
str2 = (.*)a/b/c[2]/d(.*)

answered Apr 13 '16 at 11:23

Richard Hodges

68,278
7
90
142

Thanks for writing this. Initially, I thought it was a good idea to try your code. Unfortunately, it did not work. Still giving me same 'std::regex_error'. – Hemant Bhargava Apr 13 '16 at 11:39
There is nothing to escape in OP pattern. The failure is most likely due to the old GCC version. – Wiktor Stribiżew Apr 13 '16 at 11:43
Thread model: posix gcc version 4.8.2 (Ubuntu 4.8.2-19ubuntu1) – Hemant Bhargava Apr 13 '16 at 11:44
@HemantBhargava eek. gcc 4.8 is known to be buggy. upgrade to 5.3 asap! – Richard Hodges Apr 13 '16 at 12:58

regex_match fails to find square brackets

3 Answers3