C++11 regex to match nothing

Question

I encountered a task which reuires a regex matching nothing.

C++ reference happily states that it already has such thing: http://en.cppreference.com/w/cpp/regex/basic_regex/basic_regex

1) Default constructor. Constructs an empty regular expression which will match nothing.

But in reality (clang 3.3) it happens not to be the case:

#include <string>
#include <regex>

int main(int argc, const char *argv[]) {

    std::regex re1;
    std::regex re2("");
    std::smatch rt1, rt2;

    bool r1 = std::regex_match(std::string(""), rt1, re1);
    bool r2 = std::regex_match(std::string(""), rt2, re2);
    std::cerr << "r1:" << r1 << ", r2:" << r2 << std::endl;
}

This program prints: r1:1, r2:1 What should mean that both regexes matched empty string.

Any idea what is wrong here and how to create "match nothing" regex ?

There's an empty string at the beginning of each string, that's what it's matching. — Barmar, Nov 30 '13 at 08:20
@Barmar there's also the argument that an empty match pattern represents the universally accepting DFA `({q}, S, {q*S->q}, q, {q}})`, for some abstract charset `S` and (q*S->q meaning forall s in S, qxs->q) — moshbear, Nov 30 '13 at 08:29
@Dfr: so you want a regex which never reports any match, right? As opposed to one which matches the absence of content? — John Zwinck, Nov 30 '13 at 08:55
It would be more accurate to say that the default regex matches only an empty string. — Alan Stokes, Nov 30 '13 at 09:46
@John, yes, also want to clearly understand what behaviour to expect from `re1` (regex constructed with empty constructor) and how it differs from `re2` (regex constructed with empty string) — Dfr, Nov 30 '13 at 10:44
`clang` gives me `false`. Are you linking against GNU `libstdc++` by any chance? — Cubbi, Nov 30 '13 at 15:42
@Cubbi, i guess so, i'm using clang in Fedora which itself depends on `libstdc++` — Dfr, Dec 02 '13 at 06:35
@Dfr in that case, you should use boost.regex. The GNU libstdc++ regex library was just non-functional stubs until the current 4.9 — Cubbi, Dec 02 '13 at 12:09

score 3 · Accepted Answer · answered Nov 30 '13 at 11:42

3

The default constructor for std::basic_regex constructs a regular expression that "does not match any character sequence". [re.regex.construct]/1. If your implementation matches an empty character sequence it's wrong.

answered Nov 30 '13 at 11:42

Pete Becker

74,985
8
76
165

C++11 regex to match nothing

1 Answers1