C++ regex_search is really weird

March 1, 2023

I’ve written some working code that prints out every occurrence of 2 vowels in a row.

std::regex x("(a|e|i|o|u){2}");

std::smatch r;

std::string t = some_string;

while (std::regex_search(t, r, x)) {
    std::cout << "match: " << r.str() << '\n';

    t = r.suffix();
}

But when I change the order like this:

while (std::regex_search(t, r, x)) {
    t = r.suffix();

    std::cout << "match: " << r.str() << '\n';
}

it suddenly starts giving random results. I don’t see the connection between these 2 lines, and why changing their order would affect anything. Can anyone explain this?

>Solution :

This statement:

    t = r.suffix();

isn’t creating a new string object, or changing what string object t denotes, or anything like that; rather, it’s mutating the existing string object by copying over the contents of r.suffix().

And r doesn’t hold the actual string data; it just holds index information so that the appropriate string data can be extracted from t. Mutating t essentially invalidates the old indices, so you get the wrong string data.