Here is the MCVE:
#include
#include
std::string s()
{
return \"test\";
}
int main()
{
static const std::regex regex(
There was a change going from C++11 to C++14 where std::regex_search is no longer allowed to take a r-value
template< class STraits, class SAlloc, class Alloc, class CharT, class Traits > bool regex_search( const std::basic_string<CharT,STraits,SAlloc>&&, std::match_results< typename std::basic_string<CharT,STraits,SAlloc>::const_iterator, Alloc>&, const std::basic_regex<CharT, Traits>&, std::regex_constants::match_flag_type flags = std::regex_constants::match_default ) = delete;
This was added as the overload that takes a const std::string&
is prohibited from accepting temporary strings, otherwise this function populates match_results m with string iterators that become invalid immediately.
So you can no longer pass a temporary to std::regex_search
as of C++14
To fix your code we would simply store the return from s()
into a variable in main and use that to call std::regex_search
.
#include <iostream>
#include <regex>
std::string s()
{
return "test";
}
int main()
{
static const std::regex regex(R"(\w)");
std::smatch smatch;
auto search = s();
if (std::regex_search(search, smatch, regex)) {
std::cout << smatch[0] << std::endl;
}
return 0;
}
Live Example
This changed between C++11 and C++14. If we go to the cppreference section for std::regex_search we can see that overload that takes an rvalue reference was deleted since C++14:
template< class STraits, class SAlloc,
class Alloc, class CharT, class Traits > bool regex_search( const std::basic_string<CharT,STraits,SAlloc>&&,
std::match_results<
typename std::basic_string<CharT,STraits,SAlloc>::const_iterator,
Alloc
>&,
const std::basic_regex<CharT, Traits>&,
std::regex_constants::match_flag_type flags =
std::regex_constants::match_default ) = delete;
It was changed due to LWG issue 2329: regex_match()/regex_search() with match_results should forbid temporary strings which says (emphasis mine):
Consider the following code:
const regex r(R"(meow(\d+)\.txt)"); smatch m; if (regex_match(dir_iter->path().filename().string(), m, r)) { DoSomethingWith(m[1]); }
This occasionally crashes. The problem is that dir_iter->path().filename().string() returns a temporary string, so the match_results contains invalidated iterators into a destroyed temporary string.
It's fine for regex_match/regex_search(str, reg) to accept temporary strings, because they just return bool. However, the overloads taking match_results should forbid temporary strings.
and indeed if we use a non-temporary:
std::string s1 = s() ;
if (std::regex_search(s1, smatch, regex)) {
//...
}
it compiles (see it live) and no longer exhibits undefined behavior.
Interesting to note that gcc/libstdc++ has this overload deleted in C++11 mode as well see it live. Since this is undefined behavior it seems like a good solution.
This issue also pops up in other areas of the library see Visual Studio regex_iterator Bug? which deals with the same issue but with regex_iterator/regex_token_iterator
.
This not a bug, but expected behaviour.
The reason is that s()
returns a temporary string, regex_search
makes use of regex_match
and consequently if a temporary string was utilized match results would contain iterators to a string that no longer exists. This would have been undefined behaviour. Thus, the committee abolished this regex_search
overload in C++14.
You can also confirm in the standard 28.4 Header synopsis [re.syn]:
template <class ST, class SA, class Allocator, class charT, class traits>
bool regex_search(const basic_string<charT, ST, SA>&&,
match_results<
typename basic_string<charT, ST, SA>::const_iterator,
Allocator>&,
const basic_regex<charT, traits>&,
regex_constants::match_flag_type =
regex_constants::match_default) = delete;
As you can see the overload that takes a rvalue to a basic_string
is marked deleted.