Why do a C++ regular expression code that works with “cmatch” raises an exception with “smatch”?

问题

I am new to C++ regular expressions and cannot make them work with strings instead of char*. The examples I have seen so far were always for c strings.

My real program, which I will not even try to show here, uses sub matches but I could not make them work, so I tried to modify a very simple working example but it does not work either. I use Visual Studio 2010 Ultimate.

The original - working - code:

const char *first = "abcd"; 
const char *last = first + strlen(first); 
std::cmatch mr; 
std::regex rx("abc"); 
std::regex_constants::match_flag_type fl = std::regex_constants::match_default;

std::cout << "search(f, l, \"abc\") == " << std::boolalpha 
          << regex_search(first, last, mr, rx) << std::endl; 
std::cout << "  matched: \"" << mr.str() << "\"" << std::endl; 

std::cout << "search(\"xabcd\", \"abc\") == " << std::boolalpha
          << regex_search("xabcd", mr, rx) << std::endl; 
std::cout << "  matched: \"" << mr.str() << "\"" << std::endl;

The modified code:

const string first = "abcd";     // char * => string
std::smatch mr;                  // cmatch => smatch
std::regex rx(string("abc")); 
std::regex_constants::match_flag_type fl = std::regex_constants::match_default;

               // this works:
std::cout << "search(f, l, \"abc\") == " << std::boolalpha 
          << regex_search(first, mr, rx) << std::endl; 
std::cout << "  matched: \"" << mr.str() << "\"" << std::endl; 

               // after the next line executes mr seems good to me:
               // mr[0] = {3, matched:true, first="abcd", second="d",...}
std::cout << "search(\"xabcd\", \"abc\") == " << std::boolalpha
          << regex_search(string("xabcd"), mr, rx) << std::endl; 
               // but the following line gives the error
               // "Debug assertion failed"
               // Expression: string iterators incompatible
std::cout << "  matched: \"" << mr.str() << "\"" << std::endl;

Strange that one part of the modified code works, while the next part causes an exception. I even tried to use mr[0].str() but I got the same error message. Could you help me to solve this problem?

回答1:

The issue is one of temporaries.

smatch will contain iterators into the string you are searching for.

regex_search(string("xabcd"), mr, rx) creates a temporary string that dies at the ;.

Therefore by the time you use mr at the next line, it refers to invalidated memory. The string should live longer than mr.

来源：https://stackoverflow.com/questions/10552884/why-do-a-c-regular-expression-code-that-works-with-cmatch-raises-an-exceptio