到目前为止，所有的答案似乎都有一个或多个这样的问题:(1)它可能无法在vc++上工作(2)它需要额外的依赖，如boost或fmt(3)它太复杂的自定义实现，可能没有经过很好的测试。

下面的代码解决了上述所有问题。

#include <string>
#include <cstdarg>
#include <memory>

std::string stringf(const char* format, ...)
{
    va_list args;
    va_start(args, format);
    #ifndef _MSC_VER

        //GCC generates warning for valid use of snprintf to get
        //size of result string. We suppress warning with below macro.
        #ifdef __GNUC__
        #pragma GCC diagnostic push
        #pragma GCC diagnostic ignored "-Wformat-nonliteral"
        #endif

        size_t size = std::snprintf(nullptr, 0, format, args) + 1; // Extra space for '\0'

        #ifdef __GNUC__
        # pragma GCC diagnostic pop
        #endif

        std::unique_ptr<char[]> buf(new char[ size ] ); 
        std::vsnprintf(buf.get(), size, format, args);
        return std::string(buf.get(), buf.get() + size - 1 ); // We don't want the '\0' inside
    #else
        int size = _vscprintf(format, args);
        std::string result(++size, 0);
        vsnprintf_s((char*)result.data(), size, _TRUNCATE, format, args);
        return result;
    #endif
    va_end(args);
}    

int main() {
    float f = 3.f;
    int i = 5;
    std::string s = "hello!";
    auto rs = stringf("i=%d, f=%f, s=%s", i, f, s.c_str());
    printf("%s", rs.c_str());
    return 0;
}

注:

Separate VC++ code branch is necessary because VC++ has decided to deprecate snprintf which will generate compiler warnings for other highly voted answers above. As I always run in "warnings as errors" mode, its no go for me. The function accepts char * instead of std::string. This because most of the time this function would be called with literal string which is indeed char *, not std::string. In case you do have std::string as format parameter, then just call .c_str(). Name of the function is stringf instead of things like string_format to keepup with printf, scanf etc. It doesn't address safety issue (i.e. bad parameters can potentially cause seg fault instead of exception). If you need this then you are better off with boost or fmt libraries. My preference here would be fmt because it is just one header and source file to drop in the project while having less weird formatting syntax than boost. However both are non-compatible with printf format strings so below is still useful in that case. The stringf code passes through GCC strict mode compilation. This requires extra #pragma macros to suppress false positives in GCC warnings.

以上代码已在，

GCC 4.9.2 11 / c++ / C + + 14 vc++编译器19.0 铿锵声3.7.0

你不能直接这样做，因为你没有对底层缓冲区的写访问权(直到c++ 11;见Dietrich Epp的评论)。你必须先在c-string中执行，然后将其复制到std::string中:

  char buff[100];
  snprintf(buff, sizeof(buff), "%s", "Hello");
  std::string buffAsStdStr = buff;

但我不确定为什么不直接使用字符串流?我想你有特定的理由不这么做:

  std::ostringstream stringStream;
  stringStream << "Hello";
  std::string copyOfStr = stringStream.str();

inline void format(string& a_string, const char* fmt, ...)
{
    va_list vl;
    va_start(vl, fmt);
    int size = _vscprintf( fmt, vl );
    a_string.resize( ++size );
    vsnprintf_s((char*)a_string.data(), size, _TRUNCATE, fmt, vl);
    va_end(vl);
}

这是一个特定于Windows的解决方案，旨在避免Visual Studio中的编译器警告而不消除它们。所讨论的警告是针对使用std::string和va_start，这会错误地产生警告，以及针对使用已弃用的printf变量。

template<typename ... va>
std::string Format( const std::string& format, va ... args )
{
    std::string s;
    s.resize( _scprintf( format.c_str(), args ... ) + 1 );
    s.resize( _snprintf_s( s.data(), s.capacity(), _TRUNCATE, format.c_str(), args ... ) );
    return s;
}

template<typename ... va>
std::wstring Format( const std::wstring& format, va ... args )
{
    std::wstring s;
    s.resize( _scwprintf( format.c_str(), args ... ) + 1 );
    s.resize( _snwprintf_s( s.data(), s.capacity(), _TRUNCATE, format.c_str(), args ... ) );
    return s;
}

std::string s = Format( "%hs %d", "abc", 123 );
std::wstring ws = Format( L"%hs %d", "abc", 123 );

c++ 11内部使用vsnprintf()的解决方案:

#include <stdarg.h>  // For va_start, etc.

std::string string_format(const std::string fmt, ...) {
    int size = ((int)fmt.size()) * 2 + 50;   // Use a rubric appropriate for your code
    std::string str;
    va_list ap;
    while (1) {     // Maximum two passes on a POSIX system...
        str.resize(size);
        va_start(ap, fmt);
        int n = vsnprintf((char *)str.data(), size, fmt.c_str(), ap);
        va_end(ap);
        if (n > -1 && n < size) {  // Everything worked
            str.resize(n);
            return str;
        }
        if (n > -1)  // Needed size returned
            size = n + 1;   // For null char
        else
            size *= 2;      // Guess at a larger size (OS specific)
    }
    return str;
}

一种更安全、更有效的方法(我测试过，它更快):

#include <stdarg.h>  // For va_start, etc.
#include <memory>    // For std::unique_ptr

std::string string_format(const std::string fmt_str, ...) {
    int final_n, n = ((int)fmt_str.size()) * 2; /* Reserve two times as much as the length of the fmt_str */
    std::unique_ptr<char[]> formatted;
    va_list ap;
    while(1) {
        formatted.reset(new char[n]); /* Wrap the plain char array into the unique_ptr */
        strcpy(&formatted[0], fmt_str.c_str());
        va_start(ap, fmt_str);
        final_n = vsnprintf(&formatted[0], n, fmt_str.c_str(), ap);
        va_end(ap);
        if (final_n < 0 || final_n >= n)
            n += abs(final_n - n + 1);
        else
            break;
    }
    return std::string(formatted.get());
}

fmt_str是按值传递的，以符合va_start的要求。

注意:“更安全”和“更快”的版本在某些系统上不起作用。因此，两家公司仍在上市。此外，“更快”完全取决于预分配步骤是否正确，否则strcpy会使其变慢。

我不喜欢把事情搞复杂。这是基于iFreilicht的答案，但我减少了一些噪音，使它更有效。请注意，如果您计划在接口中使用此功能，可能会添加一些模糊输入检查。

#include <iostream>
#include <string>

template<typename... Ts>
std::string string_format( const std::string& format, Ts... Args )
{
    const size_t n = std::snprintf( nullptr, 0, format.c_str(), Args ... ) + 1; // Extra space for '\0'
    std::string ret(n, '\0');
    std::snprintf( &ret.front(), n, format.c_str(), Args... );
    return ret;
}

int main()
{
    int a = 5;
    char c = 'h';
    double k = 10.3;
    std::cout << string_format("%d, %c, %.2f", a, c, k) << "\n";
}

输出:

5, h, 10.30

试着自己

(*唯一的警告，我发现性能方面是没有办法默认初始化字符串存储。这很遗憾，因为我们不需要在这里将所有的值初始化为“\0”。)