在可能的情况下，如何缩小浮动的字符串表示？

Question

我研究

std::format

试图使浮点变量的简短表示。 我的问题是我无法设置宽度，因为我不知道小数点前有多少位数字。如果我设置了精度，即使它们是唯一的尾巴数字，它也会打印零。 我猜想这可以通过宽度/精度/填充等以某种方式实现，但看不到道路。 llet说，我希望字符串适合指定的

std::string

如果可能的话；如果不是，字符串应采用最小的数字，则优先使用“ 100”而不是“ 1E2”或“ 1000”或1E03“或“ 0.01”至“ 1E-2”，等等。

可以通过std ::格式（或任何其他类型的std格式）来实现这一点，而无需手动转换float的每个数字？

fit_to

demo

.

以下版本的

#include <iostream>

std::string to_string_shortest(float v, [[maybe_unused]] std::size_t fit_to = 4) {
    return std::format("{}", v);
}

int main()
{
    std::cout << to_string_shortest(3.1415926f) << '\n';        // Want to have 3.14
    std::cout << to_string_shortest(1e7f) << '\n';              // Want to have 1e+7 (1e7 acceptable)
    std::cout << to_string_shortest(1e-7f) << '\n';             // Want to have 1e-7
    std::cout << to_string_shortest(1.2345e20f) << '\n';        // Want to have 1e20
    std::cout << to_string_shortest(1.2345e-20f) << '\n';       // Want to have 1e-20

    std::cout << to_string_shortest(3.1415926f, 8) << '\n';     // Want to have 3.141593
    std::cout << to_string_shortest(-3.1415926f, 8) << '\n';    // Want to have -3.14159
    std::cout << to_string_shortest(1.2345e20f, 8) << '\n';     // Want to have 1.234e20
    std::cout << to_string_shortest(1.2345e-20f, 8) << '\n';    // Want to have 1.23e-20

    std::cout << to_string_shortest(0.001f) << '\n';            // Want to keep as is
    std::cout << to_string_shortest(1.0f) << '\n';              // Want to keep as is
    std::cout << to_string_shortest(1e20f) << '\n';             // Want to keep as is
    std::cout << to_string_shortest(1e-20f) << '\n';            // Want to keep as is
}

可能不是完美的，但它很近。主要是，在某些指数值上有落后空间。输出周围有

to_string_shortest

显示整个场宽度。在Compilerexplorer.

Answer 1

[]

到

-e0

和修剪的尾随空间的更改指数。

-e

输出为：

auto to_string_shortest(float num, std::size_t fit_to = 4) -> std::string {

   // get digits in num
   int width = (num == 0) ? 1 : (std::log10(std::abs(num)) + 1);

   // limit width - keep > 0
   width = std::min(std::abs(width), 10);

   // '<' left justifies the output


   std::string fmt = std::format("{:<{}.{}}", num, width, fit_to);
   // replace -e0 with just -e
   auto exponent_position{fmt.rfind("e-0")};
   if (exponent_position != std::string::npos) {
      fmt.replace(exponent_position, 3, "-e");
   }

   // trim trailing white space
   auto pos = fmt.find_last_of(' ');
   while (pos != std::string::npos) {
      fmt.erase(pos);
      pos = fmt.find_last_of(' ');
   }

   return "[" + fmt + "]";  // "[ ]]" just to show no spaces
}
    

int main() {
   std::cout << to_string_shortest(3.1415926f) << '\n';     // Want to have 3.14
   std::cout << to_string_shortest(1e7f) << '\n';           // Want to have 1e+7 (1e7 acceptable)
   std::cout << to_string_shortest(3.1415926f, 8) << '\n';  // Want to have 3.141593
   std::cout << to_string_shortest(-3.1415926f, 8) << '\n'; // Want to have -3.14159
   std::cout << to_string_shortest(0.001f) << '\n';         // Want to keep as is
   std::cout << to_string_shortest(1.0f) << '\n';           // Want to keep as is

   std::cout << to_string_shortest(1e-7f) << '\n';          // Want to have 1e-7
   std::cout << to_string_shortest(1.2345e20f) << '\n';     // Want to have 1e20
   std::cout << to_string_shortest(1.2345e-20f) << '\n';    // Want to have 1e-20
   std::cout << to_string_shortest(1.2345e20f, 8) << '\n';  // Want to have 1.234e20
   std::cout << to_string_shortest(1.2345e-20f, 8) << '\n'; // Want to have 1.23e-20
   std::cout << to_string_shortest(1e20f) << '\n';          // Want to keep as is
   std::cout << to_string_shortest(1e-20f) << '\n';         // Want to keep as is
}