MetaDataRef: Make set_float preserve numbers exactly (#16090)

src/util/string.cpp

@@ -1065,3 +1065,30 @@ std::optional<v3f> str_to_v3f(std::string_view str)
 
 	return value;
 }
+
+std::string my_double_to_string(double number)
+{
+	if (std::isfinite(number)) {
+		char buf[64];
+		snprintf(buf, sizeof(buf), "%.17g", number);
+		return buf;
+	}
+	if (number < 0)
+		return "-inf";
+	if (number > 0)
+		return "inf";
+	return "nan";
+}
+
+std::optional<double> my_string_to_double(const std::string &s)
+{
+	if (s.empty())
+		return std::nullopt;
+	char *end = nullptr;
+	errno = 0;
+	// Note: this also supports hexadecimal notation like "0x1.0p+1"
+	double number = std::strtod(s.data(), &end);
+	if (end != &*s.end())
+		return std::nullopt;
+	return number;
+}

src/util/string.h

@@ -465,6 +465,10 @@ inline std::string ftos(float f)
 	return oss.str();
 }
 
+/// @brief Converts double to string. Handles high precision and inf/nan.
+std::string my_double_to_string(double number);
+/// @brief Converts string to double. Handles high precision and inf/nan.
+std::optional<double> my_string_to_double(const std::string &s);
 
 /**
  * Replace all occurrences of \p pattern in \p str with \p replacement.