Add to_string, prints a DDouble in the %e style.

include/xprec/ddouble.h

@@ -8,6 +8,7 @@
 #include <cstdint>
 #include <iosfwd>
 #include <limits>
+#include <string>
 
 #include "version.h"
 
@@ -297,6 +298,7 @@ bool isinf(DDouble x);
 bool isnan(DDouble x);
 bool isnormal(DDouble x);
 bool iszero(DDouble x);
+std::string to_string(DDouble d, size_t nDigits = 34);
 
 /**
  * Gauss-Chebyshev quadrature rule.

src/io.cxx

@@ -7,6 +7,7 @@
 #include <array>
 #include <iostream>
 #include <sstream>
+#include <cassert>
 
 #ifndef XPREC_API_EXPORT
 #define XPREC_API_EXPORT
@@ -58,4 +59,65 @@ std::ostream &operator<<(std::ostream &out, DDouble x)
     return out;
 }
 
+XPREC_API_EXPORT
+std::string to_string(DDouble d, size_t nDigits){
+	using namespace std;
+
+	//Get some basic cases out of the way.
+	if(d.hi() == 0.0)
+		return "0.0";
+
+	if(isnan(d.hi()))
+		return "NaN";
+
+	if(isinf(d.hi()))
+		if(d.hi() > 0.0)
+			return "Inf";
+		else
+			return "-Inf";
+
+
+	//Alright. Get a string started. Handle sign.
+	string s;
+
+	if(d.hi() < 0.0) s += "-";
+	d = abs(d);
+
+
+	//Get the initial tens place and reduce d by it. This way, d becomes digit.nextdigits.
+	auto n = (int)floor(log10(d.hi()));
+	if(n < 0) n++;
+
+	constexpr DDouble Ten = 10.0;
+	d /= pow(Ten, n);
+
+	//Loop over digits, print them.
+	nDigits = min(max(nDigits, (size_t)3), (size_t)34);
+
+	for(size_t i = 0; i != nDigits; i++){
+		if(i == 1) s += ".";
+
+		//Current digit to print is just the leading one.
+		auto m = (int)floor(d.hi());
+		assert(m >= 0 && m < 10);
+
+		s += std::to_string(m);
+
+		//Remove leading digit, promote by ten to retain the digit.nextdigits form.
+		d = (d - DDouble(m))*Ten;
+
+		//Emptied it - nothing more to print.
+		if(d.hi() == 0.0)
+			break;
+
+		assert(d.hi() > 0.0);
+	}
+
+	//Assuming you care more about the mantissa... always use exponential form. n == 0 imples e0.
+	if(n != 0)
+		s += "e" + to_string(n);
+
+	return s;
+}
+
 } /* namespace xprec */