mirror of
https://github.com/saitohirga/WSJT-X.git
synced 2024-11-18 01:52:05 -05:00
267 lines
10 KiB
C++
267 lines
10 KiB
C++
// Copyright 2006 Johan Rade
|
|
// Copyright 2011 Paul A. Bristow To incorporate into Boost.Math
|
|
// Copyright 2012 Paul A. Bristow with new tests.
|
|
|
|
// Distributed under the Boost Software License, Version 1.0.
|
|
// (See accompanying file LICENSE_1_0.txt
|
|
// or copy at http://www.boost.org/LICENSE_1_0.txt)
|
|
|
|
#ifdef _MSC_VER
|
|
# pragma warning(disable : 4127) // Expression is constant.
|
|
#endif
|
|
|
|
#define BOOST_TEST_MAIN
|
|
|
|
#include <boost/test/auto_unit_test.hpp>
|
|
#include <boost/math/special_functions/nonfinite_num_facets.hpp>
|
|
#include "s_.ipp" // To create test strings like std::basic_string<CharType> s = S_("0 -0");
|
|
|
|
#include <iomanip>
|
|
#include <locale>
|
|
#include <sstream>
|
|
#include <ostream>
|
|
#include <iostream>
|
|
#include <iomanip>
|
|
#include <limits>
|
|
#include <iostream>
|
|
|
|
namespace {
|
|
|
|
// Using an anonymous namespace resolves ambiguities on platforms
|
|
// with fpclassify etc functions at global scope.
|
|
|
|
using namespace boost::math;
|
|
using boost::math::signbit;
|
|
using boost::math::changesign;
|
|
using boost::math::isnan;
|
|
|
|
//------------------------------------------------------------------------------
|
|
|
|
template<class CharType, class ValType> void signed_zero_test_impl();
|
|
// Loopback tests using all built-in char and floating-point types.
|
|
|
|
BOOST_AUTO_TEST_CASE(signed_zero_test)
|
|
{
|
|
std::cout
|
|
<< "BuildInfo:" << '\n'
|
|
<< " platform " << BOOST_PLATFORM << '\n'
|
|
<< " compiler " << BOOST_COMPILER << '\n'
|
|
<< " STL " << BOOST_STDLIB << '\n'
|
|
<< " Boost version " << BOOST_VERSION/100000 << "."
|
|
<< BOOST_VERSION/100 % 1000 << "."
|
|
<< BOOST_VERSION % 100
|
|
<< std::endl;
|
|
|
|
signed_zero_test_impl<char, float>();
|
|
signed_zero_test_impl<char, double>();
|
|
signed_zero_test_impl<char, long double>();
|
|
signed_zero_test_impl<wchar_t, float>();
|
|
signed_zero_test_impl<wchar_t, double>();
|
|
signed_zero_test_impl<wchar_t, long double>();
|
|
}
|
|
|
|
template<class CharType, class ValType> void signed_zero_test_impl()
|
|
{
|
|
|
|
|
|
if (signbit(static_cast<CharType>(-1e-6f) / (std::numeric_limits<CharType>::max)()) != -0)
|
|
{
|
|
BOOST_TEST_MESSAGE("Signed zero is not supported on this platform!");
|
|
return;
|
|
}
|
|
|
|
std::locale old_locale;
|
|
std::locale tmp_locale(
|
|
old_locale, new nonfinite_num_put<CharType>(signed_zero));
|
|
std::locale new_locale(tmp_locale, new nonfinite_num_get<CharType>);
|
|
|
|
std::basic_stringstream<CharType> ss;
|
|
ss.imbue(new_locale);
|
|
|
|
std::basic_string<CharType> null = S_("");
|
|
|
|
std::basic_string<CharType> s1 = S_("123");
|
|
ss << s1 << std::endl;
|
|
ss.str(null);
|
|
|
|
|
|
BOOST_CHECK(ss.str() == null); //
|
|
|
|
ValType a1 = static_cast<ValType>(0); // zero.
|
|
ValType a2 = (changesign)(static_cast<ValType>(0)); // negative signed zero.
|
|
BOOST_CHECK(!(signbit)(a1)); //
|
|
BOOST_CHECK((signbit)(a2));
|
|
|
|
ss << a1 << ' ' << a2;
|
|
|
|
std::basic_string<CharType> s = S_("0 -0"); // Expected.
|
|
BOOST_CHECK(ss.str() == s);
|
|
|
|
ValType b1, b2;
|
|
ss >> b1 >> b2; // Read back in.
|
|
|
|
BOOST_CHECK(b1 == a1);
|
|
BOOST_CHECK(b2 == a2);
|
|
BOOST_CHECK(!(signbit)(b1));
|
|
BOOST_CHECK((signbit)(b2));
|
|
BOOST_CHECK(ss.rdstate() == std::ios_base::eofbit);
|
|
} // template<class CharType, class ValType> void signed_zero_test_impl()
|
|
|
|
// Checking output of types char using first default & then using signed_zero flag.
|
|
#define CHECKOUT(manips, expected)\
|
|
{\
|
|
{\
|
|
std::locale old_locale;\
|
|
std::locale tmp_locale(old_locale, new nonfinite_num_put<char>(0));\
|
|
std::locale new_locale(tmp_locale, new nonfinite_num_get<char>);\
|
|
std::ostringstream ss;\
|
|
ss.imbue(new_locale);\
|
|
ss << manips;\
|
|
std::basic_string<char> s = S_(expected);\
|
|
BOOST_CHECK_EQUAL(ss.str(), s);\
|
|
}\
|
|
{\
|
|
std::locale old_locale;\
|
|
std::locale tmp_locale(old_locale, new nonfinite_num_put<char>(signed_zero));\
|
|
std::locale new_locale(tmp_locale, new nonfinite_num_get<char>);\
|
|
std::ostringstream ss;\
|
|
ss.imbue(new_locale);\
|
|
ss << manips;\
|
|
std::basic_string<char> s = S_(expected);\
|
|
BOOST_CHECK_EQUAL(ss.str(), s);\
|
|
}\
|
|
}\
|
|
|
|
BOOST_AUTO_TEST_CASE(misc_output_tests)
|
|
{ // Tests of output using a variety of output options.
|
|
|
|
//
|
|
// STD libraries don't all format zeros the same,
|
|
// so figure out what the library-specific formatting is
|
|
// and then make sure that our facet produces the same...
|
|
//
|
|
bool precision_after = false; // Prints N digits after the point rather than N digits total
|
|
bool triple_exponent = false; // Has 3 digits in the exponent rather than 2.
|
|
std::stringstream ss;
|
|
ss << std::showpoint << std::setprecision(6) << 0.0;
|
|
if(ss.str().size() == 8)
|
|
precision_after = true;
|
|
ss.str("");
|
|
ss << std::scientific << 0.0;
|
|
triple_exponent = ss.str().size() - ss.str().find_first_of('e') == 5;
|
|
|
|
|
|
|
|
// Positive zero.
|
|
CHECKOUT(0, "0"); // integer zero.
|
|
CHECKOUT(0., "0"); // double zero.
|
|
CHECKOUT(std::setw(2) << 0., " 0");
|
|
CHECKOUT(std::setw(4) << 0., " 0");
|
|
CHECKOUT(std::right << std::setw(4) << 0., " 0");
|
|
CHECKOUT(std::left << std::setw(4) << 0., "0 ");
|
|
CHECKOUT(std::setw(4) << std::setfill('*') << 0., "***0");
|
|
CHECKOUT(std::setw(4) << std::internal << std::setfill('*') << 0., "***0"); // left adjust sign and right adjust value.
|
|
CHECKOUT(std::showpos << std::setw(4) << std::internal << std::setfill('*') << 0., "+**0"); // left adjust sign and right adjust value.
|
|
|
|
if(precision_after)
|
|
{
|
|
// BOOST_STDLIB == ("Dinkumware standard library version" BOOST_STRINGIZE(_CPPLIB_VER)) )
|
|
CHECKOUT(std::showpoint << 0., "0.000000"); // std::setprecision(6)
|
|
CHECKOUT(std::setprecision(2) << std::showpoint << 0., "0.00");
|
|
}
|
|
else
|
|
{
|
|
CHECKOUT(std::showpoint << 0., "0.00000"); // std::setprecision(6)
|
|
CHECKOUT(std::setprecision(2) << std::showpoint << 0., "0.0");
|
|
}
|
|
CHECKOUT(std::fixed << std::setw(5) << std::setfill('0') << std::setprecision(2) << 0., "00.00");
|
|
CHECKOUT(std::fixed << std::setw(6) << std::setfill('0') << std::setprecision(2) << 0., "000.00");
|
|
CHECKOUT(std::fixed << std::setw(6) << std::setfill('0') << std::setprecision(3) << 0., "00.000");
|
|
CHECKOUT(std::fixed << std::setw(6) << std::setfill('*') << std::setprecision(3) << 0., "*0.000");
|
|
CHECKOUT(std::fixed << std::setw(6) << std::setfill('*') << std::setprecision(2) << std::left << 0.0, "0.00**");
|
|
|
|
CHECKOUT(std::showpos << 0., "+0");
|
|
CHECKOUT(std::showpos << std::fixed << std::setw(6) << std::setfill('*') << std::setprecision(2) << std::left << 0.0, "+0.00*");
|
|
if(triple_exponent)
|
|
{
|
|
CHECKOUT(std::scientific << std::showpoint << std::setw(10) << std::setfill('*') << std::setprecision(1) << std::left << 0., "0.0e+000**");
|
|
}
|
|
else
|
|
{
|
|
CHECKOUT(std::scientific << std::showpoint << std::setw(10) << std::setfill('*') << std::setprecision(1) << std::left << 0., "0.0e+00***");
|
|
}
|
|
CHECKOUT(std::fixed << std::showpoint << std::setw(6) << std::setfill('*') << std::setprecision(3) << std::left << 0., "0.000*");
|
|
|
|
double nz = (changesign)(static_cast<double>(0)); // negative signed zero.
|
|
CHECKOUT(nz, "-0");
|
|
// CHECKOUT(std::defaultfloat << nz, "-0"); Only for C++11
|
|
CHECKOUT(std::showpos << nz, "-0"); // Ignore showpos because is negative.
|
|
CHECKOUT(std::setw(2) << nz, "-0");
|
|
CHECKOUT(std::setw(4) << nz, " -0");
|
|
CHECKOUT(std::right << std::setw(4) << nz, " -0");
|
|
CHECKOUT(std::left << std::setw(4) << nz, "-0 ");
|
|
CHECKOUT(std::setw(4) << std::setfill('*') << nz, "**-0");
|
|
CHECKOUT(std::setw(4) << std::internal << std::setfill('*') << nz, "-**0"); // Use std::internal to left adjust sign and right adjust value.
|
|
CHECKOUT(std::showpos << std::setw(4) << std::internal << std::setfill('*') << nz, "-**0");
|
|
|
|
CHECKOUT(std::fixed << std::setw(5) << std::setfill('0') << std::setprecision(2) << 0., "00.00");
|
|
CHECKOUT(std::fixed << std::setw(6) << std::setfill('0') << std::setprecision(2) << 0., "000.00");
|
|
CHECKOUT(std::fixed << std::setw(6) << std::setfill('0') << std::setprecision(3) << 0., "00.000");
|
|
CHECKOUT(std::fixed << std::setw(6) << std::setfill('*') << std::setprecision(3) << 0., "*0.000");
|
|
CHECKOUT(std::fixed << std::setw(6) << std::setfill('*') << std::setprecision(2) << std::left << 0.0, "0.00**");
|
|
CHECKOUT(std::setprecision(2) << nz, "-0"); // No showpoint, so no decimal point nor trailing zeros.
|
|
if(precision_after)
|
|
{
|
|
CHECKOUT(std::setprecision(2) << std::showpoint << nz, "-0.00"); // or "-0.0"
|
|
CHECKOUT(std::setw(1) << std::setprecision(3) << std::showpoint << nz, "-0.000"); // Not enough width for precision overflows width. or "-0.00"
|
|
}
|
|
else
|
|
{
|
|
CHECKOUT(std::setprecision(2) << std::showpoint << nz, "-0.0"); // or "-0.00"
|
|
CHECKOUT(std::setw(1) << std::setprecision(3) << std::showpoint << nz, "-0.00"); // Not enough width for precision overflows width. or "-0.000"
|
|
}
|
|
if(triple_exponent)
|
|
{
|
|
CHECKOUT(std::scientific << std::showpoint << std::setw(10) << std::setfill('*') << std::setprecision(1) << std::left << nz, "-0.0e+000*"); // -0.0e+00**
|
|
}
|
|
else
|
|
{
|
|
CHECKOUT(std::scientific << std::showpoint << std::setw(10) << std::setfill('*') << std::setprecision(1) << std::left << nz, "-0.0e+00**"); // -0.0e+000*
|
|
}
|
|
CHECKOUT(std::fixed << std::showpoint << std::setw(6) << std::setfill('*') << std::setprecision(3) << std::left << 0., "0.000*");
|
|
|
|
// Non zero values.
|
|
|
|
CHECKOUT(std::showpos << std::fixed << std::setw(6) << std::setfill('*') << std::setprecision(2) << std::left << 42., "+42.00");
|
|
CHECKOUT(std::showpos << std::fixed << std::setw(6) << std::setfill('*') << std::setprecision(2) << std::left << 4.2, "+4.20*");
|
|
CHECKOUT(std::showpos << std::fixed << std::setw(6) << std::setfill('*') << std::setprecision(2) << std::left << 1.22, "+1.22*");
|
|
CHECKOUT(std::showpos << std::fixed << std::setw(6) << std::setfill('*') << std::setprecision(2) << std::left << 0.12, "+0.12*");
|
|
|
|
CHECKOUT(std::setprecision(4) << std::showpoint << 1.2, "1.200");
|
|
|
|
}
|
|
|
|
} // anonymous namespace
|
|
|
|
/*
|
|
|
|
Output:
|
|
|
|
test_signed_zero.cpp
|
|
Running 2 test cases...
|
|
Platform: Win32
|
|
Compiler: Microsoft Visual C++ version 10.0
|
|
STL : Dinkumware standard library version 520
|
|
Boost : 1.49.0
|
|
Entering test suite "Master Test Suite"
|
|
Entering test case "signed_zero_test"
|
|
Leaving test case "signed_zero_test"; testing time: 2ms
|
|
Entering test case "misc_output_tests"
|
|
Leaving test case "misc_output_tests"; testing time: 15ms
|
|
Leaving test suite "Master Test Suite"
|
|
|
|
*** No errors detected
|
|
|
|
*/
|
|
|