path: root/src/test-unicode.cpp

#define BOOST_TEST_MODULE unicode_test

#include <boost/test/included/unit_test.hpp>
#include <boost/test/data/dataset.hpp>
#include <boost/test/data/monomorphic.hpp>
#include <boost/test/data/test_case.hpp>

#include <chrono>
#include <exception>
#include <limits>
#include <random>
#include <string>
#include <tuple>
#include <type_traits>
#include <vector>

#include <unicode.h>

using namespace std::chrono_literals;

typedef std::tuple<std::basic_string<utf8_t>, std::basic_string<char16_t>, std::basic_string<char32_t>> types_collection_type;

// create tuple of the same string, in UTF-8, UTF-16 and UTF-32
#define SUCCESS_TUPLE(x) {u8 ## x, u ## x, U ## x}

// Success cases: convert string to all other types, respectively
std::vector<types_collection_type> success_sets {
 SUCCESS_TUPLE(""),
 SUCCESS_TUPLE("ASCII string1"),
 SUCCESS_TUPLE("Täst just looks like German"),
 SUCCESS_TUPLE("\u732b is chinese for cat"),
 SUCCESS_TUPLE("\U0001F63A"),
 SUCCESS_TUPLE("\U0001F63A is a smiling cat"),
};

// Error cases: throwing upon convert to all other types
std::vector<std::basic_string<utf8_t>> failure_strings_char8_t {
 u8"\x80", // utf-8 continuation byte
 u8"\x81", // utf-8 continuation byte
 u8"\xc3\xc3\xa4", // initial byte of utf-8 "ä", followed by valid utf-8 "ä"
 u8"\xF8\x80\x80\x80\x80", // overlong encoding
 u8"\xF7\xBF\xBF\xBF", // valid encoding of invalid code point
};

std::vector<std::basic_string<char16_t>> failure_strings_char16_t {
 u"\xD801", // single high surrogate
 u"\xDFFF", // single low surrogate
 u"\xDFFF\xD801", // bad surrogate pair order
};

std::vector<std::basic_string<char32_t>> failure_strings_char32_t {
 U"blabla \xD801", // invalid unicode (surrogate half)
 U"\x10000000", // invalid unicode (number too big)
};

// output operators must be in same namespace as the type itself
namespace std {

#ifdef __cpp_char8_t
std::ostream& operator<<(std::ostream& os, std::basic_string<utf8_t> const& s)
{
 os << "[";
 for (auto& c: s)
  os << " " << std::to_string(static_cast<uint8_t>(c));
 os << "]";

 return os;
}
#endif

std::ostream& operator<<(std::ostream& os, std::basic_string<char16_t> const& s)
{
 os << "[";
 for (auto& c: s)
  os << " " << std::to_string(static_cast<uint16_t>(c));
 os << "]";

 return os;
}

std::ostream& operator<<(std::ostream& os, std::basic_string<char32_t> const& s)
{
 os << "[";
 for (auto& c: s)
  os << " " << std::to_string(static_cast<uint32_t>(c));
 os << "]";

 return os;
}

}

template<size_t i = 0, size_t j = 0, typename... Ts>
void test_utf_to_utf(std::tuple<Ts...>& t)
{
 typedef typename std::tuple_element<i,typename std::remove_reference<decltype(t)>::type>::type From;
 typedef typename std::tuple_element<j,typename std::remove_reference<decltype(t)>::type>::type To;

 // test
 To result { unicode::utf_to_utf<typename From::value_type, typename To::value_type>(std::get<i>(t)) };

 BOOST_CHECK_MESSAGE(std::get<j>(t) == result, "From " << typeid(From).name() << "(" << i << ", " << std::get<i>(t) << ") to " << typeid(To).name() << "(" << j << ", " << std::get<j>(t) << "), got " << result);

 //std::cout << std::to_string(std::tuple_size<typename std::remove_reference<decltype(t)>::type>::value) << "," << std::to_string(i) << "," << std::to_string(j) << std::endl;

 // iterate over other combinations
 if constexpr (i + 1 < std::tuple_size<typename std::remove_reference<decltype(t)>::type>::value)
  test_utf_to_utf<i + 1, j>(t);
 else if constexpr (j + 1 < std::tuple_size<typename std::remove_reference<decltype(t)>::type>::value)
  test_utf_to_utf<0, j + 1>(t);
}

// We don't use BOOST_DATA_TEST_CASE here because boost::test tries to assign
// a new variable to each tuple element which we don't want
// https://lists.boost.org/boost-bugs/2016/05/45214.php

BOOST_AUTO_TEST_CASE(utf_to_utf_success)
{
 for (auto& t: success_sets)
  test_utf_to_utf(t);
}

template<size_t i = 0, typename... Ts>
void test_is_valid_utf(std::tuple<Ts...>& t)
{
 typedef typename std::tuple_element<i,typename std::remove_reference<decltype(t)>::type>::type T;

 // test
 bool result { unicode::is_valid_utf<typename T::value_type>(std::get<i>(t)) };

 BOOST_CHECK_MESSAGE(result == true, "is_valid_utf w/ " << typeid(T).name() << "(" << i << ", " << std::get<i>(t) << "), got " << result);

 // iterate over other combinations
 if constexpr (i + 1 < std::tuple_size<typename std::remove_reference<decltype(t)>::type>::value)
  test_is_valid_utf<i + 1>(t);
}

BOOST_AUTO_TEST_CASE(is_valid_utf_success)
{
 for (auto& t: success_sets)
  test_is_valid_utf(t);
}

// iterate over std::tuple T types
template<typename From, typename Collection, size_t index = 0>
void test_utf_to_utf_failure(std::basic_string<From>& s)
{
 typedef typename std::tuple_element<index, Collection>::type::value_type To;

 try {
  unicode::utf_to_utf<From,To>(s);
  BOOST_ERROR("Expected exception at index: " << index << ", " << typeid(From).name() << " -> " << typeid(To).name());
 } catch (...) {
  // OK
 };

 // iterate over remaining types 
 if constexpr (index + 1 < std::tuple_size<Collection>::value)
  test_utf_to_utf_failure<From, Collection, index + 1>(s);
}

BOOST_AUTO_TEST_CASE(utf_to_utf_failure)
{
 for (auto& s: failure_strings_char8_t)
  test_utf_to_utf_failure<typename std::remove_reference<decltype(s)>::type::value_type, types_collection_type>(s);
 
 for (auto& s: failure_strings_char16_t)
  test_utf_to_utf_failure<typename std::remove_reference<decltype(s)>::type::value_type, types_collection_type>(s);

 for (auto& s: failure_strings_char32_t)
  test_utf_to_utf_failure<typename std::remove_reference<decltype(s)>::type::value_type, types_collection_type>(s);
}

// iterate over std::tuple T types
template<typename T, typename Collection, size_t index = 0>
void test_is_valid_utf_failure(std::basic_string<T>& s)
{
 BOOST_CHECK_MESSAGE(unicode::is_valid_utf<T>(s) == false, "Expected bad UTF at index: " << index << ", " << typeid(T).name());

 // iterate over remaining types 
 if constexpr (index + 1 < std::tuple_size<Collection>::value)
  test_is_valid_utf_failure<T, Collection, index + 1>(s);
}

BOOST_AUTO_TEST_CASE(is_valid_utf_failure)
{
 for (auto& s: failure_strings_char8_t)
  test_is_valid_utf_failure<typename std::remove_reference<decltype(s)>::type::value_type, types_collection_type>(s);
 
 for (auto& s: failure_strings_char16_t)
  test_is_valid_utf_failure<typename std::remove_reference<decltype(s)>::type::value_type, types_collection_type>(s);

 for (auto& s: failure_strings_char32_t)
  test_is_valid_utf_failure<typename std::remove_reference<decltype(s)>::type::value_type, types_collection_type>(s);
}

BOOST_AUTO_TEST_CASE(is_valid_unicode)
{
 BOOST_CHECK(unicode::is_valid_unicode('\0'));
 BOOST_CHECK(unicode::is_valid_unicode(U'a'));
 BOOST_CHECK(unicode::is_valid_unicode(U'ä'));
 BOOST_CHECK(unicode::is_valid_unicode(U'\u732b')); // cat chinese
 BOOST_CHECK(unicode::is_valid_unicode(U'\U0001F63A')); // cat chinese
 BOOST_CHECK(unicode::is_valid_unicode(0x0001F63A)); // cat smiley

 BOOST_CHECK(!unicode::is_valid_unicode(0x00110000));
 BOOST_CHECK(!unicode::is_valid_unicode(0xFFFFFFFF)); // U"\UFFFFFFFF" is invalid C++
 BOOST_CHECK(!unicode::is_valid_unicode(0x01234567));
 BOOST_CHECK(!unicode::is_valid_unicode(0x12345678));
 BOOST_CHECK(!unicode::is_valid_unicode(0xD800));
 BOOST_CHECK(!unicode::is_valid_unicode(0xD987));
 BOOST_CHECK(!unicode::is_valid_unicode(0xDFFF));
}

struct random_context {
 std::random_device rd;  // OS random number engine to seed RNG (below)
 std::mt19937 gen{rd()};
 std::uniform_int_distribution<> sequence_length{0, 100000}; // length of sequence: 0 ... 100000 code units
};

template<typename T>
T generate_random(random_context& rc, size_t length)
{
 std::uniform_int_distribution<> code_unit(0, std::numeric_limits<typename T::value_type>::max()); // code unit value
 T result;
 std::generate_n(std::back_inserter(result), length, [&](){return code_unit(rc.gen);});

 return result;
}

template<typename From, typename ToTypesCollectionType, size_t i = 0>
void test_random(random_context& rc, size_t length)
{
 //std::cerr << "LENGTH: " << length << std::endl;
 typedef typename std::tuple_element<i,ToTypesCollectionType>::type To;

 From r {generate_random<From>(rc, length)};

 try {
  To result{unicode::utf_to_utf<typename From::value_type,typename To::value_type>(r)};
 } catch (const std::runtime_error&) {
  // OK: this is an expected exception for utf_to_utf on bad input
 } catch (const std::invalid_argument&) {
  // OK: this is an expected exception for utf_to_utf on bad input
 }

 //std::cerr << "DEBUG: " << typeid(From).name() << std::endl;
 //std::cerr << " DEBUG2: " << typeid(To).name() << std::endl;

 // iterate over remaining To types
 if constexpr (i + 1 < std::tuple_size<ToTypesCollectionType>::value)
  test_random<From, ToTypesCollectionType, i + 1>(rc, length);
}

BOOST_AUTO_TEST_CASE_TEMPLATE(random_sequences, T, types_collection_type)
{
 random_context rc;

 // run for 1s (debug) 10s (release)
#ifdef _DEBUG
 const auto timeout{1.0s};
#else
 const auto timeout{10.0s};
#endif

 auto timeout_stamp { std::chrono::steady_clock::now() + (timeout / std::tuple_size<types_collection_type>::value)};

 while (!(std::chrono::steady_clock::now() > timeout_stamp)) {
  test_random<T,types_collection_type>(rc, rc.sequence_length(rc.gen));
 }
}

// TODO:
//
// char8_t, char16_t, char32_t, char, wchar_t (UTF-16 on Windows, UTF-32 on Linux)
// string, vector?
// uint8_t, uint16_t, uint32_t?