testUTF8.cpp

testUTF8.cpp

/*
  testUTF8.cpp

  https://onihusube.hatenablog.com/entry/2020/04/03/211442
*/

// defined .cbp
// #define UNICODE
// #define _UNICODE
#include <wchar.h>

#include <windows.h>

#include <testUTF8.h>

#include <exception>
#include <stdexcept>

#include <iomanip>
#include <iostream>
#include <fstream>
#include <sstream>
#include <vector>
#include <string>
#include <memory>

#if 0 // def __GNUC__
#include <string_view>
using namespace std::string_view_literals;
#endif

using namespace std;

#define TXT_WS L"(森鷗外𠮟る　🤔 😢 🙇<200d>♂️ 🎉 😰 😊 😭 😥 終端)"
#define TXT_U8S "(森鷗外𠮟る　🤔 😢 🙇<200d>♂️ 🎉 😰 😊 😭 😥 終端)"
#define TXT_U8RSV u8R"(森鷗外𠮟る　🤔 😢 🙇<200d>♂️ 🎉 😰 😊 😭 😥 終端)"sv

#define FN_WS "out_wc16.txt"
#define FN_U8 "out_utf8.txt"
#define FN_SJ "out_cp932.txt"

int main(int ac, char **av)
{
  cout.imbue(locale(""));
  wcout.imbue(locale(""));
  cout << "Hello world!" << endl;

#if 0 // def __GNUC__
  auto u8s = TXT_U8RSV;
  cout << reinterpret_cast<const char *>(u8s.data()) << endl;
#endif

   const wchar_t *ws = TXT_WS;
   const char *u8s = TXT_U8S;
   int wlen, len;

   wlen = lstrlen(ws);
   wcout << wlen << ws << endl; // 43
if(cout.fail()) cout.clear();
if(wcout.fail()) wcout.clear();

   len = WideCharToMultiByte(CP_UTF8, 0, ws, wlen, NULL, 0, NULL, NULL);
   cout << len << endl; // 79 (CP_UTF8)
   string u8scnv(len, '\0');
   len = WideCharToMultiByte(CP_UTF8, 0, ws, wlen,
     &u8scnv[0], u8scnv.length(), NULL, NULL);
   cout << len << u8scnv << endl; // 79
if(cout.fail()) cout.clear();
if(wcout.fail()) wcout.clear();
   ofstream ofsu8(FN_U8, ios::out | ios::binary);
   ofsu8.write(&u8scnv[0], len); // OK

   len = WideCharToMultiByte(CP_ACP, 0, ws, wlen, NULL, 0, NULL, NULL);
   cout << len << endl; // 50 (CP_ACP)
   string sjscnv(len, '\0');
   len = WideCharToMultiByte(CP_ACP, 0, ws, wlen,
     &sjscnv[0], sjscnv.length(), NULL, NULL);
   cout << len << sjscnv << endl; // 50
if(cout.fail()) cout.clear();
if(wcout.fail()) wcout.clear();
   ofstream ofssj(FN_SJ, ios::out | ios::binary);
   ofssj.write(&sjscnv[0], len); // skipped non ASCII (same as console)

   len = strlen(u8s);
   cout << len << u8s << endl; // 79

   wlen = MultiByteToWideChar(CP_UTF8, 0, u8s, len, NULL, 0);
   cout << wlen << endl; // 43
   wstring wscnv(wlen, L'\0');
   wlen = MultiByteToWideChar(CP_UTF8, 0, u8s, len, &wscnv[0], wscnv.length());
   wcout << wlen << wscnv << endl; // 43 wcout need wcout.imbue(locale(""));
if(cout.fail()) cout.clear();
if(wcout.fail()) wcout.clear();
   ofstream ofsws(FN_WS, ios::out | ios::binary);
   ofsws.write((char *)(&wscnv[0]), wlen * sizeof(wchar_t)); // OK (LE)

  cout << endl;
  cout << TXT_ABC << endl;
  try{
    throw runtime_error("xyz");
  }catch(runtime_error &e){
    cerr << "runtime: [" << e.what() << "]" << endl;
  }catch(exception &e){
    cerr << "exception: [" << e.what() << "]" << endl;
  }catch(...){
    cerr << "exception: [UNKNOWN]" << endl;
  }
  return 0;
}