utf8 encoding test

utf8decode.cpp

#include <iostream>
#include <cstdint>
#include <cstdio>
#include <cassert>

// (C) 2017 by Sebastian 'bueddl' Büttner

void checkCodepoint(wchar_t codepoint)
{
  // Primary requirement for validity of XML
  assert((codepoint >= 0x1 && codepoint <= 0xD7FF) ||
         (codepoint >= 0xE000 && codepoint <= 0xFFFD) ||
         (codepoint >= 0x10000 && codepoint <= 0x10FFFF) == true);
}

int main()
{
  while (!std::cin.eof()) {
    wchar_t codepoint = 0;

    wchar_t ch = std::cin.get();
    if ((ch & 0200) != 0200) {
      // u+0000 - u+007f
      codepoint = ch;

      std::printf("U+%02x\n", codepoint);
      checkCodepoint(codepoint);
      continue;
    }

    ch <<= 6;
    ch |= std::cin.get() & 077;
    if ((ch & 034'000) != 034'000) {
      // u+0080 - u+07ff
      codepoint = ch & ~034'000;

      std::printf("U+%04x\n", codepoint);
      checkCodepoint(codepoint);
      continue;
    }

    ch <<= 6;
    ch |= std::cin.get() & 077;
    if ((ch & 03'600'000) != 03'600'000) {
      // u+0800 - u+ffff
      codepoint = ch & ~03'600'000;

      std::printf("U+%04x\n", codepoint);
      checkCodepoint(codepoint);
      continue;
    }

    ch <<= 6;
    // u+10000 - u+10ffff
    codepoint = ch |= std::cin.get() & 077 & ~0370'000'000;

    std::printf("U+%06x\n", codepoint);
    checkCodepoint(codepoint);
    continue;
  }

  return 0;
}