std::pair<Iterator, char32_t> to_utf32_unchecked(Iterator p)
{
if (!is_surrogate(*p))
{
// A simple case.
char32_t result = *p++;
return std::make_pair(p, result);
}
// Otherwise have to process a surrogate pair.
{
assert(is_leading(*p));
// 0x010000 .. 0x10FFFF
// Code point - 0x10000 ==
// 3210 FEDC BA98 7654 3210
// LLLL LLLL LLTT TTTT TTTT
char32_t result = 0x10000;
result += (*p++ - 0xD800) << 10;
result += (*p++ - 0xDC00);
return std::make_pair(p, result);
}
}
std::pair<Iterator, boost::optional<char32_t>> to_utf32(Iterator p, Iterator pe)
{
char32_t result;
if (!is_surrogate(*p))
{
// A simple case.
result = *p;
return std::make_pair(++p, result);
}
// Otherwise have to process a surrogate pair.
{
// 0x010000 .. 0x10FFFF
// Code point - 0x10000 ==
// 3210 FEDC BA98 7654 3210
// LLLL LLLL LLTT TTTT TTTT
result = 0x10000;
if (!is_leading(*p))
goto error; // A lead surrogate was expected.
result += (*p - 0xD800) << 10;
++p;
if (p == pe)
goto error; // Sequence has ended prematurely.
if (!is_trailing(*p))
goto error; // A trail surrogate was expected.
result += (*p - 0xDC00);
++p;
if (!is_valid_code_point(result))
goto error; // Sequence encodes an invalid code point.
return std::make_pair(p, result);
error:
return std::make_pair(p, boost::none);
}
}
inline bool is_valid (char_t c)
{
return !(is_surrogate(c) || c.value >= char_t::max_code_point);
}