utf16.hpp

Go to the documentation of this file.

 
#ifndef JOIN_DATA_UTF16_HPP
#define JOIN_DATA_UTF16_HPP
 
// libjoin.
#include <join/utils.hpp>
 
// C.
#include <cstddef>
#include <cstdint>
 
namespace join
{
    namespace utf16
    {
        constexpr bool isSurrogate (uint32_t unit) noexcept
        {
            return (unit - 0xD800) < 0x800;
        }
 
        constexpr bool isHighSurrogate (uint32_t unit) noexcept
        {
            return (unit - 0xD800) < 0x400;
        }
 
        constexpr bool isLowSurrogate (uint32_t unit) noexcept
        {
            return (unit - 0xDC00) < 0x400;
        }
 
        constexpr uint32_t highSurrogate (uint32_t codepoint) noexcept
        {
            return 0xD800 + ((codepoint - 0x10000) >> 10);
        }
 
        constexpr uint32_t lowSurrogate (uint32_t codepoint) noexcept
        {
            return 0xDC00 + (codepoint & 0x3FF);
        }
 
        constexpr uint32_t combine (uint32_t high, uint32_t low) noexcept
        {
            return 0x10000 + (((high - 0xD800) << 10) | (low - 0xDC00));
        }
 
        inline int decode (const char16_t*& cur, const char16_t* end, uint32_t& codepoint) noexcept
        {
            uint32_t u0 = cur[0];
 
            if (JOIN_LIKELY (!isSurrogate (u0)))
            {
                codepoint = u0;
                ++cur;
                return 0;
            }
 
            if (JOIN_UNLIKELY (!isHighSurrogate (u0) || (end - cur) < 2 || !isLowSurrogate (cur[1])))
            {
                return -1;
            }
 
            codepoint = combine (u0, cur[1]);
            cur += 2;
            return 0;
        }
 
        inline size_t encode (uint32_t codepoint, char16_t* out) noexcept
        {
            if (JOIN_LIKELY (codepoint < 0x10000))
            {
                out[0] = static_cast<char16_t> (codepoint);
                return 1;
            }
 
            out[0] = static_cast<char16_t> (highSurrogate (codepoint));
            out[1] = static_cast<char16_t> (lowSurrogate (codepoint));
            return 2;
        }
    }
}
 
#endif