AuroraRuntime/Source/Locale/Encoding/UTFn/AuUTF16.hpp

/***
    Copyright (C) 2021 J Reece Wilson (a/k/a "Reece"). All rights reserved.

    File: AuUTF16.hpp
    Date: 2021-10-31
    Author: Reece
***/
#pragma once

namespace Aurora::Locale::Encoding::UTF16
{
    static void SwapU16(void *base, AuUInt32 count)
    {
        count *= 2;
        for (int i = 0; i < count; i += 2)
        {
            AuWriteU16BE(base, i, AuReadU16LE(base, i));
        }
    }

    #define AU_HIGH_SURROGATE_START  0xd800
    #define AU_HIGH_SURROGATE_END    0xdbff
    #define AU_LOW_SURROGATE_START   0xdc00
    #define AU_LOW_SURROGATE_END     0xdfff
    #define AU_IS_HIGH_SURROGATE(wch) (((wch) >= AU_HIGH_SURROGATE_START) && ((wch) <= AU_HIGH_SURROGATE_END))
    #define AU_IS_LOW_SURROGATE(wch)  (((wch) >= AU_LOW_SURROGATE_START)  && ((wch) <= AU_LOW_SURROGATE_END))
    
    static int GetLenUC2CodePoint(const AuUInt8 *in, AuUInt32 len)
    {
        if (len < 2) return 0;
        auto a = *reinterpret_cast<const AuUInt16 *>(in);
        if (!AU_IS_HIGH_SURROGATE(a)) return 2;
        
        if (len < 4) return 0;
        auto b = *reinterpret_cast<const AuUInt16 *>(in + 2);
        return AU_IS_LOW_SURROGATE(b) ? 4 : 0;
    }
    
    static int Count16(const void *base, AuUInt32 length, bool bytes = false)
    {
        AuUInt32 i {}, cps {};
        
        for (; i < length; )
        {
            auto next = GetLenUC2CodePoint(((const AuUInt8 *)base) + i, length - i);
            if (next == 0) return i;
            i += next;
            cps++;
        }

        return bytes ? i : cps;
    }
}