AuroraRuntime/Source/Locale/Encoding/Encoding.cpp

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

    File: Encoding.cpp
    Date: 2021-8-19
    Author: Reece
***/
#include <Source/RuntimeInternal.hpp>
#include "Encoding.hpp"

namespace Aurora::Locale::Encoding
{
    /// OLD ENCODING API

    AUKN_SYM AuStreamReadWrittenPair_t EncodeUTF8(const AuMemoryViewRead &utf8, const AuMemoryViewWrite & binary, ECodePage page)
    {
        TextStreamEncoder re(page);
        return re.DecodeUTF8(utf8.ptr, utf8.length, binary.ptr, binary.length);
    }

    AUKN_SYM AuStreamReadWrittenPair_t DecodeUTF8(const AuMemoryViewRead &binary, const AuMemoryViewWrite & utf8, ECodePage page)
    {
        TextStreamProcessor re(page);
        return re.EncodeUTF8(binary.ptr, binary.length, utf8.ptr, utf8.length);
    }

    AUKN_SYM AuStreamReadWrittenPair_t DecodeUTF8(const AuMemoryViewRead &binary, AuString &out, ECodePage page)
    {
        auto aaa = DecodeUTF8(binary, {}, page);
        if (!AuTryResize(out, aaa.second))
        {
            return {};
        }
        auto ret = DecodeUTF8(binary, Memory::MemoryViewWrite(out.data(), out.size()), page);
        if (!AuTryResize(out, aaa.second))
        {
            return {};
        }
        return ret;
    }
        
    AuStreamReadWrittenPair_t DecodeUTF8(void *binary, AuUInt32 binaryLength, AuString &out, ECodePage page)
    {
        auto aaa = DecodeUTF8(Memory::MemoryViewRead(binary, binaryLength), {}, page);
        if (!AuTryResize(out, aaa.second))
        {
            return {};
        }
        auto ret = DecodeUTF8(Memory::MemoryViewRead(binary, binaryLength), Memory::MemoryViewWrite(out.data(), out.size()), page);
        if (!AuTryResize(out, aaa.second))
        {
            return {};
        }
        return ret;
    }

    /// Supporting full 6 byte UTF-8, copies or returns the available streams from @param utf8 to @param utf32

    AUKN_SYM AuStreamReadWrittenPair_t ReadUTF8IntoUTF32ByteString(const AuMemoryViewRead &utf8, const AuMemoryViewWrite &utf32)
    {
        return UTF32::UTF8ToCp(utf8, utf32);
    }

    AUKN_SYM AuStreamReadWrittenPair_t ReadUTF32IntoUTF8ByteString(const AuMemoryViewRead &utf32, const AuMemoryViewWrite &utf8)
    {
        return UTF32::CPToUTF8(utf32, utf8);
    }
    
    /// Endian swap
    
    AUKN_SYM void SwapUTF32(const AuMemoryViewWrite &utf32)
    {
        UTF32::SwapU32(utf32.Begin<AuUInt32>(), utf32.ToCount<AuUInt32>());
    }
    
    AUKN_SYM void SwapUTF16(const AuMemoryViewWrite &utf32)
    {
        UTF16::SwapU16(utf32.Begin<AuUInt32>(), utf32.ToCount<AuUInt32>());
    }
    
    /// String length awareness utilities
    
    AUKN_SYM AuUInt32 CountUTF32Length(const AuMemoryViewRead &utf32, bool bytes)
    {
        return UTF32::Count32(utf32.ptr, utf32.length, bytes);
    }

    AUKN_SYM AuUInt32 CountUTF16Length(const AuMemoryViewRead &utf16, bool bytes)
    {
        return UTF16::Count16(utf16.ptr, utf16.length, bytes, true);
    }

    AUKN_SYM AuUInt32 CountUTF16BELength(const AuMemoryViewRead &utf16, bool bytes)
    {
        return UTF16::Count16(utf16.ptr, utf16.length, bytes, false);
    }

    AUKN_SYM AuUInt32 CountUTF8Length(const AuMemoryViewRead &utf8, bool bytes)
    {
        auto pair = ReadUTF8IntoUTF32ByteString(utf8, {});
        return bytes ? pair.first : pair.second / sizeof(AuUInt32);
    }

    AUKN_SYM AuUInt32 CountSJISLength(const AuMemoryViewRead &sjis, bool bytes)
    {
        return SJIS::CountSJIS(sjis.ptr, sjis.length, bytes);
    }
    
    AUKN_SYM AuUInt32 CountGBK16Length(const AuMemoryViewRead &gbk, bool bytes)
    {
        return GBK::CountGbk(gbk.ptr, gbk.length, bytes);
    }

    AUKN_SYM AuUInt32 CountEncodedStringLength(ECodePage page, const Memory::MemoryViewRead &view, bool bytes)
    {
        switch (page)
        {
        case ECodePage::eGBK:
            return CountGBK16Length(view, bytes);
        case ECodePage::eUTF8:
            return CountUTF8Length(view, bytes);
        case ECodePage::eSJIS:
            return CountSJISLength(view, bytes);
        case ECodePage::eUTF32:
        case ECodePage::eUTF32BE:
            return CountUTF32Length(view, bytes);
        case ECodePage::eUTF16:
            return CountUTF16Length(view, bytes);
        case ECodePage::eUTF16BE:
            return CountUTF16BELength(view, bytes);
        default:
            return {};
        }
    }

    AUKN_SYM AuUInt32 IterateUTF32(const Memory::MemoryViewRead &utf32)
    {
        return utf32.length < 4 ? 0 : 4;
    }

    AUKN_SYM AuUInt32 IterateUTF16(const Memory::MemoryViewRead &utf16)
    {
        return UTF16::GetLenUC2CodePointLE(utf16.ToPointer(), utf16.length);
    }

    AUKN_SYM AuUInt32 IterateUTF16BE(const Memory::MemoryViewRead &utf16)
    {
        return UTF16::GetLenUC2CodePointBE(utf16.ToPointer(), utf16.length);
    }

    AUKN_SYM AuUInt32 IterateUTF8(const Memory::MemoryViewRead &utf8)
    {
        return UTF8::IterateUTF8(utf8);
    }

    AUKN_SYM AuUInt32 IterateSJIS(const Memory::MemoryViewRead &sjis)
    {
        return SJIS::GetLenSJISCodePoint(sjis.ToPointer(), sjis.length);
    }

    AUKN_SYM AuUInt32 IterateGBK16(const Memory::MemoryViewRead &gbk)
    {
        return GBK::GetLenGBKCodePoint(gbk.ToPointer(), gbk.length);
    }

    AUKN_SYM AuUInt32 IterateEncodedString(ECodePage page, const Memory::MemoryViewRead &view)
    {
        switch (page)
        {
        case ECodePage::eGBK:
            return IterateGBK16(view);
        case ECodePage::eUTF8:
            return IterateUTF8(view);
        case ECodePage::eSJIS:
            return IterateSJIS(view);
        case ECodePage::eUTF32:
        case ECodePage::eUTF32BE:
            return IterateUTF16(view);
        case ECodePage::eUTF16:
            return IterateUTF16(view);
        case ECodePage::eUTF16BE:
            return IterateUTF16BE(view);
        default:
            return {};
        }
    }
}