AuroraRuntime/Include/Aurora/Locale/Encoding/Encoding.hpp

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

    File: Encoding.hpp
    Date: 2021-8-20
    Author: Reece
***/
#pragma once

namespace Aurora::Locale::Encoding
{
    struct BOM
    {
        ECodePage page;
        AuUInt8 length;
    };

    // Attempt to guess the code page signature of the string view provided by the binary view
    AUKN_SYM BOM DecodeBOM(const Memory::MemoryViewRead &binary);

    // General purpose arbitrary page to UTF8 (AuStrings are UTF-8 - not 16 or 32; bite me)
    AUKN_SYM AuStreamReadWrittenPair_t EncodeUTF8(const Memory::MemoryViewRead &utf8, const Memory::MemoryViewWrite &binary, ECodePage page = ECodePage::eUnsupported);
    AUKN_SYM AuStreamReadWrittenPair_t DecodeUTF8(const Memory::MemoryViewRead &binary, const Memory::MemoryViewWrite &utf8, ECodePage page = ECodePage::eUnsupported);
    AUKN_SYM AuStreamReadWrittenPair_t DecodeUTF8(const Memory::MemoryViewRead &binary, AuString &out, ECodePage page = ECodePage::eUnsupported);
    
    // Optimized UTF translation functions
    AUKN_SYM AuStreamReadWrittenPair_t ReadUTF32IntoUTF8ByteString(const Memory::MemoryViewRead &utf32, const Memory::MemoryViewWrite &utf8);
    AUKN_SYM AuStreamReadWrittenPair_t ReadUTF8IntoUTF32ByteString(const Memory::MemoryViewRead &utf8, const Memory::MemoryViewWrite &utf32);

    // Endianswap functions, could be subject to SIMD optimizations
    AUKN_SYM void SwapUTF32(const Memory::MemoryViewWrite &utf32);
    AUKN_SYM void SwapUTF16(const Memory::MemoryViewWrite &utf16);

    // Counst the amount of codepoints in a buffer, breaking when the stream is incomplete, giving you the accurate amount of bytes or relevant codepoints in a stream view
    AUKN_SYM AuUInt32 CountUTF32Length(const Memory::MemoryViewRead &utf32, bool bytes = false); // codepoint = U32 encoded; always 4 bytes per codepoint
    AUKN_SYM AuUInt32 CountUTF16Length(const Memory::MemoryViewRead &utf16, bool bytes = false); // codepoint = U32 encoded; at most: 4 bytes per codepoint, usual: 2 bytes
    AUKN_SYM AuUInt32 CountUTF8Length(const Memory::MemoryViewRead &utf8, bool bytes = false); // codepoint = U32 encoded; at most: 6 bytes per codepoint
    AUKN_SYM AuUInt32 CountSJISLength(const Memory::MemoryViewRead &sjis, bool bytes = false); // codepoint = one character
    AUKN_SYM AuUInt32 CountGBK16Length(const Memory::MemoryViewRead &gbk, bool bytes = false); // codepoint = at most; one GBK byte pair
}