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 "../Locale.hpp"
#include "Encoding.hpp"

namespace Aurora::Locale::Encoding
{
    AUKN_SYM AuOptional<AuPair<ECodePage, AuUInt8>> DecodeBOM(const Aurora::Memory::MemoryViewRead & binary)
    {
    #define ADD_PATTERN(str, code) {str, AuArraySize(str) - 1, ECodePage::code}
        AuList<std::tuple<const char *, int, ECodePage>> bows = 
        {
            ADD_PATTERN("\xFF\xFE\x00\x00", eUTF32),
            ADD_PATTERN("\x00\x00\xFE\xFF", eUTF32BE),
            ADD_PATTERN("\x84\x31\x95\x33", e18030),
            ADD_PATTERN("\xDD\x73\x66\x73", eUnsupported), // UTF-EBCDIC
            ADD_PATTERN("\xEF\xBB\xBF", eUTF8), 
            ADD_PATTERN("\xF7\x64\x4C", eUnsupported),  // UTF-1
            ADD_PATTERN("\xFB\xEE\x28", eUTF7),  // UTF-7
            ADD_PATTERN("\x2B\x2F\x76", eUnsupported),  // BOCU-1
            ADD_PATTERN("\x0E\xFE\xFF", eUnsupported),  // SCSU
            ADD_PATTERN("\xFF\xFE", eUTF16),  // UTF-16
            ADD_PATTERN("\xFE\xFF", eUTF16BE)   // UTF-16
        };
    #undef ADD_PATTERN

        for (const auto &[string, length, category] : bows)
        {
            if (binary.length < length) continue;
            if (std::memcmp(binary.ptr, string, length) != 0) continue;
            
            return AuMakePair(category, length);
        }

        return {};
    }
    
    AUKN_SYM AuStreamReadWrittenPair_t EncodeUTF8(const Aurora::Memory::MemoryViewRead & utf8, const Aurora::Memory::MemoryViewWrite & binary, ECodePage page)
    {
        TextStreamEncoder re(page);
        return re.DecodeUTF8(utf8.ptr, utf8.length, binary.ptr, binary.length);
    }

    AUKN_SYM AuStreamReadWrittenPair_t DecodeUTF8(const Aurora::Memory::MemoryViewRead & binary, const Aurora::Memory::MemoryViewWrite & utf8, ECodePage page)
    {
        TextStreamProcessor re(page);
        return re.EncodeUTF8(binary.ptr, binary.length, utf8.ptr, utf8.length);
    }

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