c++相关编码转换

十六进制转ascii

std::string hexToAscii(const std::string& hexString) {
    std::string asciiString;

    for (size_t i = 0; i < hexString.length(); i += 2) {
        if(hexString[i]=='\\')
        {
            continue;
        }
        // 从两个字符表示的十六进制数转换为整数
        int decimal = std::stoi(hexString.substr(i, 2), nullptr, 16);

        // 将整数转换为对应的 ASCII 字符
        asciiString += static_cast<char>(decimal);
    }

    return asciiString;
}

unicode转UTF8

std::string unicodeToUTF8(const std::string& unicodeStr)
{
    std::string utf8Str;
    for (std::size_t i = 0; i < unicodeStr.size(); i += 6)
    {
        std::string unicodeCode = unicodeStr.substr(i + 2, 4);
        unsigned long unicodeValue = std::stoul(unicodeCode, nullptr, 16);
        if (unicodeValue <= 0x007F)
        {
            utf8Str += static_cast<char>(unicodeValue);
        }
        else if (unicodeValue <= 0x07FF) {
            utf8Str += static_cast<char>(0xC0 | ((unicodeValue >> 6) & 0x1F));
            utf8Str += static_cast<char>(0x80 | (unicodeValue & 0x3F));
        }
        else {
            utf8Str += static_cast<char>(0xE0 | ((unicodeValue >> 12) & 0x0F));
            utf8Str += static_cast<char>(0x80 | ((unicodeValue >> 6) & 0x3F));
            utf8Str += static_cast<char>(0x80 | (unicodeValue & 0x3F));
        }
    }
    return utf8Str;
}

UTF8转unicode

std::string to_hex_string(unsigned long value, int width)
{
    std::stringstream ss;
    ss << std::hex << std::uppercase << std::setw(width) << std::setfill('0') << value;

    std::string hexStr = "\\u" + ss.str();

    // 转换为小写形式
    for (char& c : hexStr)
    {
        c = std::tolower(c);
    }

    return hexStr;
}

std::string utf8ToUnicode(const std::string& utf8Str)
{
    std::string unicodeStr;
    std::size_t i = 0;
    while (i < utf8Str.size())
    {
        unsigned char byte1 = static_cast<unsigned char>(utf8Str[i]);
        if ((byte1 & 0x80) == 0) // 1个字节
        {
            unicodeStr += utf8Str[i];
            i++;
        }
        else if ((byte1 & 0xE0) == 0xC0) // 2个字节
        {
            unsigned char byte2 = static_cast<unsigned char>(utf8Str[i + 1]);
            unsigned long unicodeValue = ((byte1 & 0x1F) << 6) | (byte2 & 0x3F);
            unicodeStr += to_hex_string(unicodeValue, 4); // 转换为Unicode编码
            i += 2;
        }
        else if ((byte1 & 0xF0) == 0xE0) // 3个字节
        {
            unsigned char byte2 = static_cast<unsigned char>(utf8Str[i + 1]);
            unsigned char byte3 = static_cast<unsigned char>(utf8Str[i + 2]);
            unsigned long unicodeValue = ((byte1 & 0x0F) << 12) | ((byte2 & 0x3F) << 6) | (byte3 & 0x3F);
            unicodeStr += to_hex_string(unicodeValue, 4); // 转换为Unicode编码
            i += 3;
        }
        else
        {
            // 不支持的UTF-8编码格式
            // 可以根据需要进行处理或抛出异常
            break;
        }
    }
    return unicodeStr;
}