copy member variable into byte vector

Question

I want to copy a 64-bit member variable into a vector byte by byte.

Please avoid telling me to use bit operation to extract each byte and then copy them into vector.

I want to do this by one line.

I use memcpy and copy methods, but both of them failed.

Here is the sample code:

#include <iostream>
#include <vector>
#include <cstdint>
#include <cstring>

using namespace std;

class A {
 public:
  A()
  : eight_bytes_data(0x1234567812345678) {
  }

  void Test() {
    vector<uint8_t> data_out;
    data_out.reserve(8);
    memcpy(data_out.data(),
           &eight_bytes_data,
           8);
    cerr << "[Test]" << data_out.size() << endl;
  }

  void Test2() {
    vector<uint8_t> data_out;
    data_out.reserve(8);
    copy(&eight_bytes_data,
         (&eight_bytes_data) + 8,
         back_inserter(data_out));
    cerr << "[Test2]" << data_out.size() << endl;
    for (auto value : data_out) {
      cerr << hex << value << endl;
    }
  }

 private:
  uint64_t eight_bytes_data;
};

int main() {
  A a;
  a.Test();
  a.Test2();
  return 0;
}

Answer 1

If you want to work with the bytes of another type structure, you could use a char* to manipulate each byte:

void Test3()
{
    vector<uint8_t> data_out;
    char* pbyte = (char*)&eight_bytes_data;

    for(int i = 0; i < sizeof(eight_bytes_data); ++i)
    {
        data_out.push_back(pbyte[i]);
    }

    cerr << "[Test]" << data_out.size() << endl;

}

Unfortunately, you requested a one-line-solution, which I don't think is viable.

Answer 2

As the others already showed where you were getting wrong, there is a one line solution that is dangeurous.

First you need to make sure that you vector has enough size to receive 8 bytes. Something like this:

data_out.resize(8);

The you can do a reinterpret_cast to force your compiler to interpret those 8 bytes from the vector to be seen as an unique type of 8 bytes, and do the copy

*(reinterpret_cast<uint64_t*>(data_out.data())) = eight_bytes_data;

I can't figure out all the possibilities of something going wrong. So use at your own risk.

Answer 3

If you are interested in more generic version:

namespace detail
{
    template<typename Byte, typename T>
    struct Impl
    {
        static std::vector<Byte> impl(const T& data)
        {
                std::vector<Byte> bytes;
                bytes.resize(sizeof(T)/sizeof(Byte));
                *(T*)bytes.data() = data;
                return bytes;
        }
    };

    template<typename T>
    struct Impl<bool, T>
    {
        static std::vector<bool> impl(const T& data)
        {
            std::bitset<sizeof(T)*8> bits(data);
            std::string string = bits.to_string();
            std::vector<bool> vector;
            for(const auto& x : string)
                vector.push_back(x - '0');
            return vector;
        }
    };
}

template<typename Byte = uint8_t,
         typename T>
std::vector<Byte> data_to_bytes(const T& data)
{
    return detail::Impl<Byte,T>::impl(data);
}

int main()
{
    uint64_t test = 0x1111222233334444ull;

    for(auto x : data_to_bytes<bool>(test))
        std::cout << std::hex << uintmax_t(x) << " ";
    std::cout << std::endl << std::endl;

    for(auto x : data_to_bytes(test))
        std::cout << std::hex << uintmax_t(x) << " ";
    std::cout << std::endl << std::endl;

    for(auto x : data_to_bytes<uint16_t>(test))
        std::cout << std::hex << uintmax_t(x) << " ";
    std::cout << std::endl << std::endl;

    for(auto x : data_to_bytes<uint32_t>(test))
        std::cout << std::hex << uintmax_t(x) << " ";
    std::cout << std::endl << std::endl;

    for(auto x : data_to_bytes<uint64_t>(test))
        std::cout << std::hex << uintmax_t(x) << " ";
    std::cout << std::endl << std::endl;

}

Output:

0 0 0 1 0 0 0 1 0 0 0 1 0 0 0 1 0 0 1 0 0 0 1 0 0 0 1 0 0 0 1 0 0 0 1 1 0 0 1 1 0 0 1 1 0 0 1 1 0 1
0 0 0 1 0 0 0 1 0 0 0 1 0 0

44 44 33 33 22 22 11 11

4444 3333 2222 1111

33334444 11112222

1111222233334444