Search code examples
ethereumsolidity

Solidity: How to represent bytes32 as string

This may be simple in other languages but I can't figure out how to do it in Solidity.
I have a bytes32 like this 0x05416460deb76d57af601be17e777b93592d8d4d4a4096c57876a91c84f4a712.

I don't want to convert the bytes to a string, rather I just want to represent the whole thing as a string, like "0x05416460deb76d57af601be17e777b93592d8d4d4a4096c57876a91c84f4a712".
How can this be done in Solidity?

Update:
Why I need to do this: Basically I connect to an oracle, which does some work off-chain and finally uploads a file to IPFS. I need to get the content identifier into my contract from the oracle. The oracle can only send bytes32 as a response, so I convert it to a multihash and send only the digest as bytes32 from oracle to contract.
So far so good, I can recreate the multihash in my contract. The problem is that after this I create an ERC721 (NFT) token and I have to store some reference to the IPFS file in the metadata, which can only be in string format. This is where I'm stuck at the moment.

Solution

  • While the answer from @Burt looks correct (didn't test it though), there is a much more efficient way to solve the same task:

    function toHex16 (bytes16 data) internal pure returns (bytes32 result) {
    result = bytes32 (data) & 0xFFFFFFFFFFFFFFFF000000000000000000000000000000000000000000000000 |
          (bytes32 (data) & 0x0000000000000000FFFFFFFFFFFFFFFF00000000000000000000000000000000) >> 64;
    result = result & 0xFFFFFFFF000000000000000000000000FFFFFFFF000000000000000000000000 |
          (result & 0x00000000FFFFFFFF000000000000000000000000FFFFFFFF0000000000000000) >> 32;
    result = result & 0xFFFF000000000000FFFF000000000000FFFF000000000000FFFF000000000000 |
          (result & 0x0000FFFF000000000000FFFF000000000000FFFF000000000000FFFF00000000) >> 16;
    result = result & 0xFF000000FF000000FF000000FF000000FF000000FF000000FF000000FF000000 |
          (result & 0x00FF000000FF000000FF000000FF000000FF000000FF000000FF000000FF0000) >> 8;
    result = (result & 0xF000F000F000F000F000F000F000F000F000F000F000F000F000F000F000F000) >> 4 |
          (result & 0x0F000F000F000F000F000F000F000F000F000F000F000F000F000F000F000F00) >> 8;
    result = bytes32 (0x3030303030303030303030303030303030303030303030303030303030303030 +
           uint256 (result) +
           (uint256 (result) + 0x0606060606060606060606060606060606060606060606060606060606060606 >> 4 &
           0x0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F) * 7);
}

function toHex (bytes32 data) public pure returns (string memory) {
    return string (abi.encodePacked ("0x", toHex16 (bytes16 (data)), toHex16 (bytes16 (data << 128))));
}

    This code produces upper case output. For lower case output, just change 7 to 39 in the code.

    Explanation

    The idea is to process 16 bytes at once using binary operations.

    The toHex16 function converts a sequence of 16 bytes represented as a bytes16 value into a sequence of 32 hexadecimal digits represented as a bytes32 value. The toHex function splits a bytes32 value into two bytes16 chunks, converts each chunk to hexadecimal representation via the toHex16 function, and finally concatenates the 0x prefix with the converted chunks using abi.encodePacked function.

    The most sophisticated part is how the toHex16 function works. Let's explain it sentence by sentence.

    The first sentence:

    result = bytes32 (data) & 0xFFFFFFFFFFFFFFFF000000000000000000000000000000000000000000000000 |
      (bytes32 (data) & 0x0000000000000000FFFFFFFFFFFFFFFF00000000000000000000000000000000) >> 64;

    Here we shift the last 64 bits of the input to the right by 64 bits, basically doing:

    0123456789abcdeffedcba9876543210
\______________/\______________/
       |               |
       |               +---------------+
 ______V_______                  ______V_______
/              \                /              \
0123456789abcdef0000000000000000fedcba9876543210

    The second sentence:

    result = result & 0xFFFFFFFF000000000000000000000000FFFFFFFF000000000000000000000000 |
      (result & 0x00000000FFFFFFFF000000000000000000000000FFFFFFFF0000000000000000) >> 32;

    Here we shift the last 32 bits of both 64-bit chunks to the right by 32 bits:

    0123456789abcdef0000000000000000fedcba9876543210
\______/\______/                \______/\______/
   |       |                       |       |
   |       +-------+               |       +-------+
 __V___          __V___          __V___          __V___
/      \        /      \        /      \        /      \
012345670000000089abcdef00000000fedcba980000000076543210

    The next sentence:

    result = result & 0xFFFF000000000000FFFF000000000000FFFF000000000000FFFF000000000000 |
      (result & 0x0000FFFF000000000000FFFF000000000000FFFF000000000000FFFF00000000) >> 16;

    does:

    012345670000000089abcdef00000000fedcba980000000076543210
\__/\__/        \__/\__/        \__/\__/        \__/\__/
 |   |           |   |           |   |           |   |
 |   +---+       |   +---+       |   +---+       |   +---+
 V_      V_      V_      V_      V_      V_      V_      V_
/  \    /  \    /  \    /  \    /  \    /  \    /  \    /  \
012300004567000089ab0000cdef0000fedc0000ba980000765400003210

    And the next one:

    result = result & 0xFF000000FF000000FF000000FF000000FF000000FF000000FF000000FF000000 |
      (result & 0x00FF000000FF000000FF000000FF000000FF000000FF000000FF000000FF0000) >> 8;

    does:

    012300004567000089ab0000cdef0000fedc0000ba980000765400003210
\/\/    \/\/    \/\/    \/\/    \/\/    \/\/    \/\/    \/\/
| |     | |     | |     | |     | |     | |     | |     | |
| +-+   | +-+   | +-+   | +-+   | +-+   | +-+   | +-+   | +-+
V   V   V   V   V   V   V   V   V   V   V   V   V   V   V   V
/\  /\  /\  /\  /\  /\  /\  /\  /\  /\  /\  /\  /\  /\  /\  /\
01002300450067008900ab00cd00ef00fe00dc00ba00980076005400320010

    The final sentence in this series is a bit different:

    result = (result & 0xF000F000F000F000F000F000F000F000F000F000F000F000F000F000F000F000) >> 4 |
      (result & 0x0F000F000F000F000F000F000F000F000F000F000F000F000F000F000F000F00) >> 8;

    It shifts odd nibbles to the right by 4 bits, and even nibbles by 8 bits:

    01002300450067008900ab00cd00ef00fe00dc00ba00980076005400320010
|\  |\  |\  |\  |\  |\  |\  |\  |\  |\  |\  |\  |\  |\  |\  |\
\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \
 \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \
 | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
 V V V V V V V V V V V V V V V V V V V V V V V V V V V V V V V V
000102030405060708090a0b0c0d0e0f0f0e0d0c0b0a09080706050403020100

    So all the nibbles of the initial data are distributed one per byte.

    Now with every byte x we need to do the following transformation:

    x` = x < 10 ? '0' + x : 'A' + (x - 10)

    Let's rewrite this formula a bit:

    x` = ('0' + x) + (x < 10 ? 0 : 'A' - '0' - 10)
x` = ('0' + x) + (x < 10 ? 0 : 1) * ('A' - '0' - 10)

    Note, that (x < 10 ? 0 : 1) could be calculated as ((x + 6) >> 4), thus we have:

    x` = ('0' + x) + ((x + 6) >> 4) * ('A' - '0' - 10)
x` = (0x30 + x) + ((x + 0x06) >> 4) * 7

    The final statement:

    result = bytes32 (0x3030303030303030303030303030303030303030303030303030303030303030 +
       uint256 (result) +
       (uint256 (result) + 0x0606060606060606060606060606060606060606060606060606060606060606 >> 4 &
       0x0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F) * 7);

    Basically performs the above calculation for every byte. The

    0x0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F0F

    mask after the right shift is needed to zero out the bits "dropped" by the right shift in the original formula.

    BTW, it would be better to ask questions like this one at https://ethereum.stackexchange.com/