CBOR byte string support

[daviddesberg]

Currently, the CBOR feature of the library only supports writing strings out as UTF-8 and doesn't support byte arrays as described in the CBOR standard. I'd like to implement COSE on top of this library, but in order to do that, byte string support is needed.

I'm happy to submit a pull request for this, but I wanted to check in advance if this is something you'd be open to

[sijohans]

This would be a nice an useful feature. This library can be used for creating tests towards a c target that parses CBOR and the library allows for writing easy to understand code when creating the data that the target should handle.

For test the following changes have been made:

• If a string start withs "0x" and the following data is a valid hex representation.Encode/decode it as CBOR bytes. Could also perhaps add a flags to the to_cbor()/from_cbor() to enable this feature.

The problem with this is the cases when sending an actual string that starts with "0x" and has a valid hex representation following. For the above mentioned use case I am fine with this corner case, but I understand that this might not be acceptable for the library in general.

Interesting is how the CBOR playground (http://cbor.me/) handles/presents this:

CBOR Representation
A1               # map(1)
   65            # text(5)
      6279746573 # "bytes"
   45            # bytes(5)
      0011223344 # "\x00\x11\"3D"

JSON Representation
{"bytes": h'0011223344'}

Can this issue be reopened?

[ecorm]

Check out the WAMP protocol for ideas on JSON-encoding a binary blob: https://wamp-proto.org/_static/gen/wamp_latest.html#binary-conversion-of-json-strings

The basic idea is to prepend a Base64-encoded string with a nul character.

[OmnipotentEntity]

For those still following this issue, #1662 addresses this issue, and is finally in a complete state, and is ready for review and merge.

[nlohmann]

The functionality was added by merging #1662.