minicbor_derive: strings as indecies?

[alex-s168]

For my application, it would be really useful if I could do #[n("key")], on both #[cbor(map)] structs, and #[cbor(index_only)] enums...

[twittner]

Would one of the serde-based CBOR crates work for you, e.g. minicbor-serde?

[alex-s168]

I considered that too, but that's not flexible enough.

[twittner]

Since it is not the first time this feature was requested I started a work-in-progress PR (#53). There may still be show stoppers and the experiment may not work out, but if you are interested you can try that branch and see if it works for you.

Edit: The documentation has not been updated yet. The syntax for string indices is #[s("...")] and string indices are not allowed under array encoding.

[ropuls]

I built on top of PR #53 and added a #[cbor(text_keys)] attribute that automatically uses field names as string keys, so you don't have to annotate every field with #[s("...")].

#[derive(Encode, Decode)]
#[cbor(text_keys)]
struct SensorData {
    temperature: i16,
    humidity: u8,
}
// encodes as {"temperature": 23, "humidity": 65}

Individual keys can be renamed with #[cbor(key = "t")] if needed.

See PR #57 for details and tests.

Thanks for minicbor and the #[s("...")] groundwork in #53, that made this straightforward to build.

[Arisa-Snowbell]

This is exactly what I needed, using str-index branch as the source for minicbor right now, hope to see it merged one day <3 thank you for all the work