feat(abigen): add EthAbiCodec proc macro

[mattsse]

Motivation

Add AbiEncode + AbiDecode implementation for custom structs.
Currently, those get derived for EthCall only.

Solution

Add EthAbiCodec as a complementary proc macro to EthAbiType that derives encode/decode.
The reason why this needs to be separate and not included in EthAbiType is because in EthCall we have an additional leading 4byte selector, so EthCall needs to implement the codec traits but requires all the tokenized support that EthAbiType provides.
So we simply add an additional EthAbiCodec derive when we generate AbiEncoderV2 structs within abigen!

PR Checklist

• Added Tests
• Added Documentation
• Updated the changelog

[mattsse]

Fixed clippy lints on #705

[sveitser]

Awesome. It works well on the struct I was testing with.

I noticed something that seems a bit strange to me. In order to abi.decode in the contract I am wrapping the struct in a one-element tuple then tokenize and encode.

If I use the new decode method on that I do get and extra 32 in front (compared to doing into_tokens followed by encode).

Tried to illustrate this below

    abigen!(
        Example,
        r#"[
        struct Foo { uint256[] a; }
        function readFoo(bytes b)
        ]"#,
        event_derives(serde::Deserialize, serde::Serialize);
    );

    let a = vec![U256::from(255)];
    let foo = Foo { a };

    fn disp(b: &[u8]) {
        hex::encode(b)
            .chars()
            .collect::<Vec<char>>()
            .chunks(64)
            .for_each(|x| println!("{}", x.iter().collect::<String>()));
        println!("");
    }

    let encoded = foo.clone().encode();
    disp(&encoded);
    // 0000000000000000000000000000000000000000000000000000000000000020
    // 0000000000000000000000000000000000000000000000000000000000000001
    // 00000000000000000000000000000000000000000000000000000000000000ff

    // same as encode above
    let from_tokens = abi::encode(&Tokenize::into_tokens(foo.clone()));
    disp(&from_tokens);
    // 0000000000000000000000000000000000000000000000000000000000000020
    // 0000000000000000000000000000000000000000000000000000000000000001
    // 00000000000000000000000000000000000000000000000000000000000000ff

    let encoded_wrapped = (foo.clone(),).encode();
    disp(&encoded_wrapped);
    // 0000000000000000000000000000000000000000000000000000000000000020  X
    // 0000000000000000000000000000000000000000000000000000000000000020  X
    // 0000000000000000000000000000000000000000000000000000000000000020
    // 0000000000000000000000000000000000000000000000000000000000000001
    // 00000000000000000000000000000000000000000000000000000000000000ff

    // this works with abi.decode(b, (Foo)) in a contract
    let from_tokens_wrapped = abi::encode(&Tokenize::into_tokens((foo.clone(),)));
    disp(&from_tokens_wrapped);
    // 0000000000000000000000000000000000000000000000000000000000000020  X
    // 0000000000000000000000000000000000000000000000000000000000000020
    // 0000000000000000000000000000000000000000000000000000000000000001
    // 00000000000000000000000000000000000000000000000000000000000000ff

[mattsse]

nice catch!
I think I fixed tuple codec now, could you try again please @sveitser?

[sveitser]

Yes. I'm getting matching outputs now for the tuple 🥳

    let encoded_wrapped = (foo.clone(),).encode();
    disp(&encoded_wrapped);
    // 0000000000000000000000000000000000000000000000000000000000000020  X
    // 0000000000000000000000000000000000000000000000000000000000000020
    // 0000000000000000000000000000000000000000000000000000000000000001
    // 00000000000000000000000000000000000000000000000000000000000000ff

[gakonst]

this tuple syntax seems like a footgun? can we get rid of the x,) requirement?

[mattsse]

it's a bit weird but this ist the required rust syntax for single-element tuple, otherwise it would just be a parenthesized expression

https://doc.rust-lang.org/1.30.0/book/second-edition/appendix-02-operators.html