eth/client: support resolving ENS domains

[mculp]

I think this may tie or interfere with the implementation of #72 also.

[q9f]

An ::EnsResolver could be part of the client. Once you instantiate a client, it should have the ability to resolve ::Ens to ::Address

[arenzel]

That would be great!

[q9f]

NameHash: https://docs.ens.domains/contract-api-reference/name-processing

Resolver: https://docs.ens.domains/dapp-developer-guide/resolving-names

[q9f]

same address on all networks: 0x00000000000C2E074eC69A0dFb2997BA6C7d2e1e

https://etherscan.io/address/0x00000000000C2E074eC69A0dFb2997BA6C7d2e1e#readContract

[mculp]

I'll work on this next

[mculp]

Working on this now.

[dansimpson]

@mculp / @q9f I was looking for this as well, and started this here: https://github.com/withfabricxyz/eth.rb/commit/7194b7e4a3af6f4bb2b1578bb410390697aafac3

The biggest issue is in normalizing the domain. The ICU libraries for ruby don't seem to do exactly what is needed. This will work for basic names, but it's not properly encoding tokens with Cyrillic characters, etc. I tried a few libraries, which generate puny code symbols, but that isn't quite enough. See: https://github.com/jcranmer/idna-uts46 , which is used by https://github.com/ensdomains/eth-ens-namehash/blob/master/index.js#L26

[q9f]

Thank you, Dan.

I would love to consider accepting this for now. Some thoughts:

• adding a test case to resolve a name on mainnet (we can set an INFURA_ID in CI)
• maybe renaming Eth::Ens::Resolver::DEFAULT_ADDRESS to Eth::Ens::DEFAULT_REGISTRY or something meaningul in the ::Ens namespace
• We already deploy the registry to our test chain, maybe we can try to set a name and resolve it straight away? https://github.com/q9f/eth.rb/blob/main/spec/eth/client_spec.rb#L107-L113

In general, we can split this up in multiple PRs if you want to only submit the resolver for now and look into the encoding issues later, I'd be happy to review it.

[mculp]

@dansimpson The encoding issue was the issue I ran into as well when trying to work on this. I was trying to follow the ENS spec to the T and couldn't seem to find a library that correctly encoded all names.

Thanks for doing this

[mculp]

@dansimpson FYI this is the recursive namehash implementation that I had. I ported it from the python reference implementation.

      def namehash(name)
        if name.empty?
          "\0" * 32
        else
          label, _, remainder = name.partition('.')
          Eth::Util.keccak256(namehash(remainder) + Eth::Util.keccak256(label))
        end
      end

[mculp]

@dansimpson If you open a PR, I have a couple pretty minor comments I'd like to make on your changes.

[musik]

@mculp

  def namehash(name)
    if name.empty?
      Eth::Util.prefix_hex("00"*32)
    else
      label, _, remainder = name.partition('.')
      Eth::Util.prefix_hex(
        sha3(Eth::Util.hex_to_bin(namehash(remainder) + sha3(label)))
      )
    end
  end  
  def sha3 str
    Eth::Util.bin_to_hex(Eth::Util.keccak256(str))
  end  
  def labelhash label
    Eth::Util.prefix_hex Eth::Util.bin_to_hex(Eth::Util.keccak256(label))
  end

[mculp]

@musik yeah, once this gets merged these conversions should be easier

[q9f]

#150

#192

infura = Eth::Client.create("https://mainnet.infura.io/v3/#{ENV['INFURA_TOKEN']}")
# => #<Eth::Client::Http:0x00005614b684fd70
infura.resolve_ens "ncwc6edqldzy6mlo.eth"
# => "0xde270e46d63b1816d1b798cff473c4ba238aca73"

[q9f]

https://github.com/q9f/eth.rb/wiki/ENS-Resolver