Feat: Add Regex Support For Email Verifier #1
Conversation
| let (is_public, dfa) = part; | ||
| let fwd = dense::DFA::from_bytes(&dfa.fwd).unwrap().0; | ||
| let rev = dense::DFA::from_bytes(&dfa.bwd).unwrap().0; | ||
|
|
||
| let re = Regex::builder().build_from_dfas(fwd, rev); | ||
| let matches: Vec<Match> = re.find_iter(email_body).collect(); | ||
|
|
||
| if !matches.is_empty() { | ||
| regex_verified = true; | ||
| if *is_public { | ||
| let substring = email_body[matches[0].start()..matches[0].end()].to_string(); | ||
| regex_matches.push(substring); | ||
| } | ||
| } |
There was a problem hiding this comment.
@shreyas-londhe
Could you explain why these processes are enough to verify the regex?
Specifically, what are roles of fwd and rev here?
There was a problem hiding this comment.
There is a bug in the regex matching which I'm fixing, and will reflect in #2.
Specifically, what are roles of
fwdandrevhere?
fwd and rev are the DFAs serialized out of the circuit and are fed to the circuit as bytes. In the circuit we just deserialize it and construct the Regex which is the most efficient for matching ops.
| #[derive(Debug, Serialize, Deserialize)] | ||
| pub struct PublicKey { | ||
| pub key: Vec<u8>, | ||
| pub key_type: String, |
There was a problem hiding this comment.
What is the role of key_type here?
There was a problem hiding this comment.
The key_type is a string which determines whether the public key is a RSA public key or ED25519 public key.
- Function where the key is used: https://github.com/risc0-labs/dkim/blob/3213315e41f426c307b8feb8fb329330a47e326c/src/lib.rs#L356
- How the key type is determined: https://github.com/zkemail/r0-zkEmail/blob/feat/regex/host/src/utils.rs#L85-L93
|
This is a comment for future PR. |
|
@SoraSuegami I understand your concern and yes I have thought of this. The reason you don't see a library implementation of the core code is because the current ZkVM implementations have different crates supported for precompiles and hence it becomes difficult to manage the dependencies properly. Once the ZkVM implementations become a bit stable, I'm planning to do an abstraction where we just write the rust and different ZkVM implementations are built automatically. Something like this - https://github.com/MatteoMer/any-zkvm |
This PR adds regex matching for email body and headers.