Move app-data-hash into app-data crate

crates/app-data/Cargo.toml

@@ -6,14 +6,15 @@ edition = "2021"
 license = "MIT OR Apache-2.0"
 
 [dependencies]
-app-data-hash = { path = "../app-data-hash" }
 bytes-hex = { path = "../bytes-hex" }
 anyhow = { workspace = true }
+tiny-keccak = { version = "2.0.2", features = ["keccak"] }
 serde = { workspace = true }
 serde_json = { workspace = true }
 serde_with = { workspace = true }
 primitive-types = { workspace = true }
 hex = { workspace = true }
+hex-literal = { workspace = true }
 
 [dev-dependencies]
 ethcontract = { workspace = true }

crates/app-data/src/app_data.rs

@@ -1,5 +1,5 @@
 use {
-    crate::{AppDataHash, Hooks},
+    crate::{app_data_hash::hash_full_app_data, AppDataHash, Hooks},
     anyhow::{anyhow, Context, Result},
     primitive_types::H160,
     serde::{de, Deserialize, Deserializer, Serialize, Serializer},
@@ -79,7 +79,7 @@ impl Validator {
             .unwrap_or_default();
 
         Ok(ValidatedAppData {
-            hash: AppDataHash(app_data_hash::hash_full_app_data(full_app_data)),
+            hash: AppDataHash(hash_full_app_data(full_app_data)),
             document,
             protocol,
         })

crates/app-data/src/app_data_hash.rs

@@ -1,3 +1,26 @@
+//! App data refers to extra information that is associated with orders. This
+//! information is not validated by the contract but it is used by other parts
+//! of the system. For example, a user could specify that they want their order
+//! to be COW only, which is something only the backend understands. Or what
+//! their intended slippage when creating the order with the frontend was, which
+//! adjusts the signed prices.
+//!
+//! On the smart contract level app data is freely choosable 32 bytes of signed
+//! order data. This isn't enough space for some purposes so we interpret those
+//! bytes as a hash of the full app data of arbitrary length. The full app data
+//! is thus signed by the user when they sign the order.
+//!
+//! This crate specifies how the hash is calculated. It takes the keccak256 hash
+//! of the input bytes. Additionally, it provides a canonical way to calculate
+//! an IPFS CID from the hash. This allows full app data to be uploaded to IPFS.
+//!
+//! Note that not all app data hashes were created this way. As of 2023-05-25 we
+//! are planning to move to the scheme implemented by this crate but orders have
+//! been created with arbitrary app data hashes until now. See [this issue][0]
+//! for more information.
+//!
+//! [0]: https://github.com/cowprotocol/services/issues/1465
+
 use {
     serde::{de, Deserializer, Serializer},
     serde_with::serde::{Deserialize, Serialize},
@@ -6,6 +29,7 @@ use {
         fmt::{self, Debug, Formatter},
         str::FromStr,
     },
+    tiny_keccak::{Hasher, Keccak},
 };
 
 /// A JSON object used to represent app data documents for uploading and
@@ -81,6 +105,29 @@ impl PartialEq<[u8; 32]> for AppDataHash {
     }
 }
 
+/// Hash full app data to get the bytes expected to be set as the contract level
+/// app data.
+pub fn hash_full_app_data(app_data: &[u8]) -> [u8; 32] {
+    let mut hasher = Keccak::v256();
+    hasher.update(app_data);
+    let mut hash = [0u8; 32];
+    hasher.finalize(&mut hash);
+    hash
+}
+
+/// Create an IPFS CIDv1 from a hash created by `hash_full_app_data`.
+///
+/// The return value is the raw bytes of the CID. It is not multibase encoded.
+pub fn create_ipfs_cid(app_data_hash: &[u8; 32]) -> [u8; 36] {
+    let mut cid = [0u8; 4 + 32];
+    cid[0] = 1; // cid version
+    cid[1] = 0x55; // raw codec
+    cid[2] = 0x1b; // keccak hash algorithm
+    cid[3] = 32; // keccak hash length
+    cid[4..].copy_from_slice(app_data_hash);
+    cid
+}
+
 #[cfg(test)]
 mod tests {
     use {super::*, serde_json::json};
@@ -125,4 +172,28 @@ mod tests {
         assert!(AppDataHash::deserialize(json!("asdf")).is_err());
         assert!(AppDataHash::deserialize(json!("0x00")).is_err());
     }
+
+    // Alternative way of calculating the expected values:
+    // cat appdata | ipfs block put --mhtype keccak-256
+    // -> bafkrwiek6tumtfzvo6yivqq5c7jtdkw6q3ar5pgfcjdujvrbzkbwl3eueq
+    // ipfs cid format -b base16
+    // bafkrwiek6tumtfzvo6yivqq5c7jtdkw6q3ar5pgfcjdujvrbzkbwl3eueq
+    // -> f01551b208af4e8c9973577b08ac21d17d331aade86c11ebcc5124744d621ca8365ec9424
+    // Remove the f prefix and you have the same CID.
+    // Or check out the cid explorer:
+    // - https://cid.ipfs.tech/#f01551b208af4e8c9973577b08ac21d17d331aade86c11ebcc5124744d621ca8365ec9424
+    // - https://cid.ipfs.tech/#bafkrwiek6tumtfzvo6yivqq5c7jtdkw6q3ar5pgfcjdujvrbzkbwl3eueq
+    #[test]
+    fn known_good() {
+        let full_app_data = r#"{"appCode":"CoW Swap","environment":"production","metadata":{"quote":{"slippageBips":"50","version":"0.2.0"},"orderClass":{"orderClass":"market","version":"0.1.0"}},"version":"0.6.0"}"#;
+        let expected_hash =
+            hex_literal::hex!("8af4e8c9973577b08ac21d17d331aade86c11ebcc5124744d621ca8365ec9424");
+        let expected_cid = hex_literal::hex!(
+            "01551b208af4e8c9973577b08ac21d17d331aade86c11ebcc5124744d621ca8365ec9424"
+        );
+        let hash = hash_full_app_data(full_app_data.as_bytes());
+        let cid = create_ipfs_cid(&hash);
+        assert_eq!(hash, expected_hash);
+        assert_eq!(cid, expected_cid);
+    }
 }

crates/e2e/Cargo.toml

@@ -44,7 +44,6 @@ web3 = { workspace = true, features = ["http"] }
 uuid = { version = "1.8.0", features = ["v4"] }
 
 [dev-dependencies]
-app-data-hash = { path = "../app-data-hash" }
 futures = { workspace = true }
 rand = { workspace = true }
 refunder = { path = "../refunder" }

crates/e2e/tests/e2e/app_data.rs

@@ -1,5 +1,5 @@
 use {
-    app_data::AppDataHash,
+    app_data::{hash_full_app_data, AppDataHash},
     e2e::{setup::*, tx},
     ethcontract::prelude::U256,
     model::{
@@ -74,7 +74,7 @@ async fn app_data(web3: Web3) {
 
     // hash matches
     let app_data = "{}";
-    let app_data_hash = AppDataHash(app_data_hash::hash_full_app_data(app_data.as_bytes()));
+    let app_data_hash = AppDataHash(hash_full_app_data(app_data.as_bytes()));
     let order1 = create_order(OrderCreationAppData::Both {
         full: app_data.to_string(),
         expected: app_data_hash,
@@ -133,7 +133,7 @@ async fn app_data(web3: Web3) {
 
     // pre-register some app-data with the API.
     let pre_app_data = r#"{"pre":"registered"}"#;
-    let pre_app_data_hash = AppDataHash(app_data_hash::hash_full_app_data(pre_app_data.as_bytes()));
+    let pre_app_data_hash = AppDataHash(hash_full_app_data(pre_app_data.as_bytes()));
     let err = services.get_app_data(pre_app_data_hash).await.unwrap_err();
     dbg!(err);
 
@@ -170,9 +170,7 @@ async fn app_data(web3: Web3) {
     // pre-registering invalid app-data fails.
     let err = services
         .put_app_data(
-            Some(AppDataHash(app_data_hash::hash_full_app_data(
-                invalid_app_data.as_bytes(),
-            ))),
+            Some(AppDataHash(hash_full_app_data(invalid_app_data.as_bytes()))),
             invalid_app_data,
         )
         .await

crates/model/Cargo.toml

@@ -11,7 +11,6 @@ doctest = false
 [dependencies]
 anyhow = { workspace = true }
 app-data = { path = "../app-data" }
-app-data-hash = { path = "../app-data-hash" }
 bytes-hex = { path = "../bytes-hex" }
 bigdecimal = { workspace = true }
 chrono = { workspace = true, features = ["serde", "clock"] }

crates/model/src/order.rs

@@ -10,7 +10,7 @@ use {
         TokenPair,
     },
     anyhow::{anyhow, Result},
-    app_data::AppDataHash,
+    app_data::{hash_full_app_data, AppDataHash},
     chrono::{offset::Utc, DateTime},
     derivative::Derivative,
     hex_literal::hex,
@@ -465,7 +465,7 @@ impl OrderCreationAppData {
         match self {
             Self::Hash { hash } => *hash,
             Self::Full { full } | Self::Both { full, .. } => {
-                AppDataHash(app_data_hash::hash_full_app_data(full.as_bytes()))
+                AppDataHash(hash_full_app_data(full.as_bytes()))
             }
         }
     }

crates/orderbook/Cargo.toml

@@ -18,7 +18,6 @@ path = "src/main.rs"
 [dependencies]
 anyhow = { workspace = true }
 app-data = { path = "../app-data" }
-app-data-hash = { path = "../app-data-hash" }
 async-trait = { workspace = true }
 bigdecimal = { workspace = true }
 cached = { workspace = true }

crates/orderbook/src/ipfs_app_data.rs

@@ -1,7 +1,7 @@
 use {
     crate::ipfs::Ipfs,
     anyhow::Result,
-    app_data::AppDataHash,
+    app_data::{create_ipfs_cid, AppDataHash},
     cached::{Cached, TimedSizedCache},
     std::sync::Mutex,
 };
@@ -121,7 +121,7 @@ impl IpfsAppData {
 }
 
 fn new_app_data_cid(contract_app_data: &AppDataHash) -> String {
-    let raw_cid = app_data_hash::create_ipfs_cid(&contract_app_data.0);
+    let raw_cid = create_ipfs_cid(&contract_app_data.0);
     multibase::encode(multibase::Base::Base32Lower, raw_cid)
 }
 

crates/shared/Cargo.toml

@@ -11,7 +11,6 @@ doctest = false
 [dependencies]
 anyhow = { workspace = true }
 app-data = { path = "../app-data" }
-app-data-hash = { path = "../app-data-hash" }
 bytes-hex = { path = "../bytes-hex" }
 async-trait = { workspace = true }
 bigdecimal = { workspace = true }