perf: switch md5 to xxhash (#262)

Closes #259.

Author: DaniPopes

Cargo.toml

@@ -46,7 +46,6 @@ alloy-json-abi = { version = "0.8", features = ["serde_json"] }
 alloy-primitives = { version = "0.8", features = ["serde", "rand"] }
 cfg-if = "1.0"
 dunce = "1.0"
-md-5 = "0.10"
 memmap2 = "0.9"
 path-slash = "0.2"
 rayon = "1.8"

crates/artifacts/solc/Cargo.toml

@@ -31,9 +31,6 @@ yansi.workspace = true
 tokio = { workspace = true, optional = true, features = ["fs"] }
 futures-util = { workspace = true, optional = true }
 
-# checksum
-md-5 = { workspace = true, optional = true }
-
 # walkdir
 walkdir = { workspace = true, optional = true }
 
@@ -50,6 +47,6 @@ foundry-compilers-core = { workspace = true, features = ["test-utils"] }
 
 [features]
 async = ["dep:tokio", "dep:futures-util"]
-checksum = ["dep:md-5"]
+checksum = ["foundry-compilers-core/hasher"]
 walkdir = ["dep:walkdir", "foundry-compilers-core/walkdir"]
 rayon = ["dep:rayon"]

crates/artifacts/solc/src/sources.rs

@@ -204,7 +204,7 @@ impl Source {
     /// Generate a non-cryptographically secure checksum of the given source.
     #[cfg(feature = "checksum")]
     pub fn content_hash_of(src: &str) -> String {
-        alloy_primitives::hex::encode(<md5::Md5 as md5::Digest>::digest(src))
+        foundry_compilers_core::utils::unique_hash(src)
     }
 }
 

crates/compilers/Cargo.toml

@@ -20,15 +20,14 @@ foundry-compilers-artifacts = { workspace = true, features = [
     "walkdir",
     "rayon",
 ] }
-foundry-compilers-core = { workspace = true, features = ["regex"] }
+foundry-compilers-core = { workspace = true, features = ["hasher", "regex"] }
 serde.workspace = true
 semver.workspace = true
 alloy-primitives.workspace = true
 serde_json.workspace = true
 tracing.workspace = true
 alloy-json-abi.workspace = true
 rayon.workspace = true
-md-5.workspace = true
 thiserror.workspace = true
 path-slash.workspace = true
 yansi.workspace = true

crates/compilers/src/buildinfo.rs

@@ -3,9 +3,7 @@
 use crate::compilers::{
     CompilationError, CompilerContract, CompilerInput, CompilerOutput, Language,
 };
-use alloy_primitives::hex;
 use foundry_compilers_core::{error::Result, utils};
-use md5::Digest;
 use semver::Version;
 use serde::{de::DeserializeOwned, Deserialize, Serialize};
 use std::{
@@ -97,22 +95,13 @@ impl<L: Language> RawBuildInfo<L> {
         let version = input.version().clone();
         let build_context = BuildContext::new(input, output)?;
 
-        let mut hasher = md5::Md5::new();
-
-        hasher.update(ETHERS_FORMAT_VERSION);
-
         let solc_short = format!("{}.{}.{}", version.major, version.minor, version.patch);
-        hasher.update(&solc_short);
-        hasher.update(version.to_string());
-
         let input = serde_json::to_value(input)?;
-        hasher.update(&serde_json::to_string(&input)?);
-
-        // create the hash for `{_format,solcVersion,solcLongVersion,input}`
-        // N.B. this is not exactly the same as hashing the json representation of these values but
-        // the must efficient one
-        let result = hasher.finalize();
-        let id = hex::encode(result);
+        let id = utils::unique_hash_many([
+            ETHERS_FORMAT_VERSION,
+            &version.to_string(),
+            &serde_json::to_string(&input)?,
+        ]);
 
         let mut build_info = BTreeMap::new();
 

crates/compilers/src/compile/output/mod.rs

@@ -607,8 +607,7 @@ impl<C: Compiler> AggregatedCompilerOutput<C> {
     ///
     /// There can be multiple `BuildInfo`, since we support multiple versions.
     ///
-    /// The created files have the md5 hash `{_format,solcVersion,solcLongVersion,input}` as their
-    /// file name
+    /// The created files have a unique identifier as their name.
     pub fn write_build_infos(&self, build_info_dir: &Path) -> Result<(), SolcError> {
         if self.build_infos.is_empty() {
             return Ok(());

crates/core/Cargo.toml

@@ -23,6 +23,11 @@ serde_json.workspace = true
 serde.workspace = true
 thiserror.workspace = true
 
+# hasher
+xxhash-rust = { version = "0.8", optional = true, default-features = false, features = [
+    "xxh3",
+] }
+
 # regex
 regex = { workspace = true, optional = true }
 
@@ -46,6 +51,7 @@ tempfile.workspace = true
 
 [features]
 async = ["dep:tokio"]
+hasher = ["dep:xxhash-rust"]
 project-util = ["dep:tempfile", "dep:fs_extra"]
 regex = ["dep:regex"]
 svm-solc = ["dep:svm", "dep:tokio"]

crates/core/src/utils/mod.rs

@@ -78,6 +78,32 @@ pub static SUPPORTS_BASE_PATH: Lazy<VersionReq> =
 pub static SUPPORTS_INCLUDE_PATH: Lazy<VersionReq> =
     Lazy::new(|| VersionReq::parse(">=0.8.8").unwrap());
 
+/// A non-cryptographic hash function for creating unique identifiers.
+///
+/// The exact algorithm being used shouldn't matter.
+// See Hardhat: https://github.com/NomicFoundation/hardhat/blob/e9ab5332a5505a6d1fe9bfbc687f5f46bdff6dd7/packages/hardhat-core/src/internal/util/hash.ts#L1-L16
+#[cfg(feature = "hasher")]
+pub fn unique_hash(input: impl AsRef<[u8]>) -> String {
+    encode_hash(xxhash_rust::xxh3::xxh3_64(input.as_ref()))
+}
+
+/// A non-cryptographic hash function for creating unique identifiers.
+///
+/// See [`unique_hash`] for more details.
+#[cfg(feature = "hasher")]
+pub fn unique_hash_many(inputs: impl IntoIterator<Item = impl AsRef<[u8]>>) -> String {
+    let mut hasher = xxhash_rust::xxh3::Xxh3Default::new();
+    for input in inputs {
+        hasher.update(input.as_ref());
+    }
+    encode_hash(hasher.digest())
+}
+
+#[cfg(feature = "hasher")]
+fn encode_hash(x: u64) -> String {
+    hex::encode(x.to_be_bytes())
+}
+
 /// Move a range by a specified offset
 pub fn range_by_offset(range: &Range<usize>, offset: isize) -> Range<usize> {
     Range {