feat: cache multiple repos

Signed-off-by: Aisha M <[email protected]>

plugins/activity/local-plugin.kdl

@@ -11,5 +11,5 @@ entrypoint {
 }
 
 dependencies {
-  plugin "mitre/git" version="0.2.0" manifest="./plugins/git/local-plugin.kdl"
+  plugin "mitre/git" version="0.3.0" manifest="./plugins/git/local-plugin.kdl"
 }

plugins/activity/plugin.kdl

@@ -11,5 +11,5 @@ entrypoint {
 }
 
 dependencies {
-  plugin "mitre/git" version="0.2.0" manifest="https://hipcheck.mitre.org/dl/plugin/mitre/git.kdl"
+  plugin "mitre/git" version="0.3.0" manifest="https://hipcheck.mitre.org/dl/plugin/mitre/git.kdl"
 }

plugins/affiliation/local-plugin.kdl

@@ -11,5 +11,5 @@ entrypoint {
 }
 
 dependencies {
-  plugin "mitre/git" version="0.2.0" manifest="./plugins/git/local-plugin.kdl"
+  plugin "mitre/git" version="0.3.0" manifest="./plugins/git/local-plugin.kdl"
 }

plugins/affiliation/plugin.kdl

@@ -11,5 +11,5 @@ entrypoint {
 }
 
 dependencies {
-  plugin "mitre/git" version="0.2.0" manifest="https://hipcheck.mitre.org/dl/plugin/mitre/git.kdl"
+  plugin "mitre/git" version="0.3.0" manifest="https://hipcheck.mitre.org/dl/plugin/mitre/git.kdl"
 }

plugins/churn/local-plugin.kdl

@@ -11,5 +11,5 @@ entrypoint {
 }
 
 dependencies {
-  plugin "mitre/git" version="0.2.0" manifest="./plugins/git/local-plugin.kdl"
+  plugin "mitre/git" version="0.3.0" manifest="./plugins/git/local-plugin.kdl"
 }

plugins/churn/plugin.kdl

@@ -11,5 +11,5 @@ entrypoint {
 }
 
 dependencies {
-  plugin "mitre/git" version="0.2.0" manifest="https://hipcheck.mitre.org/dl/plugin/mitre/git.kdl"
+  plugin "mitre/git" version="0.3.0" manifest="https://hipcheck.mitre.org/dl/plugin/mitre/git.kdl"
 }

plugins/entropy/local-plugin.kdl

@@ -11,5 +11,5 @@ entrypoint {
 }
 
 dependencies {
-  plugin "mitre/git" version="0.2.0" manifest="./plugins/git/local-plugin.kdl"
+  plugin "mitre/git" version="0.3.0" manifest="./plugins/git/local-plugin.kdl"
 }

plugins/entropy/plugin.kdl

@@ -11,5 +11,5 @@ entrypoint {
 }
 
 dependencies {
-  plugin "mitre/git" version="0.2.0" manifest="https://hipcheck.mitre.org/dl/plugin/mitre/git.kdl"
+  plugin "mitre/git" version="0.3.0" manifest="https://hipcheck.mitre.org/dl/plugin/mitre/git.kdl"
 }

plugins/git/Cargo.toml

@@ -12,6 +12,7 @@ clap = { version = "4.5.23", features = ["derive"] }
 hipcheck-sdk = { version = "0.3.0", path = "../../sdk/rust", features = [
     "macros",
 ] }
+lru = "0.12.5"
 gix = { version = "0.69.1", default-features = false, features = [
     "basic",
     "max-control",
@@ -22,6 +23,7 @@ log = "0.4.22"
 once_cell = "1.10.0"
 schemars = { version = "0.8.21", features = ["url"] }
 semver = "1.0.24"
+serde_json = "1.0.134"
 serde = { version = "1.0.215", features = ["derive", "rc"] }
 tokio = { version = "1.42.0", features = ["rt"] }
 which = { version = "7.0.1", default-features = false }

plugins/git/src/git.rs

@@ -2,6 +2,8 @@
 
 use crate::data::*;
 
+use crate::Error;
+use crate::CACHE;
 use anyhow::Context;
 use anyhow::Result;
 use gix::bstr::ByteSlice;
@@ -18,13 +20,12 @@ use gix::traverse::commit::simple::CommitTimeOrder;
 use gix::ObjectId;
 use gix::Repository;
 use jiff::Timestamp;
+use lru::LruCache;
 use std::path::Path;
 use std::path::PathBuf;
-use std::sync::Mutex;
 
 /// used to cache all of the `RawCommit` from the last repo/HEAD combination analyzed by this
-/// plugin
-type GitRawCommitCache = Option<(PathBuf, ObjectId, Vec<RawCommit>)>;
+pub type GitRawCommitCache = LruCache<(PathBuf, ObjectId), Vec<RawCommit>>;
 
 /// retrieve a handle to the git repo at this path, as well as determine the commit hash of
 /// HEAD
@@ -122,24 +123,25 @@ pub fn get_all_raw_commits<P>(repo_path: P) -> Result<Vec<RawCommit>>
 where
 	P: AsRef<Path>,
 {
-	// used to cache all of the RawCommits from the last repository analyzed
-	static ALL_RAW_COMMITS: Mutex<GitRawCommitCache> = Mutex::new(None);
-
 	let (repo, head_commit) = initialize_repo(repo_path.as_ref())?;
-	let mut cache = ALL_RAW_COMMITS.lock().unwrap();
+	let unlocked_cache = CACHE.get().ok_or(Error::UnspecifiedQueryState)?;
+	let mut cache = unlocked_cache
+		.lock()
+		.map_err(|_| Error::UnspecifiedQueryState)?;
 
 	// if there is a value in cache, and it is the same repo with the same HEAD commit, then we can use the
 	// cached value
-	if let Some(cached_value) = cache.as_ref() {
-		if cached_value.0 == repo_path.as_ref().to_path_buf() && cached_value.1 == head_commit {
-			return Ok(cached_value.2.clone());
-		}
+	if let Some(cached_value) = cache.get_mut(&(repo_path.as_ref().to_path_buf(), head_commit)) {
+		return Ok(cached_value.clone());
 	}
 
 	// otherwise the cache needs to be updated with the data from this repo_path/HEAD combination
 	let updated_value = get_all_raw_commits_inner(&repo, repo_path.as_ref(), head_commit)?;
 	let raw_commits = updated_value.2.clone();
-	*cache = Some(get_all_raw_commits_inner(&repo, repo_path, head_commit)?);
+	cache.put(
+		(repo_path.as_ref().to_path_buf(), head_commit),
+		raw_commits.clone(),
+	);
 	Ok(raw_commits)
 }
 

plugins/git/src/main.rs

@@ -12,14 +12,21 @@ use crate::{
 	},
 	git::{
 		get_all_raw_commits, get_commit_diffs, get_commits_from_date, get_contributors, get_diffs,
-		get_latest_commit,
+		get_latest_commit, GitRawCommitCache,
 	},
 };
 use clap::Parser;
 use hipcheck_sdk::{prelude::*, types::LocalGitRepo};
 use jiff::Timestamp;
+use lru::LruCache;
 use schemars::JsonSchema;
 use serde::{Deserialize, Serialize};
+use std::{
+	num::NonZero,
+	sync::{Mutex, OnceLock},
+};
+
+pub static CACHE: OnceLock<Mutex<GitRawCommitCache>> = OnceLock::new();
 
 /// A locally stored git repo, with a list of additional details
 /// The details will vary based on the query (e.g. a date, a committer e-mail address, a commit hash)
@@ -428,16 +435,37 @@ async fn commit_contributors(
 	Ok(commit_contributors)
 }
 
-#[derive(Clone, Debug)]
+#[derive(Deserialize)]
+struct Config {
+	#[serde(default = "default_commit_cache_size")]
+	commit_cache_size: usize,
+}
+
+fn default_commit_cache_size() -> usize {
+	1
+}
+
+#[derive(Clone, Debug, Default)]
 struct GitPlugin;
 
 impl Plugin for GitPlugin {
 	const PUBLISHER: &'static str = "mitre";
 
 	const NAME: &'static str = "git";
 
-	fn set_config(&self, _config: Value) -> std::result::Result<(), ConfigError> {
-		Ok(())
+	fn set_config(&self, config: Value) -> std::result::Result<(), ConfigError> {
+		// Deserialize and validate the config struct
+		let conf: Config =
+			serde_json::from_value::<Config>(config).map_err(|e| ConfigError::Unspecified {
+				message: e.to_string(),
+			})?;
+		let cache_size = conf.commit_cache_size;
+
+		CACHE
+			.set(Mutex::new(LruCache::new(NonZero::new(cache_size).unwrap())))
+			.map_err(|_e| ConfigError::Unspecified {
+				message: "config was already set".to_owned(),
+			})
 	}
 
 	fn default_policy_expr(&self) -> hipcheck_sdk::prelude::Result<String> {

plugins/identity/local-plugin.kdl

@@ -11,5 +11,5 @@ entrypoint {
 }
 
 dependencies {
-    plugin "mitre/git" version="0.2.0" manifest="./plugins/git/local-plugin.kdl"
+    plugin "mitre/git" version="0.3.0" manifest="./plugins/git/local-plugin.kdl"
 }

plugins/identity/plugin.kdl

@@ -11,5 +11,5 @@ entrypoint {
 }
 
 dependencies {
-    plugin "mitre/git" version="0.2.0" manifest="https://hipcheck.mitre.org/dl/plugin/mitre/git.kdl"
+    plugin "mitre/git" version="0.3.0" manifest="https://hipcheck.mitre.org/dl/plugin/mitre/git.kdl"
 }

site/content/docs/guide/plugins/mitre-git.md

@@ -8,3 +8,9 @@ extra:
 
 Provides access to Git commit history data. Does not define a default query
 and can't be used as a top-level plugin in a policy file.
+
+## Configuration
+
+| Parameter           | Type    | Explanation   |
+|:--------------------|:--------|:--------------|
+| `commit-cache-size` | `Integer` | Optional number of repositories to retain in cache. Defaults to one. |