Skip untracked files (#76)

* presence of .codeowners-skip-untracked-files skips untracked files

* config file as yml

* only validate tracked files using tracked files map

* updating tests

Author: perryqh

Cargo.lock

@@ -1,6 +1,6 @@
 [package]
 name = "codeowners"
-version = "0.2.11"
+version = "0.2.13"
 edition = "2024"
 
 [profile.release]
@@ -34,6 +34,6 @@ tracing-subscriber = { version = "0.3.18", features = ["env-filter"] }
 [dev-dependencies]
 assert_cmd = "2.0.16"
 rusty-hook = "^0.11.2"
-predicates = "3.1.2"
+predicates = "3.1.3"
 pretty_assertions = "1.4.1" # Shows a more readable diff when comparing objects
 indoc = "2.0.5"

Cargo.toml

@@ -24,9 +24,6 @@ pub struct Config {
 
     #[serde(default = "default_ignore_dirs")]
     pub ignore_dirs: Vec<String>,
-
-    #[serde(default = "default_skip_untracked_files")]
-    pub skip_untracked_files: bool,
 }
 
 #[allow(dead_code)]
@@ -63,10 +60,6 @@ fn vendored_gems_path() -> String {
     "vendored/".to_string()
 }
 
-fn default_skip_untracked_files() -> bool {
-    true
-}
-
 fn default_ignore_dirs() -> Vec<String> {
     vec![
         ".cursor".to_owned(),

src/config.rs

@@ -2,9 +2,9 @@ pub mod cache;
 pub(crate) mod common_test;
 pub mod config;
 pub mod crosscheck;
-pub(crate) mod files;
 pub mod ownership;
 pub(crate) mod project;
 pub mod project_builder;
 pub mod project_file_builder;
 pub mod runner;
+pub(crate) mod tracked_files;

src/files.rs

@@ -303,7 +303,6 @@ mod tests {
             vendored_gems_path: vendored_path.to_string(),
             cache_directory: "tmp/cache/codeowners".to_string(),
             ignore_dirs: vec![],
-            skip_untracked_files: false,
         }
     }
 

src/lib.rs

@@ -13,9 +13,9 @@ use tracing::{instrument, warn};
 use crate::{
     cache::Cache,
     config::Config,
-    files,
     project::{DirectoryCodeownersFile, Error, Package, PackageType, Project, ProjectFile, Team, VendoredGem, deserializers},
     project_file_builder::ProjectFileBuilder,
+    tracked_files,
 };
 
 type AbsolutePath = PathBuf;
@@ -61,16 +61,16 @@ impl<'a> ProjectBuilder<'a> {
         // Prune traversal early: skip heavy and irrelevant directories
         let ignore_dirs = self.config.ignore_dirs.clone();
         let base_path = self.base_path.clone();
-        let untracked_files = if self.config.skip_untracked_files {
-            files::untracked_files(&base_path).unwrap_or_default()
-        } else {
-            vec![]
-        };
+        let tracked_files = tracked_files::find_tracked_files(&self.base_path);
 
         builder.filter_entry(move |entry: &DirEntry| {
             let path = entry.path();
             let file_name = entry.file_name().to_str().unwrap_or("");
-            if !untracked_files.is_empty() && untracked_files.contains(&path.to_path_buf()) {
+            if let Some(tracked_files) = &tracked_files
+                && let Some(ft) = entry.file_type()
+                && ft.is_file()
+                && !tracked_files.contains_key(path)
+            {
                 return false;
             }
             if let Some(ft) = entry.file_type()

src/ownership/for_file_fast.rs

@@ -0,0 +1,58 @@
+use std::{
+    collections::HashMap,
+    path::{Path, PathBuf},
+    process::Command,
+};
+
+pub(crate) fn find_tracked_files(base_path: &Path) -> Option<HashMap<PathBuf, bool>> {
+    let output = Command::new("git")
+        .args(["ls-files", "--full-name", "-z", "--", "."])
+        .current_dir(base_path)
+        .output()
+        .ok()?;
+
+    if !output.status.success() {
+        return None;
+    }
+
+    let results: HashMap<PathBuf, bool> = output
+        .stdout
+        .split(|&b| b == b'\0')
+        .filter(|chunk| !chunk.is_empty())
+        .map(|rel| std::str::from_utf8(rel).ok().map(|s| (base_path.join(s), true)))
+        .collect::<Option<HashMap<PathBuf, bool>>>()?;
+
+    Some(results)
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_untracked_files() {
+        let tmp_dir = tempfile::tempdir().unwrap();
+        assert!(find_tracked_files(tmp_dir.path()).is_none());
+
+        std::process::Command::new("git")
+            .arg("init")
+            .current_dir(tmp_dir.path())
+            .output()
+            .expect("failed to run git init");
+
+        std::fs::write(tmp_dir.path().join("test.txt"), "test").unwrap();
+        let tracked = find_tracked_files(tmp_dir.path()).unwrap();
+        assert!(tracked.is_empty());
+
+        std::process::Command::new("git")
+            .arg("add")
+            .arg("test.txt")
+            .current_dir(tmp_dir.path())
+            .output()
+            .expect("failed to add test.txt");
+
+        let tracked = find_tracked_files(tmp_dir.path()).unwrap();
+        assert!(tracked.len() == 1);
+        assert!(tracked.get(&tmp_dir.path().join("test.txt")).unwrap());
+    }
+}