Scope Git watching and status to the sync directory (#503)

crates/yaak-git/Cargo.toml

@@ -18,3 +18,7 @@ url = "2"
 yaak-common = { workspace = true }
 yaak-models = { workspace = true }
 yaak-sync = { workspace = true }
+
+[dev-dependencies]
+tempfile = "3"
+tokio = { workspace = true, features = ["macros", "rt"] }

crates/yaak-git/bindings/gen_git.ts

@@ -19,7 +19,12 @@ export type GitStatus = "untracked" | "conflict" | "current" | "modified" | "rem
 
 export type GitStatusEntry = { relaPath: string, status: GitStatus, staged: boolean, prev: SyncModel | null, next: SyncModel | null, };
 
-export type GitStatusSummary = { path: string, headRef: string | null, headRefShorthand: string | null, entries: Array<GitStatusEntry>, origins: Array<string>, localBranches: Array<string>, remoteBranches: Array<string>, ahead: number, behind: number, };
+export type GitStatusSummary = { path: string, 
+/**
+ * The status directory relative to the repo root ("" when it IS the root).
+ * Useful for displaying entry paths relative to the sync directory
+ */
+relaDir: string, headRef: string | null, headRefShorthand: string | null, entries: Array<GitStatusEntry>, origins: Array<string>, localBranches: Array<string>, remoteBranches: Array<string>, ahead: number, behind: number, };
 
 export type GitWorktreeStatus = { entries: Array<GitWorktreeStatusEntry>, };
 

crates/yaak-git/src/commit.rs

@@ -1,11 +1,39 @@
 use crate::binary::new_binary_command;
 use crate::error::Error::GenericError;
+use crate::repository::open_repo;
+use crate::status::repo_relative_dir;
 use log::info;
 use std::path::Path;
 
+/// Commit the staged changes within `dir` (their current worktree content,
+/// matching what the commit dialog displays). Scoping the commit to the sync
+/// directory means files staged outside of it (e.g. elsewhere in a containing
+/// monorepo) are never swept into a Yaak commit — they stay staged for the
+/// user's own next commit.
 pub async fn git_commit(dir: &Path, message: &str) -> crate::error::Result<()> {
-    let out =
-        new_binary_command(dir).await?.args(["commit", "--message", message]).output().await?;
+    // Run git from the repo root: command-line pathspecs resolve relative to
+    // the working directory, and staged paths are repo-root-relative
+    let (workdir, rela_dir) = {
+        let repo = open_repo(dir)?;
+        let workdir = repo
+            .workdir()
+            .ok_or_else(|| GenericError("Repository has no worktree".to_string()))?
+            .to_path_buf();
+        (workdir, repo_relative_dir(&repo, dir))
+    };
+
+    let staged = staged_files(&workdir, rela_dir.as_deref()).await?;
+    if staged.is_empty() {
+        return Err(GenericError("No staged changes to commit".to_string()));
+    }
+
+    let mut cmd = new_binary_command(&workdir).await?;
+    // --literal-pathspecs: the staged paths are exact files, never patterns
+    cmd.arg("--literal-pathspecs");
+    cmd.args(["commit", "--message", message, "--"]);
+    cmd.args(staged);
+
+    let out = cmd.output().await?;
 
     let stdout = String::from_utf8_lossy(&out.stdout);
     let stderr = String::from_utf8_lossy(&out.stderr);
@@ -19,3 +47,60 @@ pub async fn git_commit(dir: &Path, message: &str) -> crate::error::Result<()> {
 
     Ok(())
 }
+
+/// Repo-relative paths of staged changes, limited to `rela_dir` when given.
+/// Must be run from the repo root so the pathspec resolves correctly. Uses
+/// -z for NUL separation so paths with special characters come through
+/// unquoted, --no-renames so staged renames list both sides, and
+/// --literal-pathspecs so a scope dir with glob characters isn't a pattern
+async fn staged_files(workdir: &Path, rela_dir: Option<&str>) -> crate::error::Result<Vec<String>> {
+    let mut cmd = new_binary_command(workdir).await?;
+    cmd.arg("--literal-pathspecs");
+    cmd.args(["diff", "--cached", "--name-only", "--no-renames", "-z"]);
+    if let Some(rela_dir) = rela_dir {
+        cmd.args(["--", rela_dir]);
+    }
+    let out = cmd.output().await?;
+
+    if !out.status.success() {
+        let stderr = String::from_utf8_lossy(&out.stderr);
+        return Err(GenericError(format!("Failed to list staged files: {}", stderr)));
+    }
+
+    Ok(String::from_utf8_lossy(&out.stdout)
+        .split('\0')
+        .filter(|p| !p.is_empty())
+        .map(String::from)
+        .collect())
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[tokio::test]
+    async fn test_staged_files_scoped() {
+        let tmp = tempfile::tempdir().unwrap();
+        let repo = git2::Repository::init(tmp.path()).unwrap();
+
+        // Scope dir name doubles as a glob pattern; sync1 is its glob decoy
+        let sync_dir = tmp.path().join("sync[1]");
+        std::fs::create_dir(&sync_dir).unwrap();
+        std::fs::write(sync_dir.join("yaak.req_1.yaml"), "inside").unwrap();
+        std::fs::create_dir(tmp.path().join("sync1")).unwrap();
+        std::fs::write(tmp.path().join("sync1").join("decoy.txt"), "decoy").unwrap();
+        std::fs::write(tmp.path().join("outside.txt"), "outside").unwrap();
+
+        let mut index = repo.index().unwrap();
+        index.add_path(Path::new("sync[1]/yaak.req_1.yaml")).unwrap();
+        index.add_path(Path::new("sync1/decoy.txt")).unwrap();
+        index.add_path(Path::new("outside.txt")).unwrap();
+        index.write().unwrap();
+
+        let scoped = staged_files(tmp.path(), Some("sync[1]")).await.unwrap();
+        assert_eq!(scoped, vec!["sync[1]/yaak.req_1.yaml"]);
+
+        let all = staged_files(tmp.path(), None).await.unwrap();
+        assert_eq!(all.len(), 3);
+    }
+}

crates/yaak-git/src/pull.rs

@@ -21,8 +21,11 @@ pub enum PullResult {
 
 fn has_uncommitted_changes(dir: &Path) -> Result<bool> {
     let repo = open_repo(dir)?;
-    let mut opts = git2::StatusOptions::new();
-    opts.include_ignored(false).include_untracked(false);
+    // Scoped to the sync directory: uncommitted changes elsewhere in a
+    // containing monorepo shouldn't block pulling. Git itself still refuses
+    // a merge that would clobber uncommitted files outside the sync dir
+    let mut opts = crate::status::scoped_status_options(&repo, dir);
+    opts.include_untracked(false);
     let statuses = repo.statuses(Some(&mut opts))?;
     Ok(statuses.iter().any(|e| e.status() != git2::Status::CURRENT))
 }

crates/yaak-git/src/repository.rs

@@ -6,6 +6,7 @@ use std::path::{Path, PathBuf};
 pub struct GitRepositoryPaths {
     pub workdir: PathBuf,
     pub gitdir: PathBuf,
+    pub commondir: PathBuf,
 }
 
 pub(crate) fn open_repo(dir: &Path) -> crate::error::Result<git2::Repository> {
@@ -22,7 +23,13 @@ pub fn git_repository_paths(dir: &Path) -> Result<GitRepositoryPaths> {
         .workdir()
         .ok_or_else(|| Error::GenericError("Git repository does not have a worktree".into()))?
         .to_path_buf();
-    Ok(GitRepositoryPaths { workdir, gitdir: repo.path().to_path_buf() })
+    Ok(GitRepositoryPaths {
+        workdir,
+        gitdir: repo.path().to_path_buf(),
+        // Same as gitdir except for linked worktrees, where shared refs and
+        // packed-refs live in the main repository's git directory
+        commondir: repo.commondir().to_path_buf(),
+    })
 }
 
 pub fn git_path_is_ignored(dir: &Path, rela_path: &Path) -> Result<bool> {

crates/yaak-git/src/status.rs

@@ -12,6 +12,9 @@ use yaak_sync::models::SyncModel;
 #[ts(export, export_to = "gen_git.ts")]
 pub struct GitStatusSummary {
     pub path: String,
+    /// The status directory relative to the repo root ("" when it IS the root).
+    /// Useful for displaying entry paths relative to the sync directory
+    pub rela_dir: String,
     pub head_ref: Option<String>,
     pub head_ref_shorthand: Option<String>,
     pub entries: Vec<GitStatusEntry>,
@@ -79,11 +82,8 @@ pub enum GitStatus {
 
 pub fn git_worktree_status(dir: &Path) -> crate::error::Result<GitWorktreeStatus> {
     let repo = open_repo(dir)?;
-    let mut opts = git2::StatusOptions::new();
-    opts.include_ignored(false)
-        .include_untracked(true)
-        .recurse_untracked_dirs(true)
-        .include_unmodified(false);
+    let mut opts = scoped_status_options(&repo, dir);
+    opts.include_unmodified(false);
 
     let mut entries = Vec::new();
     for entry in repo.statuses(Some(&mut opts))?.into_iter() {
@@ -115,11 +115,8 @@ pub fn git_status(dir: &Path) -> crate::error::Result<GitStatusSummary> {
     let branch_info = git_branch_info_for_repo(&repo, dir)?;
     let head_tree = repo.head().ok().and_then(|head| head.peel_to_tree().ok());
 
-    let mut opts = git2::StatusOptions::new();
-    opts.include_ignored(false)
-        .include_untracked(true) // Include untracked
-        .recurse_untracked_dirs(true) // Show all untracked
-        .include_unmodified(true); // Include unchanged
+    let mut opts = scoped_status_options(&repo, dir);
+    opts.include_unmodified(true); // Include unchanged
 
     // TODO: Support renames
 
@@ -160,6 +157,7 @@ pub fn git_status(dir: &Path) -> crate::error::Result<GitStatusSummary> {
 
     Ok(GitStatusSummary {
         entries,
+        rela_dir: repo_relative_dir(&repo, dir).unwrap_or_default(),
         path: branch_info.path,
         head_ref: branch_info.head_ref,
         head_ref_shorthand: branch_info.head_ref_shorthand,
@@ -266,6 +264,40 @@ fn git_status_from_raw(status: git2::Status) -> Option<(GitStatus, bool)> {
     Some((status, staged))
 }
 
+/// Construct StatusOptions for a walk scoped to `dir`. Yaak only cares about
+/// the sync directory, and a full walk is expensive when the containing repo
+/// is large (e.g. a sync dir inside a monorepo); scoping is a no-op when
+/// `dir` is the repo root. Always build status walks through this so a new
+/// call site can't forget the scoping.
+pub(crate) fn scoped_status_options(repo: &git2::Repository, dir: &Path) -> git2::StatusOptions {
+    let mut opts = git2::StatusOptions::new();
+    opts.include_ignored(false).include_untracked(true).recurse_untracked_dirs(true);
+    if let Some(rela) = repo_relative_dir(repo, dir) {
+        opts.pathspec(rela);
+        // Match the path literally (exact or directory prefix) instead of as
+        // a glob — directory names can contain pattern characters like [ or *
+        opts.disable_pathspec_match(true);
+    }
+    opts
+}
+
+/// The path of `dir` relative to the repo root as a forward-slash string
+/// (Git pathspecs use forward slashes even on Windows), or None when `dir`
+/// is the root itself (or outside the repo). Both sides are canonicalized so
+/// symlinked paths compare consistently.
+pub(crate) fn repo_relative_dir(repo: &git2::Repository, dir: &Path) -> Option<String> {
+    let workdir = repo.workdir()?;
+    let workdir = workdir.canonicalize().unwrap_or_else(|_| workdir.to_path_buf());
+    let canonical_dir = dir.canonicalize().unwrap_or_else(|_| dir.to_path_buf());
+    let rela = canonical_dir.strip_prefix(&workdir).ok()?;
+    if rela.as_os_str().is_empty() {
+        return None;
+    }
+    let parts: Vec<String> =
+        rela.components().map(|c| c.as_os_str().to_string_lossy().into_owned()).collect();
+    Some(parts.join("/"))
+}
+
 fn model_id_from_rela_path(path: &Path) -> Option<String> {
     let ext = path.extension()?.to_str()?;
     if ext != "yaml" && ext != "yml" && ext != "json" {
@@ -274,3 +306,70 @@ fn model_id_from_rela_path(path: &Path) -> Option<String> {
 
     path.file_stem()?.to_str()?.strip_prefix("yaak.").map(String::from)
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_worktree_status_scoped_to_subdir() {
+        let tmp = tempfile::tempdir().unwrap();
+        git2::Repository::init(tmp.path()).unwrap();
+
+        let sync_dir = tmp.path().join("sync");
+        std::fs::create_dir(&sync_dir).unwrap();
+        std::fs::write(sync_dir.join("yaak.req_1.yaml"), "inside").unwrap();
+        std::fs::write(tmp.path().join("outside.txt"), "outside").unwrap();
+
+        // Status on a subdirectory only reports that subdirectory
+        let status = git_worktree_status(&sync_dir).unwrap();
+        let paths: Vec<&str> = status.entries.iter().map(|e| e.rela_path.as_str()).collect();
+        assert_eq!(paths, vec!["sync/yaak.req_1.yaml"]);
+        assert_eq!(status.entries[0].model_id.as_deref(), Some("req_1"));
+
+        // Status on the repo root reports everything
+        let status = git_worktree_status(tmp.path()).unwrap();
+        assert_eq!(status.entries.len(), 2);
+    }
+
+    #[test]
+    fn test_worktree_status_scoped_literal_dir_name() {
+        let tmp = tempfile::tempdir().unwrap();
+        git2::Repository::init(tmp.path()).unwrap();
+
+        // A directory name that is also a valid glob pattern ([1] matches "1")
+        let sync_dir = tmp.path().join("sync[1]");
+        std::fs::create_dir(&sync_dir).unwrap();
+        std::fs::write(sync_dir.join("yaak.req_1.yaml"), "inside").unwrap();
+        std::fs::create_dir(tmp.path().join("sync1")).unwrap();
+        std::fs::write(tmp.path().join("sync1").join("decoy.txt"), "glob match").unwrap();
+
+        let status = git_worktree_status(&sync_dir).unwrap();
+        let paths: Vec<&str> = status.entries.iter().map(|e| e.rela_path.as_str()).collect();
+        assert_eq!(paths, vec!["sync[1]/yaak.req_1.yaml"]);
+    }
+
+    #[test]
+    fn test_status_scoped_to_subdir() {
+        let tmp = tempfile::tempdir().unwrap();
+        git2::Repository::init(tmp.path()).unwrap();
+
+        let sync_dir = tmp.path().join("sync");
+        std::fs::create_dir(&sync_dir).unwrap();
+        std::fs::write(sync_dir.join("yaak.req_1.yaml"), "inside").unwrap();
+        std::fs::write(sync_dir.join("README.md"), "external, but in sync dir").unwrap();
+        std::fs::write(tmp.path().join("outside.txt"), "outside").unwrap();
+
+        // The commit dialog's status only reports the sync directory
+        let status = git_status(&sync_dir).unwrap();
+        assert_eq!(status.rela_dir, "sync");
+        let mut paths: Vec<&str> = status.entries.iter().map(|e| e.rela_path.as_str()).collect();
+        paths.sort();
+        assert_eq!(paths, vec!["sync/README.md", "sync/yaak.req_1.yaml"]);
+
+        // Status on the repo root reports everything
+        let status = git_status(tmp.path()).unwrap();
+        assert_eq!(status.rela_dir, "");
+        assert_eq!(status.entries.len(), 3);
+    }
+}