feat(ops): Commit filtering support

This is a building block for faster operations by operating on less
data.

Author: epage

Cargo.toml

@@ -39,3 +39,12 @@ which = "4"
 [dev-dependencies]
 git-fixture = { version = "0.3", features = ["yaml"] }
 assert_fs = "1"
+eyre = "0.6"
+snapbox = "0.2"
+regex = "1.5.5"
+criterion = "0.3.5"
+
+[[bench]]
+harness = false
+name = "ops"
+path = "benches/ops.rs"

benches/ops.rs

@@ -0,0 +1,30 @@
+use std::collections::HashSet;
+
+fn get_repo() -> git2::Repository {
+    let repo_dir =
+        std::env::var("PATH_TO_REPO").expect("`PATH_TO_REPO` environment variable not set");
+    git2::Repository::discover(&std::path::PathBuf::from(repo_dir)).unwrap()
+}
+
+fn bench_get_changed_paths_between_trees(c: &mut criterion::Criterion) {
+    c.bench_function("get_changed_paths_between_trees", |b| {
+        let repo = get_repo();
+        let oid = repo.head().unwrap().target().unwrap();
+        let commit = repo.find_commit(oid).unwrap();
+        let parent = commit.parent(0).unwrap();
+        let parent_tree = parent.tree().unwrap();
+        let commit_tree = commit.tree().unwrap();
+
+        b.iter(|| -> HashSet<std::path::PathBuf> {
+            git2_ext::tree::get_changed_paths_between_trees(
+                &repo,
+                Some(&parent_tree),
+                Some(&commit_tree),
+            )
+            .unwrap()
+        });
+    });
+}
+
+criterion::criterion_group!(benches, bench_get_changed_paths_between_trees,);
+criterion::criterion_main!(benches);

src/bytes.rs

@@ -0,0 +1,13 @@
+// From git2 crate
+#[cfg(unix)]
+pub(crate) fn bytes2path(b: &[u8]) -> &std::path::Path {
+    use std::os::unix::prelude::*;
+    std::path::Path::new(std::ffi::OsStr::from_bytes(b))
+}
+
+// From git2 crate
+#[cfg(windows)]
+pub(crate) fn bytes2path(b: &[u8]) -> &std::path::Path {
+    use std::str;
+    std::path::Path::new(str::from_utf8(b).unwrap())
+}

src/lib.rs

@@ -6,4 +6,10 @@
 
 pub mod hooks;
 pub mod ops;
+pub mod tree;
 pub mod utils;
+
+pub(crate) mod bytes;
+
+#[cfg(test)]
+mod testing;

src/ops.rs

@@ -91,9 +91,19 @@ pub fn cherry_pick(
                     let our_path = conflict
                         .our
                         .as_ref()
-                        .map(|c| bytes2path(&c.path))
-                        .or_else(|| conflict.their.as_ref().map(|c| bytes2path(&c.path)))
-                        .or_else(|| conflict.ancestor.as_ref().map(|c| bytes2path(&c.path)))
+                        .map(|c| crate::bytes::bytes2path(&c.path))
+                        .or_else(|| {
+                            conflict
+                                .their
+                                .as_ref()
+                                .map(|c| crate::bytes::bytes2path(&c.path))
+                        })
+                        .or_else(|| {
+                            conflict
+                                .ancestor
+                                .as_ref()
+                                .map(|c| crate::bytes::bytes2path(&c.path))
+                        })
                         .unwrap_or_else(|| std::path::Path::new("<unknown>"));
                     format!("{}", our_path.display())
                 })
@@ -171,9 +181,19 @@ pub fn squash(
                 let our_path = conflict
                     .our
                     .as_ref()
-                    .map(|c| bytes2path(&c.path))
-                    .or_else(|| conflict.their.as_ref().map(|c| bytes2path(&c.path)))
-                    .or_else(|| conflict.ancestor.as_ref().map(|c| bytes2path(&c.path)))
+                    .map(|c| crate::bytes::bytes2path(&c.path))
+                    .or_else(|| {
+                        conflict
+                            .their
+                            .as_ref()
+                            .map(|c| crate::bytes::bytes2path(&c.path))
+                    })
+                    .or_else(|| {
+                        conflict
+                            .ancestor
+                            .as_ref()
+                            .map(|c| crate::bytes::bytes2path(&c.path))
+                    })
                     .unwrap_or_else(|| std::path::Path::new("<unknown>"));
                 format!("{}", our_path.display())
             })
@@ -218,16 +238,47 @@ pub fn reword(
     Ok(new_id)
 }
 
-// From git2 crate
-#[cfg(unix)]
-fn bytes2path(b: &[u8]) -> &std::path::Path {
-    use std::os::unix::prelude::*;
-    std::path::Path::new(std::ffi::OsStr::from_bytes(b))
-}
+/// Filter the directory tree of `commit_id` to just the specified file paths
+///
+/// Performance of some git2 operations scale with the number of files within the repo.  Operating on just the
+/// subset of files should dramatically speed up these operations
+pub fn filter_commit(
+    repo: &git2::Repository,
+    commit_id: git2::Oid,
+    parent_id: Option<git2::Oid>,
+    filtered_paths: &[&std::path::Path],
+) -> Result<git2::Oid, git2::Error> {
+    let commit = repo.find_commit(commit_id)?;
+    let tree = commit.tree()?;
+    let dehydrated_tree_oid = crate::tree::filter_tree(repo, &tree, filtered_paths)?;
+    let dehydrated_tree = repo.find_tree(dehydrated_tree_oid)?;
+
+    let signature = git2::Signature::new(
+        "git-branchless",
+        "[email protected]",
+        &git2::Time::new(0, 0),
+    )?;
+    let message = format!(
+        "generated by git-branchless: temporary dehydrated commit \
+                \
+                This commit was originally: {}",
+        commit_id
+    );
 
-// From git2 crate
-#[cfg(windows)]
-fn bytes2path(b: &[u8]) -> &std::path::Path {
-    use std::str;
-    std::path::Path::new(str::from_utf8(b).unwrap())
+    let dehydrated_parent: git2::Commit<'_>;
+    let parents = if let Some(parent_id) = parent_id {
+        dehydrated_parent = repo.find_commit(parent_id)?;
+        vec![&dehydrated_parent]
+    } else {
+        vec![]
+    };
+    let dehydrated_commit_oid = repo.commit(
+        None,
+        &signature,
+        &signature,
+        &message,
+        &dehydrated_tree,
+        &parents,
+    )?;
+    Ok(dehydrated_commit_oid)
 }