From c96c98c6c545378d87f59d0da847ed3a8bd69e79 Mon Sep 17 00:00:00 2001
From: Gokul Soumya <gokulps15@gmail.com>
Date: Thu, 10 Mar 2022 19:30:25 +0530
Subject: [PATCH] Avoid string allocation when git diffing

---
 Cargo.lock                 |  1 +
 helix-vcs/Cargo.toml       |  1 +
 helix-vcs/src/git.rs       | 30 +++++++++++++++---------------
 helix-vcs/src/lib.rs       |  1 +
 helix-vcs/src/rope.rs      | 20 ++++++++++++++++++++
 helix-view/src/document.rs |  2 +-
 6 files changed, 39 insertions(+), 16 deletions(-)
 create mode 100644 helix-vcs/src/rope.rs

diff --git a/Cargo.lock b/Cargo.lock
index 4a9dfc7d84a9b..11eb9a9a1cc28 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -508,6 +508,7 @@ name = "helix-vcs"
 version = "0.6.0"
 dependencies = [
  "git2",
+ "ropey",
  "similar",
  "tempfile",
 ]
diff --git a/helix-vcs/Cargo.toml b/helix-vcs/Cargo.toml
index 8ea36b40297ef..bb24d1f1e1c9b 100644
--- a/helix-vcs/Cargo.toml
+++ b/helix-vcs/Cargo.toml
@@ -13,6 +13,7 @@ homepage = "https://helix-editor.com"
 [dependencies]
 git2 = { version = "0.13", default-features = false }
 similar = "2.1"
+ropey = "1.3"
 
 [dev-dependencies]
 tempfile = "3.3"
diff --git a/helix-vcs/src/git.rs b/helix-vcs/src/git.rs
index 748dee40c74ad..63eeb5bd9b35b 100644
--- a/helix-vcs/src/git.rs
+++ b/helix-vcs/src/git.rs
@@ -5,9 +5,10 @@ use std::{
 };
 
 use git2::{Oid, Repository};
+use ropey::Rope;
 use similar::DiffTag;
 
-use crate::{LineDiff, LineDiffs, RepoRoot};
+use crate::{rope::RopeLine, LineDiff, LineDiffs, RepoRoot};
 
 pub struct Git {
     repo: Repository,
@@ -17,7 +18,7 @@ pub struct Git {
 
     /// A cache mapping absolute file paths to file contents
     /// in the HEAD commit.
-    head_cache: HashMap<PathBuf, String>,
+    head_cache: HashMap<PathBuf, Rope>,
 }
 
 impl std::fmt::Debug for Git {
@@ -54,7 +55,7 @@ impl Git {
         path.strip_prefix(&self.root).ok()
     }
 
-    pub fn read_file_from_head(&mut self, file: &Path) -> Option<&str> {
+    pub fn read_file_from_head(&mut self, file: &Path) -> Option<&Rope> {
         let current_head = Self::head_commit_id(&self.repo)?;
         // TODO: Check cache validity on events like WindowChange
         // instead of on every keypress ? Will require hooks.
@@ -70,30 +71,28 @@ impl Git {
             let blob = object.peel_to_blob().ok()?;
             let contents = std::str::from_utf8(blob.content()).ok()?;
             self.head_cache
-                .insert(file.to_path_buf(), contents.to_string());
+                .insert(file.to_path_buf(), Rope::from_str(contents));
         }
 
-        self.head_cache.get(file).map(|s| s.as_str())
+        self.head_cache.get(file)
     }
 
-    pub fn line_diff_with_head(&mut self, file: &Path, contents: &str) -> LineDiffs {
-        let base = match self.read_file_from_head(file) {
-            Some(b) => b,
+    pub fn line_diff_with_head(&mut self, file: &Path, contents: &Rope) -> LineDiffs {
+        let old = match self.read_file_from_head(file) {
+            Some(rope) => RopeLine::from_rope(rope),
             None => return LineDiffs::new(),
         };
-        let mut config = similar::TextDiff::configure();
-        config.timeout(std::time::Duration::from_millis(250));
+        let new = RopeLine::from_rope(contents);
 
         let mut line_diffs: LineDiffs = HashMap::new();
-
         let mut mark_lines = |range: Range<usize>, change: LineDiff| {
             for line in range {
                 line_diffs.insert(line, change);
             }
         };
 
-        let diff = config.diff_lines(base, contents);
-        for op in diff.ops() {
+        let diff = similar::capture_diff_slices(similar::Algorithm::Myers, &old, &new);
+        for op in diff {
             let (tag, _, line_range) = op.as_tag_tuple();
             let start = line_range.start;
             match tag {
@@ -173,15 +172,16 @@ mod test {
         exec_git_cmd("commit -m message", git_dir);
 
         let mut git = Git::discover_from_path(&file).unwrap();
+        let rope = Rope::from_str(contents);
         assert_eq!(
-            Some(contents),
+            Some(&rope),
             git.read_file_from_head(&file),
             "Wrong blob contents from HEAD on clean index"
         );
 
         fs::write(&file, "new text").expect("Could not write to file");
         assert_eq!(
-            Some(contents),
+            Some(&rope),
             git.read_file_from_head(&file),
             "Wrong blob contents from HEAD when index is dirty"
         );
diff --git a/helix-vcs/src/lib.rs b/helix-vcs/src/lib.rs
index a7abc9e39b154..4e3e4f4d32910 100644
--- a/helix-vcs/src/lib.rs
+++ b/helix-vcs/src/lib.rs
@@ -1,4 +1,5 @@
 mod git;
+mod rope;
 
 use std::{
     cell::RefCell,
diff --git a/helix-vcs/src/rope.rs b/helix-vcs/src/rope.rs
new file mode 100644
index 0000000000000..53488fa2839b7
--- /dev/null
+++ b/helix-vcs/src/rope.rs
@@ -0,0 +1,20 @@
+use std::hash::Hash;
+
+use ropey::Rope;
+
+#[derive(PartialEq, Eq, PartialOrd, Ord, Clone)]
+pub struct RopeLine<'a>(pub ropey::RopeSlice<'a>);
+
+impl Hash for RopeLine<'_> {
+    fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
+        for chunk in self.0.chunks() {
+            chunk.hash(state);
+        }
+    }
+}
+
+impl<'a> RopeLine<'a> {
+    pub fn from_rope(rope: &'a Rope) -> Vec<Self> {
+        rope.lines().into_iter().map(RopeLine).collect()
+    }
+}
diff --git a/helix-view/src/document.rs b/helix-view/src/document.rs
index 5c4aeae6b9cba..f2ce5d220ff27 100644
--- a/helix-view/src/document.rs
+++ b/helix-view/src/document.rs
@@ -624,7 +624,7 @@ impl Document {
             .as_ref()
             .and_then(|v| v.try_borrow_mut().ok());
         if let Some((mut vcs, path)) = vcs.zip(self.path()) {
-            self.line_diffs = vcs.line_diff_with_head(path, &self.text().to_string());
+            self.line_diffs = vcs.line_diff_with_head(path, self.text());
         }
     }