Coalesce adjacent blame entries

Author: cruessler

gix-blame/tests/blame.rs

@@ -732,6 +732,47 @@ mod baseline {
     }
 }
 
+/// This function merges adjacent blame entries. It merges entries that are adjacent both in the
+/// blamed file as well as in the original file that introduced them. This follows `git`’s
+/// behaviour. `libgit2`, as of 2024-09-19, only checks whether two entries are adjacent in the
+/// blamed file which can result in different blames in certain edge cases. See [the commit][1]
+/// that introduced the extra check into `git` for context.
+///
+/// [1]: https://github.com/git/git/commit/c2ebaa27d63bfb7c50cbbdaba90aee4efdd45d0a
+fn coalesce_blame_entries(lines_blamed: Vec<BlameEntry>) -> Vec<BlameEntry> {
+    // TODO
+    // It’s possible this could better be done on insertion into `lines_blamed`.
+    lines_blamed.into_iter().fold(vec![], |mut acc, entry| {
+        let previous_entry = acc.last();
+
+        if let Some(previous_entry) = previous_entry {
+            if previous_entry.commit_id == entry.commit_id
+                && previous_entry.range_in_blamed_file.end == entry.range_in_blamed_file.start
+                // As of 2024-09-19, the check below only is in `git`, but not in `libgit2`.
+                && previous_entry.range_in_original_file.end == entry.range_in_original_file.start
+            {
+                let coalesced_entry = BlameEntry {
+                    range_in_blamed_file: previous_entry.range_in_blamed_file.start..entry.range_in_blamed_file.end,
+                    range_in_original_file: previous_entry.range_in_original_file.start
+                        ..entry.range_in_original_file.end,
+                    commit_id: previous_entry.commit_id,
+                };
+
+                acc.pop();
+                acc.push(coalesced_entry);
+            } else {
+                acc.push(entry);
+            }
+
+            acc
+        } else {
+            acc.push(entry);
+
+            acc
+        }
+    })
+}
+
 fn blame_file(worktree_path: PathBuf, file_path: &BStr) -> Vec<BlameEntry> {
     // TODO
     // At a high level, what we want to do is the following:
@@ -944,7 +985,7 @@ fn blame_file(worktree_path: PathBuf, file_path: &BStr) -> Vec<BlameEntry> {
     // order on insertion.
     lines_blamed.sort_by(|a, b| a.range_in_blamed_file.start.cmp(&b.range_in_blamed_file.start));
 
-    lines_blamed
+    coalesce_blame_entries(lines_blamed)
 }
 
 macro_rules! mktest {
@@ -979,6 +1020,7 @@ mktest!(added_lines_around, "added-lines-around", 3);
 mktest!(switched_lines, "switched-lines", 4);
 mktest!(added_line_before_changed_line, "added-line-before-changed-line", 3);
 mktest!(same_line_changed_twice, "same-line-changed-twice", 2);
+mktest!(coalesce_adjacent_hunks, "coalesce-adjacent-hunks", 1);
 
 #[test]
 fn process_change_works() {

gix-blame/tests/fixtures/make_blame_repo.sh

@@ -22,8 +22,10 @@ git commit -q -m c1.2
 
 echo "line 2" >> added-lines.txt
 echo "line 2" >> added-lines-around.txt
+echo -e "line 1\nline 2" > coalesce-adjacent-hunks.txt
 git add added-lines.txt
 git add added-lines-around.txt
+git add coalesce-adjacent-hunks.txt
 git commit -q -m c1.3
 
 echo "line 2" >> simple.txt
@@ -45,7 +47,9 @@ git add added-line-before-changed-line.txt
 git commit -q -m c2.3
 
 echo -e "line 1\nline 2" > same-line-changed-twice.txt
+echo -e "line 1\nline in between\nline 2" > coalesce-adjacent-hunks.txt
 git add same-line-changed-twice.txt
+git add coalesce-adjacent-hunks.txt
 git commit -q -m c2.4
 
 echo "line 3" >> simple.txt
@@ -65,7 +69,9 @@ git add changed-line-between-unchanged-lines.txt
 git commit -q -m c3.2
 
 echo -e "line 2\nline 3" > added-line-before-changed-line.txt
+echo -e "line 1\nline 2" > coalesce-adjacent-hunks.txt
 git add added-line-before-changed-line.txt
+git add coalesce-adjacent-hunks.txt
 git commit -q -m c3.3
 
 echo -e "line 1\nline 2 changed" > same-line-changed-twice.txt
@@ -121,3 +127,4 @@ git blame --porcelain added-lines-around.txt > .git/added-lines-around.baseline
 git blame --porcelain switched-lines.txt > .git/switched-lines.baseline
 git blame --porcelain added-line-before-changed-line.txt > .git/added-line-before-changed-line.baseline
 git blame --porcelain same-line-changed-twice.txt > .git/same-line-changed-twice.baseline
+git blame --porcelain coalesce-adjacent-hunks.txt > .git/coalesce-adjacent-hunks.baseline