Code at the speed of thought – Zed is a high-performance, multiplayer code editor from the creators of Atom and Tree-sitter.
Fix multibuffer chunk splitting at UTF-8 boundaries (#57641)
Oleksiy Syvokon7ae24463e189
diff --git a/crates/multi_buffer/src/multi_buffer.rs b/crates/multi_buffer/src/multi_buffer.rsindex 74eaeef53eb..4b1231af45a 100644--- a/crates/multi_buffer/src/multi_buffer.rs+++ b/crates/multi_buffer/src/multi_buffer.rs@@ -7934,7 +7934,11 @@ impl<'a> Iterator for MultiBufferChunks<'a> {if self.range.start >= self.range.end {return None;}- if self.range.start == self.diff_transforms.end().0 {+ while self+ .diff_transforms+ .item()+ .is_some_and(|_| self.range.start >= self.diff_transforms.end().0)+ {self.diff_transforms.next();}@@ -7961,10 +7965,17 @@ impl<'a> Iterator for MultiBufferChunks<'a> {let chunk_end = self.range.start + chunk.text.len();let diff_transform_end = diff_transform_end.min(self.range.end);- if diff_transform_end < chunk_end {- let split_idx = diff_transform_end - self.range.start;+ let split_idx = if diff_transform_end < chunk_end {+ chunk+ .text+ .ceil_char_boundary(diff_transform_end - self.range.start)+ } else {+ chunk.text.len()+ };++ if split_idx < chunk.text.len() {let (before, after) = chunk.text.split_at(split_idx);- self.range.start = diff_transform_end;+ self.range.start += split_idx;let mask = 1u128.unbounded_shl(split_idx as u32).wrapping_sub(1);let chars = chunk.chars & mask;let tabs = chunk.tabs & mask;diff --git a/crates/multi_buffer/src/multi_buffer_tests.rs b/crates/multi_buffer/src/multi_buffer_tests.rsindex 3fa5ac19d1a..b90b3425616 100644--- a/crates/multi_buffer/src/multi_buffer_tests.rs+++ b/crates/multi_buffer/src/multi_buffer_tests.rs@@ -1527,6 +1527,42 @@ async fn test_basic_diff_hunks(cx: &mut TestAppContext) {);}+#[gpui::test]+fn test_text_for_range_with_diff_transform_boundary_inside_multibyte_character(cx: &mut App) {+ let buffer = cx.new(|cx| Buffer::local("タx", cx));+ let multibuffer = cx.new(|cx| MultiBuffer::singleton(buffer, cx));+ let mut snapshot = multibuffer.read(cx).snapshot(cx);++ fn ascii_summary_with_byte_len(byte_len: usize) -> MBTextSummary {+ let text = "x".repeat(byte_len);+ MBTextSummary::from(TextSummary::from(text.as_str()))+ }++ // FR-16 shown a diff transform boundary two bytes into the leading 'タ'.+ // Build that transform tree directly so this test stays focused on chunk iteration.+ let mut diff_transforms = SumTree::default();+ diff_transforms.push(+ DiffTransform::BufferContent {+ summary: ascii_summary_with_byte_len(2),+ inserted_hunk_info: None,+ },+ (),+ );+ diff_transforms.push(+ DiffTransform::BufferContent {+ summary: ascii_summary_with_byte_len("タx".len() - 2),+ inserted_hunk_info: None,+ },+ (),+ );+ snapshot.diff_transforms = diff_transforms;++ let text = snapshot+ .text_for_range(MultiBufferOffset(0)..snapshot.len())+ .collect::<String>();+ assert_eq!(text, "タx");+}+#[gpui::test]async fn test_repeatedly_expand_a_diff_hunk(cx: &mut TestAppContext) {let text = indoc!(