Fix highlight code splitting graphemes.

This resulted in phantom blank lines in files with CRLF line
endings, but could potentially have manifested with other
graphemes as well.

1 files changed, 24 insertions, 0 deletions

diff --git a/helix-core/src/graphemes.rs b/helix-core/src/graphemes.rs
index 2e6a925f..e2f7c3f3 100644
--- a/helix-core/src/graphemes.rs
+++ b/helix-core/src/graphemes.rs
@@ -121,6 +121,30 @@ pub fn next_grapheme_boundary(slice: RopeSlice, char_idx: usize) -> usize {
     nth_next_grapheme_boundary(slice, char_idx, 1)
 }
 
+/// Returns the passed char index if it's already a grapheme boundary,
+/// or the next grapheme boundary char index if not.
+pub fn ensure_grapheme_boundary(slice: RopeSlice, char_idx: usize) -> usize {
+    if char_idx == 0 {
+        0
+    } else {
+        next_grapheme_boundary(slice, char_idx - 1)
+    }
+}
+
+/// Returns the passed byte index if it's already a grapheme boundary,
+/// or the next grapheme boundary byte index if not.
+pub fn ensure_grapheme_boundary_byte(slice: RopeSlice, byte_idx: usize) -> usize {
+    // TODO: we can avoid the byte/char conversions entirely
+    // if we also make byte versions of the other functions.
+    let char_idx = slice.byte_to_char(byte_idx);
+    let fixed_char_idx = ensure_grapheme_boundary(slice, char_idx);
+    if fixed_char_idx == char_idx {
+        byte_idx
+    } else {
+        slice.char_to_byte(fixed_char_idx)
+    }
+}
+
 /// Returns whether the given char position is a grapheme boundary.
 pub fn is_grapheme_boundary(slice: RopeSlice, char_idx: usize) -> bool {
     // Bounds check