aboutsummaryrefslogtreecommitdiff
path: root/helix-core
diff options
context:
space:
mode:
authorBlaž Hrastnik2022-01-14 03:13:53 +0000
committerBlaž Hrastnik2022-01-23 07:04:26 +0000
commit2302869836eedf6bba718815fbff3a5fd7afee36 (patch)
tree46f515dceb530ed8894a61999088efb073436324 /helix-core
parentadd3be85289a701bf075307d2dfef9fd8a50f69f (diff)
fix: ensure_grapheme_boundary_next_byte needs to index at valid char
Diffstat (limited to 'helix-core')
-rw-r--r--helix-core/src/graphemes.rs32
1 files changed, 31 insertions, 1 deletions
diff --git a/helix-core/src/graphemes.rs b/helix-core/src/graphemes.rs
index a6c74f65..aa898684 100644
--- a/helix-core/src/graphemes.rs
+++ b/helix-core/src/graphemes.rs
@@ -203,7 +203,12 @@ pub fn ensure_grapheme_boundary_next_byte(slice: RopeSlice, byte_idx: usize) ->
if byte_idx == 0 {
byte_idx
} else {
- next_grapheme_boundary_byte(slice, byte_idx - 1)
+ // TODO: optimize so we're not constructing grapheme cursor twice
+ if is_grapheme_boundary_byte(slice, byte_idx) {
+ byte_idx
+ } else {
+ next_grapheme_boundary_byte(slice, byte_idx)
+ }
}
}
@@ -235,6 +240,31 @@ pub fn is_grapheme_boundary(slice: RopeSlice, char_idx: usize) -> bool {
}
}
+/// Returns whether the given byte position is a grapheme boundary.
+#[must_use]
+pub fn is_grapheme_boundary_byte(slice: RopeSlice, byte_idx: usize) -> bool {
+ // Bounds check
+ debug_assert!(byte_idx <= slice.len_bytes());
+
+ // Get the chunk with our byte index in it.
+ let (chunk, chunk_byte_idx, _, _) = slice.chunk_at_byte(byte_idx);
+
+ // Set up the grapheme cursor.
+ let mut gc = GraphemeCursor::new(byte_idx, slice.len_bytes(), true);
+
+ // Determine if the given position is a grapheme cluster boundary.
+ loop {
+ match gc.is_boundary(chunk, chunk_byte_idx) {
+ Ok(n) => return n,
+ Err(GraphemeIncomplete::PreContext(n)) => {
+ let (ctx_chunk, ctx_byte_start, _, _) = slice.chunk_at_byte(n - 1);
+ gc.provide_context(ctx_chunk, ctx_byte_start);
+ }
+ Err(_) => unreachable!(),
+ }
+ }
+}
+
/// An iterator over the graphemes of a `RopeSlice`.
#[derive(Clone)]
pub struct RopeGraphemes<'a> {